QSPIX - thoughts on interesting fund

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
larryswedroe
Posts: 15598
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: QSPIX - thoughts on interesting fund

Post by larryswedroe » Thu Sep 24, 2015 8:07 am

backpacker
A fund should be expected to have a negative alpha equal to it's ER plus trading expenses.
Note that we don't use the fund, so I don't know much about it, and thus cannot comment. With that said, the fund isn't loading that much either value or momentum
Larry

User avatar
nisiprius
Advisory Board
Posts: 34162
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: QSPIX - thoughts on interesting fund

Post by nisiprius » Thu Sep 24, 2015 9:09 am

larryswedroe wrote:...if DFA US LV set as benchmark the S&P 500 and it outperformed it would you say it beat the proper benchmark? I would hope not...
If someone advocating for some fund by noting it had beaten the S&P 500, yes, I would be skeptical about whether it was doing so by taking on more risk.

The thing that's a little bit different here is that, as the comparison chart with the Fidelity EAFE-tracking index fund shows, it really has hugged its AQR-chosen benchmark index pretty tightly.

I'm just poking around here, trying to understand what AQR can and can't do, what AQR is and isn't doing. This was an opportunity to see what it does in a space where there are funds that are familiar to me. In the case of the AQR fund lineup, it's hard to find direct comparisons with passive funds from other companies, because most of their funds have a fairly unique twist. This is one of the few that invites direct comparison, and the comparison is interesting to me. AQR says that it seeks to outperform a specific index and stay within a specified tracking-error target. That's stronger than what Dimensional claims, for example. So, I plot its return--after 0.9% (class I) or 1.25% (class N) and find that its growth chart, after expenses, is virtually identical to an index fund (Fidelity FSIIX) that tracks the same interest.

Morningstar is showing me a 10-year standard deviation of 20.41 for AQINX compared to 18.80 for FSIIX, so it is, however, taking more risk, which probably would show up if you looked at short-term volatility, although the difference between 20.41 and 18.80 would be hard to see or intuit.

In any case, relative to its goal of outperforming the MSCI EAFE index, and keeping in mind that its after-expenses returns actually have hugged the index pretty closely, there are two possible interpretations.

a) AQR really did exactly what it said its goal was--before expenses. It obtained 1% in extra returns through its management of factor exposures, without departing much from its benchmark index. However, all of those extra returns went to AQR (in the N class; most of them, in the I class) leaving nothing for the investor after expenses. Or,

b) AQR got that extra 1% in the usual boring way, by taking more risk--the factor exposures created in a slightly riskier portfolio--but it is still not passing that extra return on to investors, who get slightly more risk, even if it might not be very noticeable, without the hoped-for extra return.

The implication is that if, instead of the EAFE index they set as a goal and benchmark, one constructed a benchmark that accounted for whatever factor tilts they are using, that benchmark would have higher return than EAFE and the fund would in fact have underperformed the appropriately constructed benchmark.

In their space of unique funds, I can't tell what they're doing. In the space where you can compare them head-to-head against an index fund, they aren't doing anything remarkable after expenses. They've basically given you almost exactly the same performance, over any time period, as the Fidelity index fund. They have made 0.90% or 1.25% for themselves and it hasn't been at your expense--but it hasn't been to your benefit, either.

Reposting the chart: there's no reasonable period of time over which it would have made much difference to the investor whether they were holding AQR's AQINX, blue line, or Fidelity's FSIVX, orange line. AQR's fund after AQR's high expenses pretty much matches Fidelity's Spartan index fund after Fidelity's low expenses.

Image
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.

larryswedroe
Posts: 15598
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: QSPIX - thoughts on interesting fund

Post by larryswedroe » Thu Sep 24, 2015 9:31 am

Nisi
I think that is very fair analysis, based on my very quick look. Note that while we do approve QSPIX with 1.5% ER (though IMO the right way to look at this fund is as 0.75% ER for the long and 0.75 for the short) because it gives you a unique exposure that cannot get elsewhere at lower price and does add diversification benefits and exposure to factors with very strong historical evidence behind them. And we approved the relatively lower cost US large cap style premium fund even though it has bit higher expenses than similar DFA fund we were using. So we look at VALUE ADDED, not slavishly looking only as ER (which IMO can lead to inefficient decisions). Yet we have not approved this fund with it's higher ER. Long only funds with value and at least some momentum screening can be had quite a bit cheaper. Now if this fund would be at say 50bp or so perhaps we would approve it. I'm not involved in the regular analysis of funds until they get to the stage of discussing whether we should approve a fund or not. We have team dedicated to that. So I'm not as familiar with this fund nor most of their other funds.

To me passively managed funds once you own them if they aren't pure index funds they are really their own benchmarks and you should be looking at attribution analysis to make sure deviations from expected outcomes (based on what the factors returned) rather than comparing to some benchmark. If you wanted the benchmark you should invest in the benchmark index fund. That's the way we judge performance.

