QSPIX - thoughts on interesting fund

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
User avatar
packer16
Posts: 859
Joined: Sat Jan 04, 2014 2:28 pm

Re: QSPIX - thoughts on interesting fund

Post by packer16 » Thu Aug 13, 2015 4:54 pm

randomguy wrote:
Angst wrote:I'm in packer16's camp here: my "Show Me" sensibilities have yet to be sated and my "too good to be true" radar flickers...

I can't help having the impression that many AQR boosters posting here (save Larry, perhaps) have not fully bought into their future with QSPIX - i.e. I sense a lot of dipping of one's toes into the water but not genuine commitment.

If QSPIX is really "real", then for someone who's already "won the game" and believes in QSPIX, rather than going in the direction of the Larry Portfolio for example, shouldn't they be apt to choose something more along the lines of perhaps 75% QSPIX and 25% Treasuries? Just throwing out numbers here of course, but generally speaking, isn't QSPIX pretty much the holy grail for those who've already won the game?

As it is for now, I'm in no more rush to commit to QSPIX than I am for any bleeding edge consumer software or hardware out there in the marketplace. What's a year or two or five in a lifetime of investing? I try to base my decisions more on my confidence in their outcomes rather than on my fear of regret for not having committed as soon as possible. If QSPIX is real, it will continue to exist and will come down in price. And I will be watching!


I don't think that is the right way of thinking about it. QSPIX isn't designed to be some low volatility investment without large swings. It is designed to give stock market returns with slighlty less volatility (something like that of a 60/40 portfolio) AND low correlations with Beta. If you view the Larry portfolio as something like 30 SV/70% treasuries (yes I know it is a bit more complicated), the QSPIX version would be something like 25%SV/10% QSPIX/65% treasuries (and no I didn't calculate what that acutally would return and expected volatility). QSPIX gives up some return compared to SV, so you need more risky exposure but you hope the low correlations keeps the swings in check.

Obviously you can wait but it is pretty much impossible to wait long enought. Lets say in 10 years, it has outperformed stocks by 5% with half the volatility. Is that a good time to buy or is it just a sign that we finished up a decade where the strategy worked great and about hit a decade where it trails stocks by 5% (i.e. we end 20 years with the predicated performance of similiar return to stocks)? We have people debating the SV premium with close to 100 years of data to look at. And I am not sure how much cheaper funds like this will get. There is a cost to these strategies and while I am sure they will get cheaper, I don't think we are going to be seeing them in .15% funds in the near future.


I don't see why a wait to see if the modeled performance actually generates positive returns after fees for 5 years is not a good idea. I think by not seeing how these guys actually perform you are taking on risks that cannot be seen from modeled performance and why would you pay more for that than other factor funds. I think the argument about more factor exposure is based upon the assumptions that these guys have distilled these factors so that they can be dialed up and down at will. I think the factors are not that well behaved (see SCV over time and Bogle's telltale chart) that this can actually be done.

Packer
Buy cheap and something good might happen

rpj2004
Posts: 18
Joined: Sat Feb 27, 2010 4:17 pm

Re: QSPIX - thoughts on interesting fund

Post by rpj2004 » Thu Aug 13, 2015 5:10 pm

HomerJ wrote:
randomguy wrote:Lets say in 10 years, it has outperformed stocks by 5% with half the volatility. Is that a good time to buy or is it just a sign that we finished up a decade where the strategy worked great and about hit a decade where it trails stocks by 5% (i.e. we end 20 years with the predicated performance of similiar return to stocks)?


That exactly is the problem... Even 10 years of good performance may turn out to just be luck. The 10 years may just be good years for those combinations of factors.

Very hard to know for sure... I definitely don't trust backtesting... EVERY new strategy ever invented back-tested well. And pretty much all of them crashed and burned when something different came along.

This is EXACTLY what happened to Asness in 2008 with his other AQR fund. He had lots of factors in that one too... Sure, different factors, but he tried to minimize downturns by devising a fund with a ton of factors that back-tested well against the 2000 crash... And then something new happened.

Are you sure this new fund of his will do just fine if yet another new thing happens?


Along the lines of this...on the surface a crash in which this fund does very poorly wouldn't actually be that bad. After all, this fund is mainly used to diversify your portfolio with tolerable returns. But, this fund uses leveraging. My fear is that while the four magical factors in the fund may not correlate right now, during that hypothetical crash they will, and with deleveraging the whole fund will go belly up.

randomguy
Posts: 4585
Joined: Wed Sep 17, 2014 9:00 am

Re: QSPIX - thoughts on interesting fund

Post by randomguy » Thu Aug 13, 2015 5:15 pm

packer16 wrote:I don't see why a wait to see if the modeled performance actually generates positive returns after fees for 5 years is not a good idea. I think by not seeing how these guys actually perform you are taking on risks that cannot be seen from modeled performance and why would you pay more for that than other factor funds. I think the argument about more factor exposure is based upon the assumptions that these guys have distilled these factors so that they can be dialed up and down at will. I think the factors are not that well behaved (see SCV over time and Bogle's telltale chart) that this can actually be done.

Packer


Problem is that 5 years isn't going to give you any clarity. Lets say you read the Farma-French paper in 1993 and decided to wait 5 years to see how it performed. Well in 1999 the results were pretty bad with it majorly laggin the S&P 500 (something like 8%). Did you invest just as SV was about to have its day in the sun leading to a 2% outperformance over the 20 year period or did you say that strategy obviously no longer works? If you are worried about something unexpected, that risk will still exist. Things like 2008 (or asian crisis of the 90s, and so on) don't happen very often and you would need the right one to sink the fund. Unless luck and the fund collapses, you are going to be in the same boat as far as judging risk.

User avatar
matjen
Posts: 1670
Joined: Sun Nov 20, 2011 11:30 pm

Re: QSPIX - thoughts on interesting fund

Post by matjen » Thu Aug 13, 2015 5:16 pm

Yawn...

QSPIX up .49% today. Total Bond (BND) down .21%. Total World Stock (VT) down .15%

Must be eating you fellas alive...a little part of you dying as QSPIX performs.
A man is rich in proportion to the number of things he can afford to let alone.

lack_ey
Posts: 5542
Joined: Wed Nov 19, 2014 11:55 pm

Re: QSPIX - thoughts on interesting fund

Post by lack_ey » Thu Aug 13, 2015 6:13 pm

All we can tell from the performance so far is that daily, weekly, etc. volatility is lower than that for stocks and that correlations with asset classes seem to be very low, at least for the kinds of market conditions we've seen since late 2013. That's all we're getting out of the play-by-play. Live returns in such a short period, or even 5 years or a decade aren't really enough to give a sense of what the fund does.

Basically, the backtest says this:
Image

And there are decades of data for some of the styles (particularly in equities) before that too. It is primarily our interpretation of the past that gets a heavy weighting and information about current results a very small one (other than the fact that it seems to check out that volatility over calmer market conditions is roughly as expected, and correlations likewise are as expected).

One could readily be fooled that this indicates long-term positive returns:
Image

But it doesn't really. I mean, the most popular commodities fund (DBC) dropped 40% over the same period. Based on just a graph of that you might conclude this is an asset that loses money fast, over the long run.


rpj2004 wrote:
HomerJ wrote:
randomguy wrote:Lets say in 10 years, it has outperformed stocks by 5% with half the volatility. Is that a good time to buy or is it just a sign that we finished up a decade where the strategy worked great and about hit a decade where it trails stocks by 5% (i.e. we end 20 years with the predicated performance of similiar return to stocks)?


That exactly is the problem... Even 10 years of good performance may turn out to just be luck. The 10 years may just be good years for those combinations of factors.

Very hard to know for sure... I definitely don't trust backtesting... EVERY new strategy ever invented back-tested well. And pretty much all of them crashed and burned when something different came along.

This is EXACTLY what happened to Asness in 2008 with his other AQR fund. He had lots of factors in that one too... Sure, different factors, but he tried to minimize downturns by devising a fund with a ton of factors that back-tested well against the 2000 crash... And then something new happened.

Are you sure this new fund of his will do just fine if yet another new thing happens?


Along the lines of this...on the surface a crash in which this fund does very poorly wouldn't actually be that bad. After all, this fund is mainly used to diversify your portfolio with tolerable returns. But, this fund uses leveraging. My fear is that while the four magical factors in the fund may not correlate right now, during that hypothetical crash they will, and with deleveraging the whole fund will go belly up.

It would be unusual for all the styles to blow up simultaneously. There's just a whole lot of data across asset classes, countries, time, etc. that shows value and momentum somewhat reliably not dying together, never mind the other two.

Also, that's mainly what the volatility targeting is for. They adjust the size of the positions based on recent volatility. No, it's not reliable, but it has generally worked in the past, because bad things and extreme drops tend not to happen out of nowhere—they're usually preceded by market jitters and higher volatility, so scaling positions down may reduce the impact a bit. Larry had an article about volatility targeting for momentum a couple months ago:
http://www.etf.com/sections/index-inves ... utperforms

No guarantees, definitely.

User avatar
packer16
Posts: 859
Joined: Sat Jan 04, 2014 2:28 pm

Re: QSPIX - thoughts on interesting fund

Post by packer16 » Thu Aug 13, 2015 7:30 pm

It is interesting looking at performance since inception data, to me it is a disappointment. From inception to date this fund has had an annualized return of 9.9% versus 11.4% for the S&P 500 and 7.8% for the Vanguard Balanced Fund. I am not sure where the backtested outperformance went but alot of it disappeared. To be fair, this is a short period of time and in most cases of simulated performance I have seen similar results. Now if you add back the fee of 150bp you get equity returns for less volatility. To date, it looks like AQR is taking the excess return and the investor is left with the big goose egg versus the index.

Packer
Buy cheap and something good might happen

lack_ey
Posts: 5542
Joined: Wed Nov 19, 2014 11:55 pm

Re: QSPIX - thoughts on interesting fund

Post by lack_ey » Thu Aug 13, 2015 7:48 pm

packer16 wrote:It is interesting looking at performance since inception data, to me it is a disappointment. From inception to date this fund has had an annualized return of 9.9% versus 11.4% for the S&P 500 and 7.8% for the Vanguard Balanced Fund. I am not sure where the backtested outperformance went but alot of it disappeared. To be fair, this is a short period of time and in most cases of simulated performance I have seen similar results. Now if you add back the fee of 150bp you get equity returns for less volatility. To date, it looks like AQR is taking the excess return and the investor is left with the big goose egg versus the index.