Larry

User avatar
grap0013
Posts: 1867
Joined: Thu Mar 18, 2010 1:24 pm

Re: QSPIX - thoughts on interesting fund

Post by grap0013 » Thu Sep 24, 2015 9:36 am

grok87 wrote:
nisiprius wrote:QSPIX is one of thirty-three AQR funds, and one of ten AQR "alternative" funds. Larry Swedroe called attention to QSPIX and so far it's been pretty much the only AQR fund to get much attention. Presumably they are all managed by smart people.

They have a lot of flavors that would appeal to various coteries of the "anything-but-cap-weighted" faction, including "defensive" funds, a long-short equity fund, a "market neutral" fund, a "Multi-Strategy Alternative Fund" ("Diversified exposure to nine classic hedge fund strategies"), a "risk-balanced" fund, and many "multi-strategy" funds besides QSPIX.

One of them caught my eye because it was one of the plainest and most directly comparable to a Vanguard fund I personally hold, and thus potentially of direct interest to me. There are two "Benchmark-Oriented Equity Funds: Seek to outperform their benchmarks subject to a specified tracking-error target."

"AQR International Equity I," AQIIX, is an institutional-class fund probably not available to most individual investors, but we'll use it. We'll compare it to the fund I hold, Vanguard's International Stock Index Fund, Admiral shares, minimum $10,000 purchase. The time period is since inception of AQIIX.

Source: Morningstar

It essentially tied the benchmark and Vanguard's fund. It slightly lagged the benchmark index and slightly outperformed the Vanguard fund. The amount of the outperformance relative to Vanguard is about 0.07% annualized.

It's pretty impressive that they can virtually tie the return of the benchmark index after 0.90% expenses, but it pretty much fits a common pattern: yes, it appears that they really might have created alpha before expenses, but they took virtually all of it for themselves and gave none to their investors.

Image

thanks nisi- great post.


grok, I know you are a factor junkie and am surprised you do not like the "pure meth" factor loadings of QSPIX.

I respectfully disagree on Nisi's post. Nobody said all of AQR's funds are magical. Just QSPIX. :wink:

We are talking about QSPIX not AQIIX and those charts do not add value to this discussion. All that matters is QSPIX live returns of investors who are actually holding the fund. Sorry to rub salt in the wound to many people's dismay but it has performed exactly as advertised and despite the 1.5% ER we have more :moneybag overall with an allocation to it.
There are no guarantees, only probabilities.

grok87
Posts: 7575
Joined: Tue Feb 27, 2007 9:00 pm

Re: QSPIX - thoughts on interesting fund

Post by grok87 » Thu Sep 24, 2015 11:13 am

grap0013 wrote:
grok, I know you are a factor junkie and am surprised you do not like the "pure meth" factor loadings of QSPIX.

I respectfully disagree on Nisi's post. Nobody said all of AQR's funds are magical. Just QSPIX. :wink:

We are talking about QSPIX not AQIIX and those charts do not add value to this discussion. All that matters is QSPIX live returns of investors who are actually holding the fund. Sorry to rub salt in the wound to many people's dismay but it has performed exactly as advertised and despite the 1.5% ER we have more :moneybag overall with an allocation to it.


I don't like the 7x leverage. If it were 2x leveraged with a lower e.r. I might consider buying it...
"...people always live for ever when there is any annuity to be paid them"- Jane Austen

User avatar
grap0013
Posts: 1867
Joined: Thu Mar 18, 2010 1:24 pm

Re: QSPIX - thoughts on interesting fund

Post by grap0013 » Thu Sep 24, 2015 12:55 pm

grok87 wrote:
I don't like the 7x leverage. If it were 2x leveraged with a lower e.r. I might consider buying it...


I wouldn't get too hung up on this. If you were levering something with higher baseline volatility then I would be concerned. The factors themselves do not have that much volatility compared to equities and the zero correlation between the factors helps mitigate the leverage and volatility as well. When my equities have a big down day even when QSPIX goes down with them it is not by near as much. This has been very consistent since I've been following this fund. Hence, AQR has done an excellent job of formulaically engineering this fund and I'd focus on the standard deviation of ~10 much more than 2X vs. 7X leverage. It is a better predictor of downside losses IMO.
There are no guarantees, only probabilities.

grok87
Posts: 7575
Joined: Tue Feb 27, 2007 9:00 pm

Re: QSPIX - thoughts on interesting fund

Post by grok87 » Thu Sep 24, 2015 2:53 pm

grap0013 wrote:
grok87 wrote:
I don't like the 7x leverage. If it were 2x leveraged with a lower e.r. I might consider buying it...


I wouldn't get too hung up on this. If you were levering something with higher baseline volatility then I would be concerned. The factors themselves do not have that much volatility compared to equities and the zero correlation between the factors helps mitigate the leverage and volatility as well. When my equities have a big down day even when QSPIX goes down with them it is not by near as much. This has been very consistent since I've been following this fund. Hence, AQR has done an excellent job of formulaically engineering this fund and I'd focus on the standard deviation of ~10 much more than 2X vs. 7X leverage. It is a better predictor of downside losses IMO.

Well let's just say we may be thinking about risk in somewhat different ways.

if you go to the french data library you can get the value and momentum premiums for US stocks. there were 12 cases in the last 88 years when the average of value and momentum was negative- basically 1 in 7.

year mom value =average(val,mom)
2009 -82.91 -5.57 -44.24
2003 -24.47 5.05 -9.71
1939 0.88 -19.17 -9.145
1938 -1.69 -12.18 -6.935
1932 -21.25 10.16 -5.545
1934 17.32 -27.37 -5.025
1975 -18.88 8.9 -4.99
1937 -4.36 -3.97 -4.165
1971 3.55 -11.18 -3.815
1987 -3.68 -1.63 -2.655
1949 0.36 -4.55 -2.095
2014 1.38 -3.17 -0.895

the most recent of these cases was also the worst: in 2009 momentum was -83% and value -6% with the two averaging to roughly -44%.

now obviously QSPIX invests in other asset classes than US stocks and in other strategies than just value and momentum. So hopefully that would mitigate the risk somewhat. hopefully if a bad year like 2009 happens then the funds strategies/investments might lose less than that, maybe only -30% on an unleveraged basis.

now enter leverage:

QSPIX is 7 times levered or 3.5 x per side (i.e. 3.5 long, 3.5 short). So we would need to multiply the loss by 3.5x. now the math here is pretty uncompromising: 3.5 x (-30%) = -105%- i.e. the funds NAV goes to zero.

let's say one thinks the funds diversification lead to even more reduction in downside risk. So instead of -44% or -30%, let's say you think the downside on an unleveraged basis is only say -15%. well 3.5 x (-15%) = -52.5%. or down by just over half.

if even that less extreme scenario happens, i think QSPIX is done, i.e. the fund will be wound down. QSPIX is basically a hedge-fund like investment. It is widely thought in the hedge fund world that if you lose half your investors money you are done. investors figure if you can go down by half, you can lose it all. investors will pull their money, and there will be massive redemptions. at that point QSPIX is a forced seller and the realized loss may be even worse than the mathematical -52.5%.
"...people always live for ever when there is any annuity to be paid them"- Jane Austen

lack_ey
Posts: 5715
Joined: Wed Nov 19, 2014 11:55 pm

Re: QSPIX - thoughts on interesting fund

Post by lack_ey » Thu Sep 24, 2015 3:04 pm

Keep in mind that the 3.5x / -3.5x is not static and is wound down considerably if underlying markets are that volatile to meet the aggregate projected ~10% volatility target. (Possibly too late in some cases, though. In fact, this is one of the risks of the fund, that some other fund that's much more aggressively leveraged is taking similar positions but is forced to unwind, causing a chain reaction of similar funds unwinding as a result of the volatility targeting programs.)

By dollar allocations I'm not even sure the equity program is net 1x / -1x by dollar amount most of the times but would need to check. Recall it's a bit over 1/2 in equities by risk, not by dollar amounts.

Your French data library example of US stocks momentum going -83% is basically what you get if you don't have volatility targeting and dutifully construct UMD by shorting financials with the same allocations all the way through the rebound, and not ever thinking to cut losses somewhere. A cautionary tale but not quite what the fund is reproducing.

I tried hashing out some of these details myself at one point and came to the conclusion that the amount of leverage and volatility targeted is about all I'm comfortable with, getting into the uncomfortable range. This is not like a long-only asset class with some kind of inherent rebound mechanism from drops, either. For example, if you get slaughtered in bonds at least you now have high rates, which is some consolation, especially if inflation hasn't spiked too.

I think that 1x / -1x total would be far too low, though, considering all the bonds and other stuff in the fund. It's too diversified and taking too many kinds of positions, particularly in many not-very-volatile assets, for that to be nearly risky enough, IMHO.