Packer

So to be clear, you are benchmarking it against US equity despite it being net hedged and not long stocks? (and not global equity)

User avatar
packer16
Posts: 859
Joined: Sat Jan 04, 2014 2:28 pm

Re: QSPIX - thoughts on interesting fund

Post by packer16 » Thu Aug 13, 2015 8:17 pm

Yes, I am because I have am not confident that the historical simulation on the volatility side will be any better than on the return side. I have two buckets, bonds and CDs for 3 years of expenses and growth investments. This would fall into the growth bucket so with this framework this is an expensive option to get returns. For this investment to make sense for me I would have to sell equity which would result in a lower return by the amount of the expenses. BTW the fund does appear to be meeting its goal before expenses (equity returns with less risk) but it is the expenses that cause it to underperform.

Packer
Buy cheap and something good might happen

Angst
Posts: 1641
Joined: Sat Jun 09, 2007 11:31 am
Location: St Louis, MO

Re: QSPIX - thoughts on interesting fund

Post by Angst » Thu Aug 13, 2015 8:18 pm

lack_ey wrote:I would say that the composite backtest numbers (17.4% annualized excess return above risk-free rate with 10.3% standard deviation, -0.09/0.09/0.13 correlation to equities/bonds/commodities), even after discounting for transaction costs or lost opportunities from real-world implementation that is cognizant of costs, are too good to be true. You can't expect factors and strategies discovered to work well to do quite as well in the future for all the usual reasons. But you don't need returns like that. If you just get something like 60/40 stock/bond returns and volatility but with frequently low correlation to the combination, you have a winner. Even with slightly worse risk/return characteristics, as long as the correlation is low, you're slightly better off with the inclusion.

randomguy wrote:Obviously you can wait but it is pretty much impossible to wait long enough. Lets say in 10 years, it has outperformed stocks by 5% with half the volatility. Is that a good time to buy or is it just a sign that we finished up a decade where the strategy worked great and about hit a decade where it trails stocks by 5% (i.e. we end 20 years with the predicated performance of similar return to stocks)? We have people debating the SV premium with close to 100 years of data to look at. And I am not sure how much cheaper funds like this will get. There is a cost to these strategies and while I am sure they will get cheaper, I don't think we are going to be seeing them in .15% funds in the near future.

(My underlines above)

One has expectations for matching a 60/40 portfolio but with less volatility, while another can accept the potential for 10 years of 5% under-performance vs. equities. Are those not mutually exclusive expectations? Perhaps not, but they don't come across in concert. Does the backtesting suggest there's potential for many years of under-performance vs. equities?

From my somewhat weathered investing perspective (1987 seems like yesterday), crashes and crises are a dime a dozen, and it's been a good while since 2008 (well, we did have that little taper tantrum in 2011), so not only do I have no trouble waiting a few years to see what I think of QSPIX and hold off on paying that 1.5% ER for the time being, there's no need to compare this with waiting for the Value premium to show up. Even then though, QSPIX is supposed to capture a basket of somewhat uncorrelated premia independent of beta, and as such ought to be more reliable and stable than the value premium on its own. (Consider the graph lack_ey posted above.) Regardless, I'm simply more concerned with the effective implementation of the fund's objectives, reliably capturing these premia in the face of crashes and crises and so forth, not with the factors always outperforming the market over any particular stretch of time. So for me, waiting a while is prudent.

lack_ey
Posts: 5542
Joined: Wed Nov 19, 2014 11:55 pm

Re: QSPIX - thoughts on interesting fund

Post by lack_ey » Thu Aug 13, 2015 8:42 pm

packer16 wrote:Yes, I am because I have am not confident that the historical simulation on the volatility side will be any better than on the return side. I have two buckets, bonds and CDs for 3 years of expenses and growth investments. This would fall into the growth bucket so with this framework this is an expensive option to get returns. For this investment to make sense for me I would have to sell equity which would result in a lower return by the amount of the expenses. BTW the fund does appear to be meeting its goal before expenses (equity returns with less risk) but it is the expenses that cause it to underperform.

Packer

But on the other hand, it invests globally and has beaten global equities over the period. Though as we both know, ~2 years of performance results are close to meaningless.

In any case, if you stick to a framework like yours of growth buckets and lose any nuance of in-between assets, this is very far from the usual portfolio theory constructs that would lead someone to consider (and perhaps reject, but consider in the first place) commodities, gold, other alternatives, etc. If you have no interest in those, you'll probably not like this. As Larry said a long while back, it's an alternative to alternatives.


Angst wrote:[...]One has expectations for matching a 60/40 portfolio but with less volatility, while another can accept the potential for 10 years of 5% under-performance vs. equities. Are those not mutually exclusive expectations? Perhaps not, but they don't come across in concert. Does the backtesting suggest there's potential for many years of under-performance vs. equities?

From my somewhat weathered investing perspective (1987 seems like yesterday), crashes and crises are a dime a dozen, and it's been a good while since 2008 (well, we did have that little taper tantrum in 2011), so not only do I have no trouble waiting a few years to see what I think of QSPIX and hold off on paying that 1.5% ER for the time being, there's no need to compare this with waiting for the Value premium to show up. Even then though, QSPIX is supposed to capture a basket of somewhat uncorrelated premia independent of beta, and as such ought to be more reliable and stable than the value premium on its own. (Consider the graph lack_ey posted above.) Regardless, I'm simply more concerned with the effective implementation of the fund's objectives, reliably capturing these premia in the face of crashes and crises and so forth, not with the factors always outperforming the market over any particular stretch of time. So for me, waiting a while is prudent.

Both quoted scenarios given were hypotheticals. Targeted volatility is something around the range of 10% annually. That's about where 60/40 has been over the long run, so that's why I brought it up. If you have two assets (here one is actually a combination, 60/40) that have similar returns and risks but low correlation, that's better than one asset with those characteristics. If the fund has significantly worse performance than 60/40 then you're just getting deworsification. Slightly worse, and it's about even or there's some mild benefit.

The return figure thrown around is about 7%, though I have no idea if that means real, nominal, nominal above the risk-free rate, or what. I frankly don't think people quite know what to expect. Personally, I would be happy enough with 3-4% in this environment with what bonds yield these days.

In any case, alternatives are definitely not replacements for equities. And the best alternative in practice may be no alternatives. Who knows.

I suppose I'm just not that concerned with implementation issues. The strategy has some decent turnover compared to what some people may be used to, but isn't exactly trading furiously. The backtest restricted itself to the most liquid stocks, futures, etc.* The fund company has a reasonable history of execution, if not things going as planned all the time. The live performance attribution data shows different categories working at different times. Getting exposure to the styles under the rules they came up with doesn't seem all that hard. They run some screens and figure out what needs to be bought and what shorted.

*recall that momentum is said to be stronger in small caps. No small caps in the backtest. No obscure commodity futures nobody actually trades. The capacity of the styles as tested is reasonably large.

randomguy
Posts: 4585
Joined: Wed Sep 17, 2014 9:00 am

Re: QSPIX - thoughts on interesting fund

Post by randomguy » Thu Aug 13, 2015 9:36 pm

Angst wrote:[

One has expectations for matching a 60/40 portfolio but with less volatility, while another can accept the potential for 10 years of 5% under-performance vs. equities. Are those not mutually exclusive expectations? Perhaps not, but they don't come across in concert. Does the backtesting suggest there's potential for many years of under-performance vs. equities?



I don't see how they remotely mutual exclusive. QSPIX gets it returns form different factors than a 60/40 fund. That is the whole point of it. 60/40 over 10 year rolling periods you will see huge variation in performance (best is like 16%, worse is like 0% with an average of around 9.) If the stock/bonds have a good period (call it 12%) and QSPIX has an slightly below average one (call it 7), you end up with a big gap. And obviously it works in reverse. If QSPIX has lower volatility (say the range is 6% to 12% instead of 0-16), you still can get this outperformance.

SnowSkier
Posts: 262
Joined: Fri Mar 16, 2012 12:21 pm
Location: Austin

Re: QSPIX - thoughts on interesting fund

Post by SnowSkier » Thu Aug 13, 2015 9:59 pm

FYI, here's an interesting article from AQR exploring Portfolio Construction, including Alt Risk Premia (such as a diversified portfolio of
long/short style premia)...aka QSPIX

the link, where you can download the pdf article:
https://www.aqr.com/library/aqr-publica ... nstruction

or just directly download the pdf article:
https://www.aqr.com/~/media/files/paper ... g-2q15.pdf

longinvest
Posts: 2447
Joined: Sat Aug 11, 2012 8:44 am

Re: QSPIX - thoughts on interesting fund

Post by longinvest » Thu Aug 13, 2015 10:54 pm

I don't understand how, regardless of whatever promise this fund's promoters make, one would just go ahead and give 1.5% of his portfolio to AQR every year! The compound costs are just incredible:

Let's assume a compound growth, before fees, of 5% per year over 30 years.
Without fees, $10,000 would grow (5%) to $43,219.42 in 30 years.
With 1.5% fees, $10,000 would grow (3.5%) to $28,067.94 in 30 years.

So, out of $33,219.42 in growth, the investor is giving away $15,151.48 to AQR. That's 45% of the growth, almost half of it! :oops:
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VAB/ZRR

randomguy
Posts: 4585
Joined: Wed Sep 17, 2014 9:00 am

Re: QSPIX - thoughts on interesting fund

Post by randomguy » Thu Aug 13, 2015 11:35 pm

longinvest wrote:I don't understand how, regardless of whatever promise this fund's promoters make, one would just go ahead and give 1.5% of his portfolio to AQR every year! The compound costs are just incredible:

Let's assume a compound growth, before fees, of 5% per year over 30 years.
Without fees, $10,000 would grow (5%) to $43,219.42 in 30 years.
With 1.5% fees, $10,000 would grow (3.5%) to $28,067.94 in 30 years.

So, out of $33,219.42 in growth, the investor is giving away $15,151.48 to AQR. That's 45% of the growth, almost half of it! :oops:


And if QSPIX returns 7% after fees you end up with 76k. Would you rather give AQR a ton of money and have an extra 33k or give them no money and have 33k less? :) Yes low fees are great but it isn't like you can choose between paying 1.5% and .5%(or .05 if all you want is long beta) and have the same product like you can do with stock funds. If there is another product that suggests it is going to return 7% (after fees), have a volatility of around 10% and have 0 correlation with beta I would love to hear about it. Some of the commodities come close (low correlation but high volatility) but they are also expensive. Low fees are great but they are not the only thing to consider.

If you buy the premis of the fund (decent returns, low correlation), the question you have to ask is does the product make sense for you? For a 25 year old with a 40 year investment horizon, going 100% stocks might be a much better alternative as they don't care about volatility and any rebalancing bonus you get isn't going to add up to much. For the 60 year old retiree, going 60/40 with 20% of the equity allocation in QSPIX might reduce the sequence of risk enough to let you eke out another .3% SWR (unfortunately we have no clue how QSPIX would have performed in the 70s. But we know international diversification was good enough for another .3%) while still sleeping well at night.

User avatar
HomerJ
Posts: 10157
Joined: Fri Jun 06, 2008 12:50 pm

Re: QSPIX - thoughts on interesting fund

Post by HomerJ » Fri Aug 14, 2015 12:25 am

randomguy wrote:And if QSPIX returns 7% after fees you end up with 76k. Would you rather give AQR a ton of money and have an extra 33k or give them no money and have 33k less? :) Yes low fees are great but it isn't like you can choose between paying 1.5% and .5%(or .05 if all you want is long beta) and have the same product like you can do with stock funds. If there is another product that suggests it is going to return 7% (after fees), have a volatility of around 10% and have 0 correlation with beta I would love to hear about it.


Oh, all kinds of funds can SUGGEST they are going to return 7% after fees... That's the easy part.

User avatar
tarheel
Posts: 304
Joined: Fri Sep 17, 2010 8:33 am

Re: QSPIX - thoughts on interesting fund

Post by tarheel » Fri Aug 14, 2015 5:30 am

Angst wrote:I'm in packer16's camp here: my "Show Me" sensibilities have yet to be sated and my "too good to be true" radar flickers...

I can't help having the impression that many AQR boosters posting here (save Larry, perhaps) have not fully bought into their future with QSPIX - i.e. I sense a lot of dipping of one's toes into the water but not genuine commitment.

If QSPIX is really "real", then for someone who's already "won the game" and believes in QSPIX, rather than going in the direction of the Larry Portfolio for example, shouldn't they be apt to choose something more along the lines of perhaps 75% QSPIX and 25% Treasuries? Just throwing out numbers here of course, but generally speaking, isn't QSPIX pretty much the holy grail for those who've already won the game?

As it is for now, I'm in no more rush to commit to QSPIX than I am for any bleeding edge consumer software or hardware out there in the marketplace. What's a year or two or five in a lifetime of investing? I try to base my decisions more on my confidence in their outcomes rather than on my fear of regret for not having committed as soon as possible. If QSPIX is real, it will continue to exist and will come down in price. And I will be watching!


My IPS has 7.5% allocation to QSPIX with at least a permanent 10% allocation long term. If that's not commitment I don't know what is.

User avatar
packer16
Posts: 859
Joined: Sat Jan 04, 2014 2:28 pm

Re: QSPIX - thoughts on interesting fund

Post by packer16 » Fri Aug 14, 2015 5:38 am

randomguy wrote:
longinvest wrote:I don't understand how, regardless of whatever promise this fund's promoters make, one would just go ahead and give 1.5% of his portfolio to AQR every year! The compound costs are just incredible:

Let's assume a compound growth, before fees, of 5% per year over 30 years.
Without fees, $10,000 would grow (5%) to $43,219.42 in 30 years.
With 1.5% fees, $10,000 would grow (3.5%) to $28,067.94 in 30 years.

So, out of $33,219.42 in growth, the investor is giving away $15,151.48 to AQR. That's 45% of the growth, almost half of it! :oops:


And if QSPIX returns 7% after fees you end up with 76k. Would you rather give AQR a ton of money and have an extra 33k or give them no money and have 33k less? :) Yes low fees are great but it isn't like you can choose between paying 1.5% and .5%(or .05 if all you want is long beta) and have the same product like you can do with stock funds. If there is another product that suggests it is going to return 7% (after fees), have a volatility of around 10% and have 0 correlation with beta I would love to hear about it. Some of the commodities come close (low correlation but high volatility) but they are also expensive. Low fees are great but they are not the only thing to consider.

If you buy the premis of the fund (decent returns, low correlation), the question you have to ask is does the product make sense for you? For a 25 year old with a 40 year investment horizon, going 100% stocks might be a much better alternative as they don't care about volatility and any rebalancing bonus you get isn't going to add up to much. For the 60 year old retiree, going 60/40 with 20% of the equity allocation in QSPIX might reduce the sequence of risk enough to let you eke out another .3% SWR (unfortunately we have no clue how QSPIX would have performed in the 70s. But we know international diversification was good enough for another .3%) while still sleeping well at night.


But the excess return is just in theory. To date, QSPIX has delivered below equity market returns and the difference has all gone to the asset manager, so you have paid 100% of the outperformance to AQR, unless you put this into something other than a equity or risk category in your portfolio (if you do this IMO you are stretching the utility of this for simple folks like me). At this point it is show me fund.

Packer
Buy cheap and something good might happen

User avatar
tarheel
Posts: 304
Joined: Fri Sep 17, 2010 8:33 am

Re: QSPIX - thoughts on interesting fund

Post by tarheel » Fri Aug 14, 2015 5:47 am

randomguy wrote:
longinvest wrote:I don't understand how, regardless of whatever promise this fund's promoters make, one would just go ahead and give 1.5% of his portfolio to AQR every year! The compound costs are just incredible:

Let's assume a compound growth, before fees, of 5% per year over 30 years.
Without fees, $10,000 would grow (5%) to $43,219.42 in 30 years.
With 1.5% fees, $10,000 would grow (3.5%) to $28,067.94 in 30 years.

So, out of $33,219.42 in growth, the investor is giving away $15,151.48 to AQR. That's 45% of the growth, almost half of it! :oops:


And if QSPIX returns 7% after fees you end up with 76k. Would you rather give AQR a ton of money and have an extra 33k or give them no money and have 33k less? :) Yes low fees are great but it isn't like you can choose between paying 1.5% and .5%(or .05 if all you want is long beta) and have the same product like you can do with stock funds. If there is another product that suggests it is going to return 7% (after fees), have a volatility of around 10% and have 0 correlation with beta I would love to hear about it. Some of the commodities come close (low correlation but high volatility) but they are also expensive. Low fees are great but they are not the only thing to consider.

If you buy the premis of the fund (decent returns, low correlation), the question you have to ask is does the product make sense for you? For a 25 year old with a 40 year investment horizon, going 100% stocks might be a much better alternative as they don't care about volatility and any rebalancing bonus you get isn't going to add up to much. For the 60 year old retiree, going 60/40 with 20% of the equity allocation in QSPIX might reduce the sequence of risk enough to let you eke out another .3% SWR (unfortunately we have no clue how QSPIX would have performed in the 70s. But we know international diversification was good enough for another .3%) while still sleeping well at night.


First of all, looking at 5% (no fees) versus 3.5% (with 1.5% ER) with any investment is pointless because it is not viewed in the context of the portfolio.

I believe that the expense of QSPIX is well worth it given projected returns and low correlation. Do the math for yourself. I hope you really believe that "the rebalancing bonus isn't going to add up to much". I couldn't disagree more.

For those of you really caught up on the 1.5% ER, I suggest you calculate how much a 10% allocation to QSPIX would effect your overall portfolio ER. For me it was about 5 basis points.

I really hope that every one of you that has issue with the ER does not use a financial advisor. If you don't and three-fund then fine, argue away. If you do, then how about you calculate the effect of that on your CAGR and compare it to QSPIX.

User avatar
tarheel
Posts: 304
Joined: Fri Sep 17, 2010 8:33 am

Re: QSPIX - thoughts on interesting fund

Post by tarheel » Fri Aug 14, 2015 5:48 am

packer16 wrote:
randomguy wrote:
longinvest wrote:I don't understand how, regardless of whatever promise this fund's promoters make, one would just go ahead and give 1.5% of his portfolio to AQR every year! The compound costs are just incredible:

Let's assume a compound growth, before fees, of 5% per year over 30 years.
Without fees, $10,000 would grow (5%) to $43,219.42 in 30 years.
With 1.5% fees, $10,000 would grow (3.5%) to $28,067.94 in 30 years.

So, out of $33,219.42 in growth, the investor is giving away $15,151.48 to AQR. That's 45% of the growth, almost half of it! :oops:


And if QSPIX returns 7% after fees you end up with 76k. Would you rather give AQR a ton of money and have an extra 33k or give them no money and have 33k less? :) Yes low fees are great but it isn't like you can choose between paying 1.5% and .5%(or .05 if all you want is long beta) and have the same product like you can do with stock funds. If there is another product that suggests it is going to return 7% (after fees), have a volatility of around 10% and have 0 correlation with beta I would love to hear about it. Some of the commodities come close (low correlation but high volatility) but they are also expensive. Low fees are great but they are not the only thing to consider.

If you buy the premis of the fund (decent returns, low correlation), the question you have to ask is does the product make sense for you? For a 25 year old with a 40 year investment horizon, going 100% stocks might be a much better alternative as they don't care about volatility and any rebalancing bonus you get isn't going to add up to much. For the 60 year old retiree, going 60/40 with 20% of the equity allocation in QSPIX might reduce the sequence of risk enough to let you eke out another .3% SWR (unfortunately we have no clue how QSPIX would have performed in the 70s. But we know international diversification was good enough for another .3%) while still sleeping well at night.


But the excess return is just in theory. To date, QSPIX has delivered below equity market returns and the difference has all gone to the asset manager, so you have paid 100% of the outperformance to AQR, unless you put this into something other than a equity or risk category in your portfolio (if you do this IMO you are stretching the utility of this for simple folks like me). At this point it is show me fund.

Packer


The point of this fund is not to achieve equity-like returns and none of us that invest in it are expecting that.

User avatar
packer16
Posts: 859
Joined: Sat Jan 04, 2014 2:28 pm

Re: QSPIX - thoughts on interesting fund

Post by packer16 » Fri Aug 14, 2015 6:01 am

Then you are paying an alot for a balanced fund rate of return, at least 25% of the return if they are successful at obtaining 6% , much more if they cannot. As some have mentioned above, this is an alternative to the alternative. I would never invest in alternatives as the fees are crazy and the returns less than equity, so comparing this to a terrible investment I think is a little silly.

The alternatives I like can yield me almost 7% plus 3 to 4% appreciation in areas like NNN, nursing home and container ship leasing versus maybe a 7% return in an expensive black box investment.