User avatar
matjen
Posts: 1694
Joined: Sun Nov 20, 2011 11:30 pm

Re: QSPIX - thoughts on interesting fund

Post by matjen » Thu Sep 24, 2015 3:16 pm

grok87 wrote:if you go to the french data library you can get the value and momentum premiums for US stocks. there were 12 cases in the last 88 years when the average of value and momentum was negative- basically 1 in 7.

year mom value =average(val,mom)
2009 -82.91 -5.57 -44.24
2003 -24.47 5.05 -9.71
1939 0.88 -19.17 -9.145
1938 -1.69 -12.18 -6.935
1932 -21.25 10.16 -5.545
1934 17.32 -27.37 -5.025
1975 -18.88 8.9 -4.99
1937 -4.36 -3.97 -4.165
1971 3.55 -11.18 -3.815
1987 -3.68 -1.63 -2.655
1949 0.36 -4.55 -2.095
2014 1.38 -3.17 -0.895

the most recent of these cases was also the worst: in 2009 momentum was -83% and value -6% with the two averaging to roughly -44%.

now obviously QSPIX invests in other asset classes than US stocks and in other strategies than just value and momentum. So hopefully that would mitigate the risk somewhat. hopefully if a bad year like 2009 happens then the funds strategies/investments might lose less than that, maybe only -30% on an unleveraged basis.

now enter leverage:

QSPIX is 7 times levered or 3.5 x per side (i.e. 3.5 long, 3.5 short). So we would need to multiply the loss by 3.5x. now the math here is pretty uncompromising: 3.5 x (-30%) = -105%- i.e. the funds NAV goes to zero.

let's say one thinks the funds diversification lead to even more reduction in downside risk. So instead of -44% or -30%, let's say you think the downside on an unleveraged basis is only say -15%. well 3.5 x (-15%) = -52.5%. or down by just over half.

if even that less extreme scenario happens, i think QSPIX is done, i.e. the fund will be wound down. QSPIX is basically a hedge-fund like investment. It is widely thought in the hedge fund world that if you lose half your investors money you are done. investors figure if you can go down by half, you can lose it all. investors will pull their money, and there will be massive redemptions. at that point QSPIX is a forced seller and the realized loss may be even worse than the mathematical -52.5%.


What were the numbers for 2008? Because in 2008 AQR figures it would have been down only 5%. More here:

http://www.forbes.com/sites/phildemuth/ ... en-israel/

Q for Ronen Israel: How would a style premia type of strategy have performed in 2008?

Israel: Based on our research, a broadly diversified, market-neutral style premia strategy would have been down about 5% in 2008. Yet, the strategy would have likely been close to flat in the months of September through December of that year, when markets really sold off, further emphasizing its market neutrality and diversification benefits to a traditional equity portfolio. This market neutrality is further evident when you look at longer term correlations of a broadly diversified, market-neutral style premia strategy to traditional markets. For example, the estimated correlation between such a style premia strategy and the MSCI World Index is 0.01 over the period 1990 through September 2013.

Note that every crisis is different. During the steep equity market sell off from March of 2000 through September 2002, when the equity market was down around 60%, according to our research a broadly diversified, market-neutral style premia strategy would have been up decently, providing diversification to a traditional equity portfolio. Market-neutrality does not mean you make money in every crisis – the word is “neutral” not “short”! But it does mean, unlike most traditional investments, you should excel in some of these periods, and have no tendency to get really killed in them in general. Our tests really bear this out.


Obviously your scenario assumes investors look at the fund on its own and not in the context of the overall portfolio. In your 2009 hypothetical I think the fact the fund held up reasonably well in 2008 would be worth something behaviorally.
A man is rich in proportion to the number of things he can afford to let alone.

grok87
Posts: 7575
Joined: Tue Feb 27, 2007 9:00 pm

Re: QSPIX - thoughts on interesting fund

Post by grok87 » Thu Sep 24, 2015 3:23 pm

matjen wrote:
What were the numbers for 2008? Because in 2008 AQR figures it would have been down only 5%. More here:

http://www.forbes.com/sites/phildemuth/ ... en-israel/


in 2008 mom was +13 and value +1. so for US stocks the average of value/momentum was +7.