Packer
Buy cheap and something good might happen

User avatar
matjen
Posts: 1670
Joined: Sun Nov 20, 2011 11:30 pm

Re: QSPIX - thoughts on interesting fund

Post by matjen » Fri Aug 14, 2015 6:51 am

packer16 wrote:To date, QSPIX has delivered below equity market returns and the difference has all gone to the asset manager, so you have paid 100% of the outperformance to AQR, unless you put this into something other than a equity or risk category in your portfolio (if you do this IMO you are stretching the utility of this for simple folks like me). At this point it is show me fund.

Packer


NO, NO, NO. Come on Packer! As Tarheel said (as have countless others over, and over and over and over) this fund isn't expected to outperform equities. Regardless, your ability to discuss things after the fact is absurd and disingenuous. It's like someone coming on this board and saying indexing is for losers because he picked Apple, Facebook, and Home Depot the last 5 years. We have gone through this multiple times. Spare me your Sequoia or Total US (Didn't Lack_ey call you on the US v. Int'l distinction just a handful of posts ago?). Why don't you pick International, Emerging Markets or some other slice of equities? We both know the answer don't we. And how many times do you have to be reminded that QSPIX isn't considered equity. You can consider it all you want to be equity but no one else does. It's like me saying Total Bond should be equity because 50% of it is comprised of nutty stuff like securitized and corporates. In my personal definition of bonds only US Treasuries/TIPS count....Discussing generally accepted things and changing the definitions along your own personal whims isn't very fruitful it seems to me.

Again, to the extent you are going to even suggest QSPIX be compared to equities (for giggles only I might add) the only fair benchmark would be total world. In that context QSPIX is actually outperforming.

As we used to say at high school basketball games. SCOREBOARD, SCOREBOARD, SCOREBOARD*...QSPIX, Total World, Total Bond, And Vanguard Market Neutral.

Image

*It should be pointed out that I fully expect VT to pull ahead eventually and that QSPIX should be viewed in the context of an entire portfolio not on its own...but this is fun. ;-)

The hilarious part about all this is that "if" QSPIX performs well over 10+ years and we have this discussion, folks like you and Homer and longinvest will then insist that it is too late and that the market will correct for the now known strategy...
Last edited by matjen on Fri Aug 14, 2015 8:57 am, edited 3 times in total.
A man is rich in proportion to the number of things he can afford to let alone.

User avatar
grap0013
Posts: 1861
Joined: Thu Mar 18, 2010 1:24 pm

Re: QSPIX - thoughts on interesting fund

Post by grap0013 » Fri Aug 14, 2015 6:57 am

packer16 wrote:But the excess return is just in theory. To date, QSPIX has delivered below equity market returns and the difference has all gone to the asset manager, so you have paid 100% of the outperformance to AQR, unless you put this into something other than a equity or risk category in your portfolio (if you do this IMO you are stretching the utility of this for simple folks like me). At this point it is show me fund.

Packer


Highly incorrect. https://www.portfoliovisualizer.com/ass ... +VT%2C+BND

QSPIX has whooped up on VT (total world) BND (total bond) and a 60:40 combo of the two since inception. You can throw out all that 1.5% ER and high turnover business in this regard. All that matters are real returns in your pocket and how this fund fits in to overall portfolio return.

Your logic is analogous to avoiding a 400K salary just to avoid paying high taxes. You're too hung up on fees IMO. "Letting the tax tail wag the income dog". I think debating this with you is likely futile though. Hopefully others may benefit from this discussion.
There are no guarantees, only probabilities.

randomguy
Posts: 4585
Joined: Wed Sep 17, 2014 9:00 am

Re: QSPIX - thoughts on interesting fund

Post by randomguy » Fri Aug 14, 2015 7:35 am

HomerJ wrote:
randomguy wrote:And if QSPIX returns 7% after fees you end up with 76k. Would you rather give AQR a ton of money and have an extra 33k or give them no money and have 33k less? :) Yes low fees are great but it isn't like you can choose between paying 1.5% and .5%(or .05 if all you want is long beta) and have the same product like you can do with stock funds. If there is another product that suggests it is going to return 7% (after fees), have a volatility of around 10% and have 0 correlation with beta I would love to hear about it.


Oh, all kinds of funds can SUGGEST they are going to return 7% after fees... That's the easy part.


Sure. But give me a list of any liquid fund that will theorically return 7%, have 10% volatility and low/no beta exposure. All of the ones I am aware of have higher ER (2%+ isn't unheard for various market neutral funds and the hedge funds 2/20), I would be willing to look into investing in any of them. If you don't believe the product can perform, that is a good reason not to invest. If you are worried about someone else making money off you, so you make suboptimal choices to avoid that, I think you have other money issues.

Angst
Posts: 1641
Joined: Sat Jun 09, 2007 11:31 am
Location: St Louis, MO

Re: QSPIX - thoughts on interesting fund

Post by Angst » Fri Aug 14, 2015 9:08 am

tarheel wrote:
Angst wrote:I'm in packer16's camp here: my "Show Me" sensibilities have yet to be sated and my "too good to be true" radar flickers...

I can't help having the impression that many AQR boosters posting here (save Larry, perhaps) have not fully bought into their future with QSPIX - i.e. I sense a lot of dipping of one's toes into the water but not genuine commitment.

If QSPIX is really "real", then for someone who's already "won the game" and believes in QSPIX, rather than going in the direction of the Larry Portfolio for example, shouldn't they be apt to choose something more along the lines of perhaps 75% QSPIX and 25% Treasuries? Just throwing out numbers here of course, but generally speaking, isn't QSPIX pretty much the holy grail for those who've already won the game?

As it is for now, I'm in no more rush to commit to QSPIX than I am for any bleeding edge consumer software or hardware out there in the marketplace. What's a year or two or five in a lifetime of investing? I try to base my decisions more on my confidence in their outcomes rather than on my fear of regret for not having committed as soon as possible. If QSPIX is real, it will continue to exist and will come down in price. And I will be watching!

My IPS has 7.5% allocation to QSPIX with at least a permanent 10% allocation long term. If that's not commitment I don't know what is.

Thanks for the info. I too would consider 10% to be a commitment, even just 7.5%. Of the "many AQR boosters posting here", I think that's exceptional. According to your IPS, when will 7.5% change to 10%? Is that based on an age milestone? Retirement? SS/pension?

User avatar
packer16
Posts: 859
Joined: Sat Jan 04, 2014 2:28 pm

Re: QSPIX - thoughts on interesting fund

Post by packer16 » Fri Aug 14, 2015 9:09 am

You guys seem to be convinced you need high fee alternatives to which I am skeptical. Lets talk in few years and after a time of distress and see where QSPIX is and how it compares to the balanced fund benchmark you want to compare it to. BTW if you look under the hood to see what this fund actually holds I don't see how you can not say it is a black box (all kinds of indices and futures and such).

Packer
Buy cheap and something good might happen

lack_ey
Posts: 5542
Joined: Wed Nov 19, 2014 11:55 pm

Re: QSPIX - thoughts on interesting fund

Post by lack_ey » Fri Aug 14, 2015 9:36 am

Angst wrote:
tarheel wrote:
Angst wrote:I'm in packer16's camp here: my "Show Me" sensibilities have yet to be sated and my "too good to be true" radar flickers...

I can't help having the impression that many AQR boosters posting here (save Larry, perhaps) have not fully bought into their future with QSPIX - i.e. I sense a lot of dipping of one's toes into the water but not genuine commitment.

If QSPIX is really "real", then for someone who's already "won the game" and believes in QSPIX, rather than going in the direction of the Larry Portfolio for example, shouldn't they be apt to choose something more along the lines of perhaps 75% QSPIX and 25% Treasuries? Just throwing out numbers here of course, but generally speaking, isn't QSPIX pretty much the holy grail for those who've already won the game?

As it is for now, I'm in no more rush to commit to QSPIX than I am for any bleeding edge consumer software or hardware out there in the marketplace. What's a year or two or five in a lifetime of investing? I try to base my decisions more on my confidence in their outcomes rather than on my fear of regret for not having committed as soon as possible. If QSPIX is real, it will continue to exist and will come down in price. And I will be watching!

My IPS has 7.5% allocation to QSPIX with at least a permanent 10% allocation long term. If that's not commitment I don't know what is.

Thanks for the info. I too would consider 10% to be a commitment, even just 7.5%. Of the "many AQR boosters posting here", I think that's exceptional. According to your IPS, when will 7.5% change to 10%? Is that based on an age milestone? Retirement? SS/pension?

I have about 6%. It would be higher (closer to 10%) but I didn't want to use my entire IRA space because then there wouldn't really be much opportunity to rebalance. I would prefer to have more like 15% total in risky alts including P2P lending so as to not rely solely on stocks for long-term returns. I figure 60-70% is already a large enough bet there if using a high savings rate. Sometimes the stock market doesn't bounce right back, and there could theoretically be stagnation in the long run where other assets do just fine. The goal isn't necessarily to outperform on average but to potentially be in a better position if stocks are disappointing for a couple decades. If they do well, with 60-70% I think that will make me happy enough.


packer16 wrote:You guys seem to be convinced you need high fee alternatives to which I am skeptical. Lets talk in few years and after a time of distress and see where QSPIX is and how it compares to the balanced fund benchmark you want to compare it to. BTW if you look under the hood to see what this fund actually holds I don't see how you can not say it is a black box (all kinds of indices and futures and such).

Packer

I would prefer low-fee alternatives but that's how it goes. A few years is still a terrible time to do performance evaluation. It may be interesting to see what happens, but the "few years" timeline is not long enough and gets everyone into trouble. That's when people ditch underperforming funds and switch to new ones, except some mean reversion means their old manager outperforms the new one going forward.

"Need" is too strong a word in any case. I wouldn't go beyond "probably useful, especially in some middling outcomes for the futures, but maybe not and possibly disastrous."

For all the talk of the fund being a black box—if that's what you want to call it, fine by me—the disclosures are far and away much greater than for most funds out there and most competitors. For example, if the minimum weren't so high I might consider Vanguard's market neutral fund but I don't know what their strategy really is and how they pick securities because they don't say beyond the wishy-washy qualifiers of undervalued vs. overvalued. That and I think multialternatives are a better idea than sticking with a single asset class or strategy.

afan
Posts: 2849
Joined: Sun Jul 25, 2010 4:01 pm

Re: QSPIX - thoughts on interesting fund

Post by afan » Fri Aug 14, 2015 10:42 am

Accept that one should only consider adding an asset in the context of the overall portfolio. Looking at it in isolation is meaningless.
However, in this case we have lots of backtesting and simulation, with very little real world experience.

To something like this one can have two reactions
"This is great. I have to be in this space NOW!!!"

or
"That's interesting. Most brilliant market innovations do not work. The back tested data look good due to data mining, chance, or market conditions that do not repeat. I'll make a note to take a look at real world performance in 10 or 20 years, when there have been dozens of funds in this area, and many market cycles. No hurry."

I am certainly in the second group. On the small chance that I will miss out on 20 years of somewhat improved risk adjusted returns, well, the world will keep turning.

There are two reactions to the 1.5% ER and perhaps costs on top of that for an advisor to give you access

"cost does not matter. This is great. I have to be in this space NOW!!!"

Or

"If this works out well over then next 10-20 years, Vanguard will offer it at a reasonable fee of <20 basis points. If it works out well and Vanguard offers it at the unconscionable fee of 50 basis points, I will pass. If it does not work out, Vanguard will never offer it. In any case, there is no way I will ever pay anyone 1.5% of assets to manage financial investments. EVER. The world would stop turning, it would kill me, and I would be twisting in my grave."
We don't know how to beat the market on a risk-adjusted basis, and we don't know anyone that does know either | --Swedroe | We assume that markets are efficient, that prices are right | --Fama

randomguy
Posts: 4585
Joined: Wed Sep 17, 2014 9:00 am

Re: QSPIX - thoughts on interesting fund

Post by randomguy » Fri Aug 14, 2015 10:49 am

afan wrote:"If this works out well over then next 10-20 years, Vanguard will offer it at a reasonable fee of <20 basis points. If it works out well and Vanguard offers it at the unconscionable fee of 50 basis points, I will pass. If it does not work out, Vanguard will never offer it. In any case, there is no way I will ever pay anyone 1.5% of assets to manage financial investments. EVER. The world would stop turning, it would kill me, and I would be twisting in my grave."


Except vanguard will not be able to offer this fund for .2 ER. The overhead of leverage and shorts will bump it up. You will end up with something like the market neutral fund where the ER is 1.64 but you get the *" Excluding the effect of expenses attributable to borrowing and dividend expenses on short sales, Total Annual Operating Expense Ratio would be 0.25%.".

afan
Posts: 2849
Joined: Sun Jul 25, 2010 4:01 pm

Re: QSPIX - thoughts on interesting fund

Post by afan » Fri Aug 14, 2015 11:07 am

randomguy wrote:
Except vanguard will not be able to offer this fund for .2 ER.



Then I will pass.

Same reason I am never investing in any other form of hedge fund.

Global warming or giant asteroids may kill us all, but lack of an actively managed fund making revolving factor bets has yet to be fatal to any person or portfolio.
We don't know how to beat the market on a risk-adjusted basis, and we don't know anyone that does know either | --Swedroe | We assume that markets are efficient, that prices are right | --Fama

afan
Posts: 2849
Joined: Sun Jul 25, 2010 4:01 pm

Re: QSPIX - thoughts on interesting fund

Post by afan » Fri Aug 14, 2015 11:28 am

One post shy of 1,000.

Among the many problems with this fund- the proponents argue that it helps smooth out the volatility of your overall portfolio due to its low correlation with the rest of your assets. it also has low correlation with the rest of your financial life. But since it varies its factor bets, and is free to drop some and adopt others, you never know what the factor composition of the portfolio will be. Perhaps it was a good counterbalance to the rest of your life, for some period of time, with some combination of bets. Then that changes. Of course, you don't even know about the change, let alone how the new combo fits with your portfolio.

How can you construct a portfolio like this? You put money in, say, stocks because you think they make a favorable long term contribution. Their returns are very noisy, but you think they have good risk/return characteristics and you want some exposure.

Now you have a fund whose investments are quite literally unknown. Whatever they may be now WILL change and you do not know how. How can you possibly decide what to do with it? I have an IPS that allocates X to stocks, Y to fixed income and Z to the mystery investment?? How much should Z be? 1%? 10%? 100%? What basis could there be for a decision to make an allocation to something when I don't know what it is?

Even if the fee were 20bp, it is hard to imagine investing in this hedge fund.
We don't know how to beat the market on a risk-adjusted basis, and we don't know anyone that does know either | --Swedroe | We assume that markets are efficient, that prices are right | --Fama

Johno
Posts: 1883
Joined: Sat May 24, 2014 4:14 pm

Re: QSPIX - thoughts on interesting fund

Post by Johno » Fri Aug 14, 2015 1:40 pm

The general theme of 'let's wait and see' also has limited logic to it, even before the time span got to 15 or 20yrs and added in the imaginary future VQSPIX fund which is going have an ER of 0.20% :D . Back testing says this array of strategies tends to work. That doesn't mean they will work in the future, although OTOH 'data mining' is a overly or loosely used term which doesn't apply to any and all cases of referencing historical results. If some of the strategies in QSPIX are just a product of 'data mining', you could say the same thing about any particular level, especially a pleasing and remunerative level, of the equity risk premium itself. If you say 'data mining' you should come armed with specific riposte to the statistics in question. If you mean 'the distribution of returns is not stationary so things that worked in past might not in future', say so, perhaps more succinctly, but 'data mining' is a malapropism if all you mean is 'past performance doesn't gtee future results'.

So the next 15-20yrs of these strategies working, if so, will mean relatively little more for the following 15-20 than the back testing means now for the coming 15-20yrs. It won't mean zero if one believes AQR is cooking the books on the backtesting, since if so they'd be less likely to get away with presenting cooked fund results in 15 or 20yrs. But I don't personally believe they're cooking the back testing.

These are a diversifed set of persistent anomalies invested in via long/short with leverage (by their nature). They will persist or not in terms of generating net return. But still nobody has given a good argument (just repeating the letters L, T, C and M endlessly is not it) why this fund would become highly correlated with the equity risk premium, and the basic idea is to diversify from the ERP as a source of return, not to beat the ERP in level of return. Though that does still relate to the perennial debate about E[r] of equities. The periodic bump of the thread with quotes from Bogle appearing to say 'don't do this' are typically by people who don't accept the basic concept of equity risk premium as just one source of risky return you might not want to rely too heavily on. They tend to see the ERP *the* source of risky return, and also tend to implicitly set E[r] of stocks as the (unusually) excellent realized return of US equities over the last 100yrs or so. They don't accept the basic argument that would make one even look for alternative sources of risky return in the first place.

lack_ey
Posts: 5542
Joined: Wed Nov 19, 2014 11:55 pm

Re: QSPIX - thoughts on interesting fund

Post by lack_ey » Fri Aug 14, 2015 2:02 pm

afan wrote:One post shy of 1,000.

Among the many problems with this fund- the proponents argue that it helps smooth out the volatility of your overall portfolio due to its low correlation with the rest of your assets. it also has low correlation with the rest of your financial life. But since it varies its factor bets, and is free to drop some and adopt others, you never know what the factor composition of the portfolio will be. Perhaps it was a good counterbalance to the rest of your life, for some period of time, with some combination of bets. Then that changes. Of course, you don't even know about the change, let alone how the new combo fits with your portfolio.

How can you construct a portfolio like this? You put money in, say, stocks because you think they make a favorable long term contribution. Their returns are very noisy, but you think they have good risk/return characteristics and you want some exposure.

Now you have a fund whose investments are quite literally unknown. Whatever they may be now WILL change and you do not know how. How can you possibly decide what to do with it? I have an IPS that allocates X to stocks, Y to fixed income and Z to the mystery investment?? How much should Z be? 1%? 10%? 100%? What basis could there be for a decision to make an allocation to something when I don't know what it is?

Even if the fee were 20bp, it is hard to imagine investing in this hedge fund.

It's not a hedge fund. It's a mutual fund and much more heavily regulated under all the applicable rules. This includes not being able to drastically change the strategy and objectives as outlined in the prospectus without you knowing.

Regardless of what the underlying holdings are at any given time, anyone should be able to plan around the property of this usually having low correlation to all the asset classes.

If you have some very vague idea of the risk and return and relationship to other assets (or in the extreme, the return distributions, covariances, etc.), you can apply the usual modern portfolio theory constructs to figure out asset weightings, or any other kind of portfolio analysis and construction method you believe in. If that results in a zero weighting here, then sure, move on. Same with everything else.

User avatar
oneleaf
Posts: 2246
Joined: Mon Feb 19, 2007 5:48 pm

Re: QSPIX - thoughts on interesting fund

Post by oneleaf » Fri Aug 14, 2015 2:05 pm

If Vanguard ever offered it at a low enough minimum, I would commit 5% or more of my portfolio. Having read Ilmanen's book and looked over AQR's papers, I strongly believe in the diversification benefits, despite the fees. My biggest hesitation is that AQR ever closes it and there is no alternative to replace it with.

I refuse to move my assets back to Fidelity since I already went through the hassle of moving it all FROM Fidelity to Vanguard. Consolidation of assets is too important to me to go after this fund, but I will keep an eye on whether it becomes more widely available.

User avatar
grap0013
Posts: 1861
Joined: Thu Mar 18, 2010 1:24 pm

Re: QSPIX - thoughts on interesting fund

Post by grap0013 » Fri Aug 14, 2015 2:07 pm

Johno wrote:The general theme of 'let's wait and see' also has limited logic to it, even before the time span got to 15 or 20yrs and added in the imaginary future VQSPIX fund which is going have an ER of 0.20% :D . Back testing says this array of strategies tends to work. That doesn't mean they will work in the future, although OTOH 'data mining' is a overly or loosely used term which doesn't apply to any and all cases of referencing historical results. If some of the strategies in QSPIX are just a product of 'data mining', you could say the same thing about any particular level, especially a pleasing and remunerative level, of the equity risk premium itself. If you say 'data mining' you should come armed with specific riposte to the statistics in question. If you mean 'the distribution of returns is not stationary so things that worked in past might not in future', say so, perhaps more succinctly, but 'data mining' is a malapropism if all you mean is 'past performance doesn't gtee future results'.

So the next 15-20yrs of these strategies working, if so, will mean relatively little more for the following 15-20 than the back testing means now for the coming 15-20yrs. It won't mean zero if one believes AQR is cooking the books on the backtesting, since if so they'd be less likely to get away with presenting cooked fund results in 15 or 20yrs. But I don't personally believe they're cooking the back testing.