here's the full data for US Stocks:
year mom value =average(val,mom)
2009 -82.91 -5.57 -44.24
2003 -24.47 5.05 -9.71
1939 0.88 -19.17 -9.145
1938 -1.69 -12.18 -6.935
1932 -21.25 10.16 -5.545
1934 17.32 -27.37 -5.025
1975 -18.88 8.9 -4.99
1937 -4.36 -3.97 -4.165
1971 3.55 -11.18 -3.815
1987 -3.68 -1.63 -2.655
1949 0.36 -4.55 -2.095
2014 1.38 -3.17 -0.895
1969 9.89 -9.86 0.015
1991 14.79 -14.48 0.155
1999 34.66 -34.29 0.185
2011 7.21 -6.56 0.325
1994 3.19 -0.81 1.19
2010 5.68 -3.29 1.195
1957 9.9 -6.42 1.74
1958 -9.56 13.14 1.79
1942 -15.52 19.87 2.175
1940 5.8 -0.82 2.49
1951 11.62 -5.62 3
2006 -7.74 14.44 3.35
1953 16.32 -8.7 3.81
2013 7.94 0 3.97
1990 17.65 -9.64 4.005
1946 5.31 2.98 4.145
2007 21.54 -12.31 4.615
1988 -5.46 14.88 4.71
2012 1.56 8.03 4.795
1966 10.46 -0.86 4.8
1931 23.96 -14.29 4.835
2004 -0.33 10.16 4.915
1996 6.74 3.21 4.975
1983 -10.14 20.67 5.265
1980 37.1 -25.14 5.98
1960 17.19 -4.67 6.26
1978 12.07 0.52 6.295
1952 9.01 4.05 6.53
1998 23.39 -10.19 6.6
1930 25.75 -12.28 6.735
1967 22.48 -8.92 6.78
2008 13.3 0.83 7.065
1964 4.66 10.09 7.375
1948 11.86 3.52 7.69
1985 14.81 1.37 8.09
1981 -8.32 25.21 8.445
1972 15.57 1.79 8.68
1986 8.38 9.26 8.82
1961 10.75 7.11 8.93
1956 19.04 -1.07 8.985
1974 8.32 10.02 9.17
1970 -3.25 22.46 9.605
1995 17.62 1.65 9.635
1927 23.23 -3.75 9.74
1955 14.05 5.71 9.88
1941 9.14 11.13 10.135
1959 19.23 1.51 10.37
1968 2.25 18.59 10.42
1962 11.97 9.29 10.63
1928 28.21 -6.15 11.03
2001 4.31 18.33 11.32
1947 13.84 9.76 11.8
1989 28.1 -4.27 11.915
2005 14.95 9.18 12.065
1997 11.96 13.15 12.555
1977 17.86 7.47 12.665
1979 27 -1.46 12.77
1944 10.92 15.61 13.265
1945 15.55 11.4 13.475
1963 11.66 15.48 13.57
1992 3.15 24.22 13.685
1965 20.56 7.54 14.05
1984 9.15 19.35 14.25
1976 7.39 24.74 16.065
1929 20.73 11.83 16.28
1954 10.34 25.74 18.04
2002 25.65 10.49 18.07
1935 22.47 14.73 18.6
1936 5.89 35.86 20.875
1950 15.39 27.01 21.2
1993 23.72 18.96 21.34
1973 29.26 18.19 23.725
1982 34.69 13.62 24.155
1933 20.04 28.27 24.155
1943 14.24 38.97 26.605
2000 14.98 39.19 27.085
"...people always live for ever when there is any annuity to be paid them"- Jane Austen

User avatar
matjen
Posts: 1694
Joined: Sun Nov 20, 2011 11:30 pm

Re: QSPIX - thoughts on interesting fund

Post by matjen » Thu Sep 24, 2015 3:43 pm

grok87 wrote:in 2008 mom was +13 and value +1. so for US stocks the average of value/momentum was +7.


Thanks Grok. I am sure AQR has the backtested info for 2009 if they had it for 2008. Bet it wasn't as good either. ;-) Wonder if Larry S. can get that question answered....just to see how realistic your thoughts are.
A man is rich in proportion to the number of things he can afford to let alone.

lack_ey
Posts: 5715
Joined: Wed Nov 19, 2014 11:55 pm

Re: QSPIX - thoughts on interesting fund

Post by lack_ey » Thu Sep 24, 2015 3:55 pm

matjen wrote:
grok87 wrote:in 2008 mom was +13 and value +1. so for US stocks the average of value/momentum was +7.


Thanks Grok. I am sure AQR has the backtested info for 2009 if they had it for 2008. Bet it wasn't as good either. ;-) Wonder if Larry S. can get that question answered....just to see how realistic your thoughts are.

With the painfully obvious caveat that nobody publishes a paper or presents a fund that includes a faceplant somewhere in the backtest,

Image
from the JOIM paper.

The modest dip in carry around the financial crisis may be surprisingly small—it's well known currency carry got hit hard—until you realize there are other asset classes involved.

Stepping back from historical simulations, I agree with grok87's line of analysis, but the right figures and considerations need to be plugged in.

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 » Thu Sep 24, 2015 4:00 pm

grok87 wrote:
matjen wrote:
What were the numbers for 2008? Because in 2008 AQR figures it would have been down only 5%. More here:

http://www.forbes.com/sites/phildemuth/ ... en-israel/


in 2008 mom was +13 and value +1. so for US stocks the average of value/momentum was +7.