These are a diversifed set of persistent anomalies invested in via long/short with leverage (by their nature). They will persist or not in terms of generating net return. But still nobody has given a good argument (just repeating the letters L, T, C and M endlessly is not it) why this fund would become highly correlated with the equity risk premium, and the basic idea is to diversify from the ERP as a source of return, not to beat the ERP in level of return. Though that does still relate to the perennial debate about E[r] of equities. The periodic bump of the thread with quotes from Bogle appearing to say 'don't do this' are typically by people who don't accept the basic concept of equity risk premium as just one source of risky return you might not want to rely too heavily on. They tend to see the ERP *the* source of risky return, and also tend to implicitly set E[r] of stocks as the (unusually) excellent realized return of US equities over the last 100yrs or so. They don't accept the basic argument that would make one even look for alternative sources of risky return in the first place.


+1 Best post in this thread. Well said.
There are no guarantees, only probabilities.

afan
Posts: 2849
Joined: Sun Jul 25, 2010 4:01 pm

Re: QSPIX - thoughts on interesting fund

Post by afan » Fri Aug 14, 2015 2:28 pm

lack_ey wrote: If you have some very vague idea of the risk and return and relationship to other assets (or in the extreme, the return distributions, covariances, etc.)....


That is precisely the problem. It seems they promise you that the composition will change, so how could you possibly have an idea of risk and return relationships, return distributions, or anything else? You could have it for some set of bets, but if the factor bets change, these expectations will change.
Last edited by afan on Fri Aug 14, 2015 2:55 pm, edited 1 time in total.
We don't know how to beat the market on a risk-adjusted basis, and we don't know anyone that does know either | --Swedroe | We assume that markets are efficient, that prices are right | --Fama

afan
Posts: 2849
Joined: Sun Jul 25, 2010 4:01 pm

Re: QSPIX - thoughts on interesting fund

Post by afan » Fri Aug 14, 2015 2:50 pm

grap0013 wrote: If you say 'data mining'


I really mean "data mining". Looking backward to find a set of relationships that exist over a certain data set, but which may just as easily not be found in other sets. For finance, this typically means out of sample, which is hard to come by. Hence the solution of waiting a long time.

Data mining is much worse as the number of variables increases. Postulating a continuation of the market premium does not require assumptions about persistent relationships among a large number of variables. The whole point of data mining is that it is a concern when one has looked at a large number of variables, excluded an unknown number, and "found" an interesting set of relationships among those that survived the screens. If you had proof the relationships were artifacts, then the question would never come up. The challenge is deciding how likely it is that the results will withstand test on out of sample data before those data are available. In many circumstances, one is forced to make a decision before the oos analysis can be done. In the case of this fund, there is no compulsion at all to do anything. Perhaps make a note to take another look in 10-20 years. If it works over that time- great! But life will go on if it is a reliable strategy and I am late to the party.

grap0013 wrote:So the next 15-20yrs of these strategies working, if so, will mean relatively little more for the following 15-20 than the back testing means now for the coming 15-20yrs.


There I have to disagree. Particularly if there are numerous funds over the next few decades to reduce the effect of individual managers, one might well have enough out of sample data to make conclusions about whether the original results were the product of "data mining" by the conventional definition.

Misleading results from backtesting do not require dishonesty by the investigators. It happens all the time by accident when people are trying as hard as they can to avoid it. It happens all the time when people are not getting paid massive sums to come up with a reason for others to invest with them. In addition to the purely statistical problem with data mining, this particular sort of back testing requires that you actually know how the real world investment strategy would work in detail, even though no one had been doing it. You have to assume you know about all the odd covariances that blew up LTCM. People keep bringing them up because they thought they knew how to take their backtests and run them forward. They thought they knew the covariances. They were wrong. That is why you need oos data.

You have to assume you know the full variance covariance matrix of this approach with essentially anything that might be correlated with the markets somewhere. For this fund, since they promise to vary their factor bets not only among the current ones, but adding new ones in the future, you have to assume that you know what the matrix will look like when you start betting on factors that you currently do not even know exist. Just as the practice referred to above is commonly known as "data mining" this knowledge of the behavior of unknown factors is commonly known as "magic."

grap0013 wrote: why this fund would become highly correlated with the equity risk premium


I am not concerned that it would do that. It is easy to ensure that it does not. Straightforward futures and options can take that correlation down to any arbitrary level. My concern is that it is impossible to use even backtesting to predict what will be the performance of the strategy in the future, since the strategy itself will change. Even if they got lucky and the backtesting was not spurious, they promise that they will not keep doing the same thing they backtested.

My concerns have nothing to do with the equity risk premium. Whatever it turns out to be in the future has nothing to do with the challenge of guessing the contribution of a fund that is based on backtesting a complex strategy, with unknown number of alternative strategies that were tested and rejected, and which will change in the future. Equities could be great or terrible going forward, but that has nothing to do with how one could determine a role for a fund like this.
We don't know how to beat the market on a risk-adjusted basis, and we don't know anyone that does know either | --Swedroe | We assume that markets are efficient, that prices are right | --Fama

User avatar
tarheel
Posts: 304
Joined: Fri Sep 17, 2010 8:33 am

Re: QSPIX - thoughts on interesting fund

Post by tarheel » Sat Aug 15, 2015 5:55 am

Angst wrote:
tarheel wrote:
Angst wrote:I'm in packer16's camp here: my "Show Me" sensibilities have yet to be sated and my "too good to be true" radar flickers...

I can't help having the impression that many AQR boosters posting here (save Larry, perhaps) have not fully bought into their future with QSPIX - i.e. I sense a lot of dipping of one's toes into the water but not genuine commitment.

If QSPIX is really "real", then for someone who's already "won the game" and believes in QSPIX, rather than going in the direction of the Larry Portfolio for example, shouldn't they be apt to choose something more along the lines of perhaps 75% QSPIX and 25% Treasuries? Just throwing out numbers here of course, but generally speaking, isn't QSPIX pretty much the holy grail for those who've already won the game?

As it is for now, I'm in no more rush to commit to QSPIX than I am for any bleeding edge consumer software or hardware out there in the marketplace. What's a year or two or five in a lifetime of investing? I try to base my decisions more on my confidence in their outcomes rather than on my fear of regret for not having committed as soon as possible. If QSPIX is real, it will continue to exist and will come down in price. And I will be watching!

My IPS has 7.5% allocation to QSPIX with at least a permanent 10% allocation long term. If that's not commitment I don't know what is.

Thanks for the info. I too would consider 10% to be a commitment, even just 7.5%. Of the "many AQR boosters posting here", I think that's exceptional. According to your IPS, when will 7.5% change to 10%? Is that based on an age milestone? Retirement? SS/pension?


Age 40 milestone.....I've been tinkering way too much lately and I am at a place now where I am happy with the way things are.

5% Vanguard 500
1.25% Vanguard Extended Market
6.25% Vanguard Total International
15% Total Bond Market
(I have to have these based on my screwy 401k rules, but I'm happy with them)
17.5% QSMLX
11.25% QCELX
11.25% QICLX
9% QEELX
8.5% VSS
7.5% FSRVX
7.5% QSPIX

Maybe I'm AQR fan #1! Letting this ride for the next five years and then slowly lowering equity % glide path

rpj2004
Posts: 18
Joined: Sat Feb 27, 2010 4:17 pm

Re: QSPIX - thoughts on interesting fund

Post by rpj2004 » Sat Aug 15, 2015 7:16 am

Johno wrote:These are a diversifed set of persistent anomalies invested in via long/short with leverage (by their nature). They will persist or not in terms of generating net return. But still nobody has given a good argument (just repeating the letters L, T, C and M endlessly is not it) why this fund would become highly correlated with the equity risk premium, and the basic idea is to diversify from the ERP as a source of return, not to beat the ERP in level of return. Though that does still relate to the perennial debate about E[r] of equities. The periodic bump of the thread with quotes from Bogle appearing to say 'don't do this' are typically by people who don't accept the basic concept of equity risk premium as just one source of risky return you might not want to rely too heavily on. They tend to see the ERP *the* source of risky return, and also tend to implicitly set E[r] of stocks as the (unusually) excellent realized return of US equities over the last 100yrs or so. They don't accept the basic argument that would make one even look for alternative sources of risky return in the first place.


Correlation to equity is only one risk though. Due to leverage, the internal fund correlation of the four factors matters also. If all four factors do badly at the same time, the fund fails. Sure, from back testing it appears that these four factors never correlate during downturns. But as others have pointed out, the past data has been specifically mined for design. Looking at the back tested returns for the factors, there are definitely periods where they all do correllate during boom periods. Has AQR really found the four mysterious factors that can only correlate during upswings but not downswings? I think not. The next black swan event could hit this fund, and the fund would die.

That said, if there really was a way to capture these factors without leverage, I might consider it.

User avatar
czeckers
Posts: 901
Joined: Thu May 17, 2007 3:49 pm
Location: Upstate NY

Re: QSPIX - thoughts on interesting fund

Post by czeckers » Sat Aug 15, 2015 7:32 am

The problem with using backrest results is that they are heavily skewed by recent performance. Any asset class with recent good performance will look like a good diversifier.

Over the years I've seen here heated discussions about CCFs, gold, precious metals mining funds, emerging market funds, small cap value, and REITs all touted as great diversifiers and even being worth the extra ERs. Uniformly, these periods have then been followed by some pretty dismal performance.

I'm not saying that they aren't good diversifiers, but the price you buy them does matter. Diversification goes both ways. For example: VG's VGPMX (precious mining fund) has been very negatively correlated with VTSMX (US Total Stock Market). VTSMX has gone up substantially in the last 5 years while VGPMX has lost over 80% of its value. Yet, people were much more excited about its diversifying benefits at $38 than they are now at $7.

Remember that past performance is not predictive of future returns.
The Espresso portfolio: | | 16% LCV, 16% SCV, 16% EM, 8% Int'l Value, 8% Int'l Sm, 8% US REIT, 8% Int'l REIT, 20% Inter-term US Treas | | "A journey of a thousand miles begins with a single step."

User avatar
Maynard F. Speer
Posts: 2139
Joined: Wed Mar 18, 2015 10:31 am

Re: QSPIX - thoughts on interesting fund

Post by Maynard F. Speer » Sat Aug 15, 2015 8:17 am

We're talking about backtesting a lot, but with QSPIX we're talking about some very well known, fairly universal factors, that we were debating in the 70s and 80s, that we do have a lot of out-of-sample data for now, and that don't seem to go away

Unlike a quant hedge fund algorithm based on trading volumes and lunar cycles, it would presumably take an enormous fund to wipe out factors like value and momentum - although these factors routinely take 7-10 year holidays, and that might be partly why they persist

I have been sceptical of ETF-style hedge fund strategies - I've not seen it work persistently before .. The closest to QSPIX would presumably be a managed futures ETF? (where you've got systematic trend-following strategies) .. Perhaps the improved diversification of QSPIX will avoid these long periods of lacklustre performance most systematic strategies tend to go through - big hedge funds often run half-a-dozen strategies at once, completely independently ... I'd certainly put some money in QSPIX if it were available in the UK - but I'd probably want to hold about 6 similar funds (which is what I do with hedge funds) so I can monitor how they're performing and control exposure accordingly
"Economics is a method rather than a doctrine, an apparatus of the mind, a technique of thinking, which helps its possessor to draw correct conclusions." - John Maynard Keynes

Johno
Posts: 1883
Joined: Sat May 24, 2014 4:14 pm

Re: QSPIX - thoughts on interesting fund

Post by Johno » Sat Aug 15, 2015 4:32 pm

afan wrote:
grap0013 wrote: If you say 'data mining'


1. I really mean "data mining". Looking backward to find a set of relationships that exist over a certain data set, but which may just as easily not be found in other sets. For finance, this typically means out of sample, which is hard to come by. Hence the solution of waiting a long time.

grap0013 wrote:So the next 15-20yrs of these strategies working, if so, will mean relatively little more for the following 15-20 than the back testing means now for the coming 15-20yrs.


2. There I have to disagree. Particularly if there are numerous funds over the next few decades to reduce the effect of individual managers, one might well have enough out of sample data to make conclusions about whether the original results were the product of "data mining" by the conventional definition.

3. You have to assume you know about all the odd covariances that blew up LTCM. People keep bringing them up because they thought they knew how to take their backtests and run them forward. They thought they knew the covariances. They were wrong. That is why you need oos data.

grap0013 wrote: why this fund would become highly correlated with the equity risk premium


4. I am not concerned that it would do that. It is easy to ensure that it does not. Straightforward futures and options can take that correlation down to any arbitrary level. My concern is that it is impossible to use even backtesting to predict what will be the performance of the strategy in the future, since the strategy itself will change.

5. My concerns have nothing to do with the equity risk premium. ...Equities could be great or terrible going forward, but that has nothing to do with how one could determine a role for a fund like this.

The quoted post was mine not grap0013's response, just to clarify.

1. I don't think that changes my point. If you meant by 'data mining' simply 'past performance doesn't gtee futures results', which is actually implied by a lot of your other statements, that's obvious. The world and process of returns can change. If you mean 'data mining' in the sense of flawed over-fitted statistics, you haven't provided any evidence. A statement like that can be debated in terms of the actual stats presented, and other threads or maybe even this one way back, have gone over the actual results. I don't think it was convincingly shown they constituted 'data mining' in the proper sense. I don't recall any real attempt. And you've provided nothing but a claim.

2. If numerous other funds arbitrage away premia which previously actually existed, established themselves because the premia actually existed, that's more evidence you really only mean 'past performance is not gtee' when you say 'data mining', whatever you think you mean. But aside from that I don't see much logic in that response. If anything it cuts against what you've said otherwise. If the premia exist and have been 'too large', that's all the more reason it makes no particular sense to wait another 15-20 yrs. Now there's backtesting data, which you've done nothing to actually demonstrate is 'data mining'. If we assume AQR's honesty, that's *at least* as good evidence the strategies will work in 2015-2030 as the funds good performance in 2015-2030, if so, will be of its likely performance in 2030-2045. Again, the 'wait 15-20 yrs tangent' came from QSPIX skeptics.

3. There really should be an automatic penalty for blind out of context reference to LTCM, kind of like whoever compares to Hitler first loses. :D Seriously, LTCM had huge short positions in equity option volatility and risk markets went south. There is no big mystery there.

4. There is no way to be sure the strategies will continue to work because the world and thus process of returns generation may change. This is true of any risky investment. However you keep repeating 'but the strategies may change'. But as has been pointed out, it's a mutual fund. If the strategy basically changes they need to report that. Otherwise I may not be alone in being puzzled what exactly you mean by that repeated statement.

5. It has to do with the equity risk premium because a fund such as this is only rationally considered in the context of a whole portfolio. Either the investor seeks all his/her return (except the minuscule real return of safe bonds) in the equity risk premium, or attempts to diversify among sources of return not completely correlated with the ERP. And in reality an alternative like this would never replace a very large % of the risk budget committed to the ERP. So the issue is extremely high or just high reliance on the ERP. I see clear benefit in reducing dependence on the ERP. Whether or not I can find true lack of correlation to the ERP in investments I'm pretty sure also have real return (eg. my direct real estate investments) or real return in investments I'm pretty sure are uncorrelated with the ERP (eg. QSPIX) is the uncertainty. But I think this statement of yours 'it has nothing to do with the ERP' with due respect shows a basic missing of the point.

afan
Posts: 2849
Joined: Sun Jul 25, 2010 4:01 pm

Re: QSPIX - thoughts on interesting fund

Post by afan » Sat Aug 15, 2015 8:52 pm

Johno wrote: If you mean 'data mining' in the sense of flawed over-fitted statistics, you haven't provided any evidence.


I am not sure you get what "data mining" means.

It is not an insult. It is not a careless way of saying one does not like something. It has nothing to do with flawed statistics (unless someone presents spurious significance metrics).

The problem is that you can always find relationships with enough backtesting. If you take an introductory course in statistics, you will do an exercise in which you generate numerous sets of random numbers, then run a multiple regression of combinations of the random data sets against yet another random data set. By hunting through this, you will find some high correlations. These are not "flawed". You did not do it wrong. This is what happens when you runs lots of tests on the same data sets, ignore the the runs that do not produce an interesting result and cull out only the ones that do. Basic statistics.

As for an "argument", as I indicated. the only way to know whether you have this situation is to run the analysis on out of sample data. Here, we have no out of sample data. Hence the need to accumulate some. My estimate of 10 to 20 years is only a guess. Longer if we are dealing with only one fund. Perhaps less if there are lots of funds.

As I keep pointing out, the managers of this fund promise to keep doing backtests and keep changing the factor bets. If they do this, then there may never be a long enough period with a stable set of bets to determine whether the strategy is working. You MIGHT decide the fund did well over a long period of time, or decide it did poorly, but that would be a test of the success of the active management choices of factor bets, along with whatever else they may have done.

We can discuss later what "did well" means when they are careful to indicate that Sharpe ratio is irrelevant. You might call "did well" to be "high volatility with low correlation with the return sequences of stocks, bonds or any other asset". Or "a volatile, random, return sequence".

They are producing a standard multiple testing. false discovery problem. It would certainly come up in an introductory course in statistics, see above, and approaches to it remain under active study.

Campbell Harvey has worked on it specifically in the economic context.

"Multiple Testing in Economics"
Campbell R. Harvey
Yan Liu∗

http://poseidon01.ssrn.com/delivery.php ... 90&EXT=pdf
We don't know how to beat the market on a risk-adjusted basis, and we don't know anyone that does know either | --Swedroe | We assume that markets are efficient, that prices are right | --Fama

lack_ey
Posts: 5542
Joined: Wed Nov 19, 2014 11:55 pm

Re: QSPIX - thoughts on interesting fund

Post by lack_ey » Sat Aug 15, 2015 9:15 pm

So which factors are data mined, or more generally, which do you have concerns about robustness and legitimacy?

The point is that the styles chosen are decades old, pointed out and even exploited with live trading for years (perhaps not with some of the specific style/asset class combinations but for a good deal of them). For example, value in equities goes back way before 1980, the start of the composite simulation they did. Even in the Fama-French sense it has been tested out of sample now for a while and in multiple world equity markets.

The specific composite mix of styles at any given time to not vary all that much to make a difference in these types of broader concerns of data mining. Let's say they dynamically increase the weighting of momentum in commodities from X to 1.2X. That's not going to make or break the fund. What matters is if the style works in the first place at all.

afan
Posts: 2849
Joined: Sun Jul 25, 2010 4:01 pm

Re: QSPIX - thoughts on interesting fund

Post by afan » Sun Aug 16, 2015 8:06 am

lack_ey wrote:
The point is that the styles chosen are decades old, pointed out and even exploited with live trading for years (perhaps not with some of the specific style/asset class combinations but for a good deal of them).


Not exactly. that is true for SOME of them, and as time has gone by the success of these has varied, perhaps gone away. Other factors are quite new, identified with backtesting and have little or no record of real world application.

lack_ey wrote: For example, value in equities goes back way before 1980, the start of the composite simulation they did. Even in the Fama-French sense it has been tested out of sample now for a while and in multiple world equity markets.


Yes, and even F&F now have found that new factors subsume their original factors.

lack_ey wrote:The specific composite mix of styles at any given time to not vary all that much to make a difference in these types of broader concerns of data mining. Let's say they dynamically increase the weighting of momentum in commodities from X to 1.2X. That's not going to make or break the fund. What matters is if the style works in the first place at all.


Only data will tell how much difference variation in the weights will matter going forward.

And again, they PROMISE to change the factors and weights in the future. This seems to require one to believe that there are NO changes in factors or weights that would change the success of the approach. That is quite an optimistic assumption.
We don't know how to beat the market on a risk-adjusted basis, and we don't know anyone that does know either | --Swedroe | We assume that markets are efficient, that prices are right | --Fama

User avatar
packer16
Posts: 859
Joined: Sat Jan 04, 2014 2:28 pm

Re: QSPIX - thoughts on interesting fund

Post by packer16 » Sun Aug 16, 2015 8:54 am

matjen wrote:
packer16 wrote:To date, QSPIX has delivered below equity market returns and the difference has all gone to the asset manager, so you have paid 100% of the outperformance to AQR, unless you put this into something other than a equity or risk category in your portfolio (if you do this IMO you are stretching the utility of this for simple folks like me). At this point it is show me fund.