here's the full data for US Stocks:
year mom value =average(val,mom)
2009 -82.91 -5.57 -44.24
2003 -24.47 5.05 -9.71
1939 0.88 -19.17 -9.145
1938 -1.69 -12.18 -6.935
1932 -21.25 10.16 -5.545
1934 17.32 -27.37 -5.025
1975 -18.88 8.9 -4.99
1937 -4.36 -3.97 -4.165
1971 3.55 -11.18 -3.815
1987 -3.68 -1.63 -2.655
1949 0.36 -4.55 -2.095
2014 1.38 -3.17 -0.895
1969 9.89 -9.86 0.015
1991 14.79 -14.48 0.155
1999 34.66 -34.29 0.185
2011 7.21 -6.56 0.325
1994 3.19 -0.81 1.19
2010 5.68 -3.29 1.195
1957 9.9 -6.42 1.74
1958 -9.56 13.14 1.79
1942 -15.52 19.87 2.175
1940 5.8 -0.82 2.49
1951 11.62 -5.62 3
2006 -7.74 14.44 3.35
1953 16.32 -8.7 3.81
2013 7.94 0 3.97
1990 17.65 -9.64 4.005
1946 5.31 2.98 4.145
2007 21.54 -12.31 4.615
1988 -5.46 14.88 4.71
2012 1.56 8.03 4.795
1966 10.46 -0.86 4.8
1931 23.96 -14.29 4.835
2004 -0.33 10.16 4.915
1996 6.74 3.21 4.975
1983 -10.14 20.67 5.265
1980 37.1 -25.14 5.98
1960 17.19 -4.67 6.26
1978 12.07 0.52 6.295
1952 9.01 4.05 6.53
1998 23.39 -10.19 6.6
1930 25.75 -12.28 6.735
1967 22.48 -8.92 6.78
2008 13.3 0.83 7.065
1964 4.66 10.09 7.375
1948 11.86 3.52 7.69
1985 14.81 1.37 8.09
1981 -8.32 25.21 8.445
1972 15.57 1.79 8.68
1986 8.38 9.26 8.82
1961 10.75 7.11 8.93
1956 19.04 -1.07 8.985
1974 8.32 10.02 9.17
1970 -3.25 22.46 9.605
1995 17.62 1.65 9.635
1927 23.23 -3.75 9.74
1955 14.05 5.71 9.88
1941 9.14 11.13 10.135
1959 19.23 1.51 10.37
1968 2.25 18.59 10.42
1962 11.97 9.29 10.63
1928 28.21 -6.15 11.03
2001 4.31 18.33 11.32
1947 13.84 9.76 11.8
1989 28.1 -4.27 11.915
2005 14.95 9.18 12.065
1997 11.96 13.15 12.555
1977 17.86 7.47 12.665
1979 27 -1.46 12.77
1944 10.92 15.61 13.265
1945 15.55 11.4 13.475
1963 11.66 15.48 13.57
1992 3.15 24.22 13.685
1965 20.56 7.54 14.05
1984 9.15 19.35 14.25
1976 7.39 24.74 16.065
1929 20.73 11.83 16.28
1954 10.34 25.74 18.04
2002 25.65 10.49 18.07
1935 22.47 14.73 18.6
1936 5.89 35.86 20.875
1950 15.39 27.01 21.2
1993 23.72 18.96 21.34
1973 29.26 18.19 23.725
1982 34.69 13.62 24.155
1933 20.04 28.27 24.155
1943 14.24 38.97 26.605
2000 14.98 39.19 27.085


I'm not an expert on how QSPIX is implemented (as I can't invest in it anyway, being in the UK)

But I'd assume the momentum data there is measuring positive momentum? (The momentum premium most long-only investors feel) .. Obviously with everything going down, there wasn't much of that

But hedge funds can go long and short momentum - and I gather QSPIX does both - so things falling consistently can be just as profitable as things rising ... I've often mentioned the Medallion fund being a mathematical exercise in risk management - which allows them to leverage up to 17x .. And if QSPIX is suitably thought out, you can make substantial losses extremely improbable
"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

larryswedroe
Posts: 15598
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: QSPIX - thoughts on interesting fund

Post by larryswedroe » Thu Sep 24, 2015 4:05 pm

grok
Re the leverage, if the leverage was say 7:1 on the equities that might raise flags. But the higher leverage is likely due to leverage on things like ST interest rates, where it is MUCH LESS risky. And note that as vol rises the fund lowers it's leverage--and vice versa.
Larry

lack_ey
Posts: 5715
Joined: Wed Nov 19, 2014 11:55 pm

Re: QSPIX - thoughts on interesting fund

Post by lack_ey » Thu Sep 24, 2015 4:09 pm

Maynard F. Speer wrote:[quoted dataset]
I'm not an expert on how QSPIX is implemented (as I can't invest in it anyway, being in the UK)

But I'd assume the momentum data there is measuring positive momentum? (The momentum premium most long-only investors feel) .. Obviously with everything going down, there wasn't much of that

But hedge funds can go long and short momentum - and I gather QSPIX does both - so things falling consistently can be just as profitable as things rising ... I've often mentioned the Medallion fund being a mathematical exercise in risk management - which allows them to leverage up to 17x .. And if QSPIX is suitably thought out, you can make substantial losses extremely improbable

Nah, the French dataset there is UMD (up minus down), the pure factor exposure, not just long. Though only US stocks, if it's the series I think was used. AQR's styla premia fund is long-short (back to 0 net exposure) momentum across stocks and equity indexes as well as other asset classes.

The blowup in momentum was on the short side, as I alluded to earlier. Pure UMD had you shorting all those financial companies that had been doing terribly and riding that all the way to ruin as they rebounded up. Long-only momentum isn't as risky.

This issue is covered in an article by Larry Swedroe and is noted throughout the literature:
http://www.etf.com/sections/index-inves ... utperforms

This is part of the motivation for volatility targeting.

grok87
Posts: 7575
Joined: Tue Feb 27, 2007 9:00 pm

Re: QSPIX - thoughts on interesting fund

Post by grok87 » Thu Sep 24, 2015 4:21 pm

larryswedroe wrote:grok
Re the leverage, if the leverage was say 7:1 on the equities that might raise flags. But the higher leverage is likely due to leverage on things like ST interest rates, where it is MUCH LESS risky. And note that as vol rises the fund lowers it's leverage--and vice versa.
Larry

Thanks Larry. I take your point.

I think it is interesting though that, up until 2009, the average of the value and momentum premiums for us stocks never did worse than down 10%. Even through the great depression, etc. But then in 2009 there was a -44%, driven of course by momentum at -83%.

So my own view is that risk can show up unexpectedly and that things that we all thought were low risk can suddenly become quite risky. I do admit that i am probably a bit of a chicken in this regard. It's probably for people like me that AQR rolled out QSLIX, their less leveraged (And less expensive) version of the fund...
"...people always live for ever when there is any annuity to be paid them"- Jane Austen

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 » Thu Sep 24, 2015 4:33 pm

lack_ey wrote:
Maynard F. Speer wrote:[quoted dataset]
I'm not an expert on how QSPIX is implemented (as I can't invest in it anyway, being in the UK)

But I'd assume the momentum data there is measuring positive momentum? (The momentum premium most long-only investors feel) .. Obviously with everything going down, there wasn't much of that

But hedge funds can go long and short momentum - and I gather QSPIX does both - so things falling consistently can be just as profitable as things rising ... I've often mentioned the Medallion fund being a mathematical exercise in risk management - which allows them to leverage up to 17x .. And if QSPIX is suitably thought out, you can make substantial losses extremely improbable

Nah, the French dataset there is UMD (up minus down), the pure factor exposure, not just long. Though only US stocks, if it's the series I think was used. AQR's styla premia fund is long-short (back to 0 net exposure) momentum across stocks and equity indexes as well as other asset classes.

The blowup in momentum was on the short side, as I alluded to earlier. Pure UMD had you shorting all those financial companies that had been doing terribly and riding that all the way to ruin as they rebounded up. Long-only momentum isn't as risky.

This issue is covered in an article by Larry Swedroe and is noted throughout the literature:
http://www.etf.com/sections/index-inves ... utperforms

This is part of the motivation for volatility targeting.


Ah, got you .. I wasn't sure how that was measured - or actually implemented .. Assuming however they measure momentum, they switch positions fairly quickly as indicators cross? Rather than riding losses all the way down ... You saw the long-short equity funds do that here over 2008/09 - blowing up briefly before switching
"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

User avatar
nisiprius
Advisory Board
Posts: 34162
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: QSPIX - thoughts on interesting fund

Post by nisiprius » Thu Sep 24, 2015 4:35 pm

(With regard to my peeking at a completely different AQR fund, not QSPIX)
larryswedroe wrote:Nisi
I think that is very fair analysis, based on my very quick look.
I appreciate the comment, thanks.

Apologies to those who think I derailed the thread by bringing in another fund. I should have started a separate thread.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.

larryswedroe
Posts: 15598
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: QSPIX - thoughts on interesting fund

Post by larryswedroe » Thu Sep 24, 2015 4:44 pm

grok
Further to why I think your concerns are overstated (though not misguided).
Remember AQR will scale MOM down as vol goes up and the vol crashes are at turn arounds which occur at very high vol So the leverage applied to MOM at least (where the big losses would have been) would have been scaled way down.

In fact there is good academic literature on scaling MOM strategies which has shown the big losses are cut dramatically without impacting returns (if memory serves/traveling today and tomorrow). This isn't an issue for long only MOM as that side doesn't have "crashes."

Larry

lack_ey
Posts: 5715
Joined: Wed Nov 19, 2014 11:55 pm

Re: QSPIX - thoughts on interesting fund

Post by lack_ey » Thu Sep 24, 2015 4:59 pm

Maynard F. Speer wrote:
lack_ey wrote:
Maynard F. Speer wrote:[quoted dataset]
I'm not an expert on how QSPIX is implemented (as I can't invest in it anyway, being in the UK)

But I'd assume the momentum data there is measuring positive momentum? (The momentum premium most long-only investors feel) .. Obviously with everything going down, there wasn't much of that

But hedge funds can go long and short momentum - and I gather QSPIX does both - so things falling consistently can be just as profitable as things rising ... I've often mentioned the Medallion fund being a mathematical exercise in risk management - which allows them to leverage up to 17x .. And if QSPIX is suitably thought out, you can make substantial losses extremely improbable

Nah, the French dataset there is UMD (up minus down), the pure factor exposure, not just long. Though only US stocks, if it's the series I think was used. AQR's styla premia fund is long-short (back to 0 net exposure) momentum across stocks and equity indexes as well as other asset classes.