Packer


NO, NO, NO. Come on Packer! As Tarheel said (as have countless others over, and over and over and over) this fund isn't expected to outperform equities. Regardless, your ability to discuss things after the fact is absurd and disingenuous. It's like someone coming on this board and saying indexing is for losers because he picked Apple, Facebook, and Home Depot the last 5 years. We have gone through this multiple times. Spare me your Sequoia or Total US (Didn't Lack_ey call you on the US v. Int'l distinction just a handful of posts ago?). Why don't you pick International, Emerging Markets or some other slice of equities? We both know the answer don't we. And how many times do you have to be reminded that QSPIX isn't considered equity. You can consider it all you want to be equity but no one else does. It's like me saying Total Bond should be equity because 50% of it is comprised of nutty stuff like securitized and corporates. In my personal definition of bonds only US Treasuries/TIPS count....Discussing generally accepted things and changing the definitions along your own personal whims isn't very fruitful it seems to me.

Again, to the extent you are going to even suggest QSPIX be compared to equities (for giggles only I might add) the only fair benchmark would be total world. In that context QSPIX is actually outperforming.

As we used to say at high school basketball games. SCOREBOARD, SCOREBOARD, SCOREBOARD*...QSPIX, Total World, Total Bond, And Vanguard Market Neutral.

Image

*It should be pointed out that I fully expect VT to pull ahead eventually and that QSPIX should be viewed in the context of an entire portfolio not on its own...but this is fun. ;-)

The hilarious part about all this is that "if" QSPIX performs well over 10+ years and we have this discussion, folks like you and Homer and longinvest will then insist that it is too late and that the market will correct for the now known strategy...


As to comparing to equity benchmarks, maybe I missed something but in Larry's article about QSPIX it clearly states that the expected return is equity-like (http://www.advisorperspectives.com/arti ... th-style/3), not balanced fund returns. So far the returns are in between balanced and equity. This makes a significant difference as the fees/expected returns with a 5% equity returns is 30% versus 40% for a balanced fund versus lets say 20% for Sequoia. Sequoia BTW has delivered about 2% over market returns (similar to PRIMECAP). If QSPIX does deliver equity type returns and does not blow up in the next crisis I would be very much interested in investing in it. We will see and I wish you well with QSPIX.

Packer
Buy cheap and something good might happen

lack_ey
Posts: 5542
Joined: Wed Nov 19, 2014 11:55 pm

Re: QSPIX - thoughts on interesting fund

Post by lack_ey » Sun Aug 16, 2015 9:33 am

afan wrote:
lack_ey wrote:
The point is that the styles chosen are decades old, pointed out and even exploited with live trading for years (perhaps not with some of the specific style/asset class combinations but for a good deal of them).


Not exactly. that is true for SOME of them, and as time has gone by the success of these has varied, perhaps gone away. Other factors are quite new, identified with backtesting and have little or no record of real world application.

You're still not being specific. Which ones are "some" and which ones are "others"? Also, even if one style is a dud, do you think the expected return (before transaction costs) to be sharply negative? If the other three work, the whole will do fine.

afan wrote:
lack_ey wrote: For example, value in equities goes back way before 1980, the start of the composite simulation they did. Even in the Fama-French sense it has been tested out of sample now for a while and in multiple world equity markets.


Yes, and even F&F now have found that new factors subsume their original factors.

This is irrelevant. Ignoring for now the fact that these are just models without the ability to explain all the data, and that a five-way decomposition is going to cover more space than a three-way analysis, this doesn't mean that there's anything wrong with value. If you reject the old model and take the new one as gospel, a fund that misguidedly loads on value will end up having weights in the five true factors. It may not quite be as effective as trying to load on the five factors directly, but a deviation that causes a positive loading is supposed to give you better returns over the long run, and if it still does, it doesn't matter if you achieved it through running p/b and other value screens or if you didn't.

afan wrote:
lack_ey wrote:The specific composite mix of styles at any given time to not vary all that much to make a difference in these types of broader concerns of data mining. Let's say they dynamically increase the weighting of momentum in commodities from X to 1.2X. That's not going to make or break the fund. What matters is if the style works in the first place at all.


Only data will tell how much difference variation in the weights will matter going forward.

And again, they PROMISE to change the factors and weights in the future. This seems to require one to believe that there are NO changes in factors or weights that would change the success of the approach. That is quite an optimistic assumption.

If they make a big change like adding or dropping another style, as mentioned previously, we will know about it. Even supposing these guys could keep themselves from talking or writing papers about it, a big change would have to be disclosed in fund paperwork.

It doesn't matter right now if they actually only stick to the current strategy as described for another five years. You can always sell if you think the change is a bad idea. But it's not like anybody knows with certainty that the current construction will deliver good results in the long term, and the same can be said about anything else they might potentially change to in the future. At all times, you can only look at all available information to date.

longinvest
Posts: 2447
Joined: Sat Aug 11, 2012 8:44 am

Re: QSPIX - thoughts on interesting fund

Post by longinvest » Sun Aug 16, 2015 9:51 am

packer16 wrote:As to comparing to equity benchmarks, maybe I missed something but in Larry's article about QSPIX it clearly states that the expected return is equity-like (http://www.advisorperspectives.com/arti ... th-style/3), not balanced fund returns. So far the returns are in between balanced and equity. This makes a significant difference as the fees/expected returns with a 5% equity returns is 30% versus 40% for a balanced fund versus lets say 20% for Sequoia. Sequoia BTW has delivered about 2% over market returns (similar to PRIMECAP). If QSPIX does deliver equity type returns and does not blow up in the next crisis I would be very much interested in investing in it. We will see and I wish you well with QSPIX.


Wow! What an incredible claim. Packer, you are right: he actually wrote that! Let me cite the text:

Larry Swedroe wrote:Using QSPIX

The equity-like expected return (about 7% net of fees) makes QSPIX an excellent diversifier, reducing the tail risk relative to long-only portfolios. If an investor uses the fund to reduce their allocation to the equity side of their portfolio, the expectation should be for an increased Sharpe ratio – similar portfolio returns with lower portfolio volatility (risk).


Let's look at a growth chart, just to verify this:

Morningstar: AQR Style Premia Alternative I (QSPIX) blue, Vanguard Total Stock Mkt Idx Adm (VTSAX) orange
Image

The only thing that is sure, with QSPIX, is that the investor will give at least 1.5% of his investment to AQR every single year, year after year after year.

You can believe the hype that only journalists can claim (as the SEC would probably sue AQR for making such claims) at the risk of your own wealth!
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VAB/ZRR

larryswedroe
Posts: 15597
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: QSPIX - thoughts on interesting fund

Post by larryswedroe » Sun Aug 16, 2015 10:14 am

Exactly, based on the historical premiums and the leverage applied the expected return is equity like with about half the volatility of equity. The return is expected to be more than bond like with about 2x volatility of a high quality bond fund. But of course that is no guarantee, any more than one saying the expected return to stocks is say 6.5% nominal with about 20% vol which is based on historical evidence and current valuations.
Without the leverage the returns would not be equity like, but then the volatility would also be lower.
The evidence supporting the premiums is not only persistent and pervasive across time and asset classes but comes along with logical explanations for each of the premiums, either risk or behavioral.
Now the future might look different, just as it might look very different for US stocks. Since we don't know what the future holds IMO its prudent to diversify across factors that have shown persistent and pervasive premiums.
Anyone investing in the fund so far I'm confident has been more than pleased with the results. Time will tell if that is the case going forward.

User avatar
packer16
Posts: 859
Joined: Sat Jan 04, 2014 2:28 pm

Re: QSPIX - thoughts on interesting fund

Post by packer16 » Sun Aug 16, 2015 12:52 pm

I was also surprised when I read the annual report and it looks the drag on performance was due to passive exposure to factors but the gains was due to "trading". Here is an excerpt from the 2014 annual report:

Performance was mostly positive across asset groups with fixed income and equity indices experiencing losses. Stock trading produced the best results, contributing almost 100% to the positive performance for the period. Stock trading benefited from Defensive and Value themes, while Momentum was flat. The next-best contributor was commodities, followed by currencies trading. In commodities, both Value and Momentum offered positive contributions. The biggest asset group detractor was fixed income, driven by Value and Carry. Commodities and currencies, which had positive contributions for the year, roughly offset the negative performance from equity indices and fixed income.

With 145% turnover and hundreds if not thousands of individual positions how can anyone have confidence these guys know what these underlying positions are? This is probably why I am not interested in investing in hedge funds, momentum mutual funds or other trading vehicles and why an alternative to these would not appeal to me either unless it is based upon something I understand like low price versus value.

Packer
Buy cheap and something good might happen

lack_ey
Posts: 5542
Joined: Wed Nov 19, 2014 11:55 pm

Re: QSPIX - thoughts on interesting fund

Post by lack_ey » Sun Aug 16, 2015 1:07 pm

packer16 wrote:I was also surprised when I read the annual report and it looks the drag on performance was due to passive exposure to factors but the gains was due to "trading". Here is an excerpt from the 2014 annual report:

Performance was mostly positive across asset groups with fixed income and equity indices experiencing losses. Stock trading produced the best results, contributing almost 100% to the positive performance for the period. Stock trading benefited from Defensive and Value themes, while Momentum was flat. The next-best contributor was commodities, followed by currencies trading. In commodities, both Value and Momentum offered positive contributions. The biggest asset group detractor was fixed income, driven by Value and Carry. Commodities and currencies, which had positive contributions for the year, roughly offset the negative performance from equity indices and fixed income.

With 145% turnover and hundreds if not thousands of individual positions how can anyone have confidence these guys know what these underlying positions are? This is probably why I am not interested in investing in hedge funds, momentum mutual funds or other trading vehicles and why an alternative to these would not appeal to me either unless it is based upon something I understand like low price versus value.

Packer

They don't mean "trading" in the sense of flitting in and out of positions and to distinguish from passive (if you call it that) exposures. "Stock trading" means individual stocks, as opposed to country equity indexes. Part of the money is invested in equity index futures, so for example if the S&P 500 has negative momentum and low value relative to the FTSE 100 and Nikkei 225 and so on, they go short S&P 500 futures and long the ones with better attributes according to the criteria. "Stock trading" just means the contribution from the long/short program in individual stocks based on value, momentum, and defensive (carry is skipped for equities, as it overlaps with value there).

I'm pretty sure the managers don't know what the underlying positions are, but why would they need to in order to execute the strategy as described? It's algorithmic at the core. They run the screens on the data and see what needs to be traded. The low-level execution is proprietary and not described, but how and when to get in and out of a position doesn't really depend on whether or not somebody at AQR knows offhand if it's currently long Chipotle or short bunds.

Post Reply