The blowup in momentum was on the short side, as I alluded to earlier. Pure UMD had you shorting all those financial companies that had been doing terribly and riding that all the way to ruin as they rebounded up. Long-only momentum isn't as risky.

This issue is covered in an article by Larry Swedroe and is noted throughout the literature:
http://www.etf.com/sections/index-inves ... utperforms

This is part of the motivation for volatility targeting.


Ah, got you .. I wasn't sure how that was measured - or actually implemented .. Assuming however they measure momentum, they switch positions fairly quickly as indicators cross? Rather than riding losses all the way down ... You saw the long-short equity funds do that here over 2008/09 - blowing up briefly before switching

Hm... Good question. I would need to check how frequently the portfolio is reconstituted (monthly, I believe). The problem there is that momentum is only being measured based on trailing twelve months I think (or was it trailing twelve excepting the last? something like that). So if something has massive recent losses in the last twelve months but a considerable rebound in the last few, it may still count as negative momentum and would be shorted.

I believe most practitioners use multiple momentum measures, not just trailing twelve months but maybe also last three, last one. That would be quicker to catch on to a changing position. Many use some human input as well, not just algorithmic.

In any case, the raw data being quoted is not really indicative of the particular implementation that AQR is going for, even in the narrow slice of the momentum/equities combination.


nisiprius wrote:(With regard to my peeking at a completely different AQR fund, not QSPIX)
larryswedroe wrote:Nisi
I think that is very fair analysis, based on my very quick look.
I appreciate the comment, thanks.

Apologies to those who think I derailed the thread by bringing in another fund. I should have started a separate thread.

I only speak for myself but I think it's relevant. It's an AQR factor fund, just a long-only implementation. There are some takeaways that apply to some extent.

But for that fund, the majority of the exposure is just the market beta, which is unhedged in a long-only fund. All the other factor exposures can get largely swamped out. With a hedged, leveraged long-short fund, the non-market exposures can be amplified, for better or worse. You are getting a lot more non-market-beta factor exposures per dollar invested in the style premia fund, so even though that has a higher ER there is a greater chance for the alternative exposures to provide returns that beat the ER. If all goes well and the factors actually provide positive returns, that is.


larryswedroe wrote:grok
Further to why I think your concerns are overstated (though not misguided).
Remember AQR will scale MOM down as vol goes up and the vol crashes are at turn arounds which occur at very high vol So the leverage applied to MOM at least (where the big losses would have been) would have been scaled way down.

In fact there is good academic literature on scaling MOM strategies which has shown the big losses are cut dramatically without impacting returns (if memory serves/traveling today and tomorrow). This isn't an issue for long only MOM as that side doesn't have "crashes."

Larry

I remember you wrote about that for etf.com a while back. I linked it above:
http://www.etf.com/sections/index-inves ... utperforms

grok87
Posts: 7575
Joined: Tue Feb 27, 2007 9:00 pm

Re: QSPIX - thoughts on interesting fund

Post by grok87 » Thu Sep 24, 2015 5:09 pm

larryswedroe wrote:grok
Further to why I think your concerns are overstated (though not misguided).
Remember AQR will scale MOM down as vol goes up and the vol crashes are at turn arounds which occur at very high vol So the leverage applied to MOM at least (where the big losses would have been) would have been scaled way down.

In fact there is good academic literature on scaling MOM strategies which has shown the big losses are cut dramatically without impacting returns (if memory serves/traveling today and tomorrow). This isn't an issue for long only MOM as that side doesn't have "crashes."

Larry

Thanks Larry, safe travels.

I was going to ask for a pointer to one of those papers ("good academic literature") but i see lack_ey has already linked to your article discussing them.
"...people always live for ever when there is any annuity to be paid them"- Jane Austen

grok87
Posts: 7575
Joined: Tue Feb 27, 2007 9:00 pm

Re: QSPIX - thoughts on interesting fund

Post by grok87 » Thu Sep 24, 2015 5:10 pm

lack_ey wrote:I remember you wrote about that for etf.com a while back. I linked it above:
http://www.etf.com/sections/index-inves ... utperforms

Thanks for the link lack_ey!
"...people always live for ever when there is any annuity to be paid them"- Jane Austen

lack_ey
Posts: 5715
Joined: Wed Nov 19, 2014 11:55 pm

Re: QSPIX - thoughts on interesting fund

Post by lack_ey » Thu Sep 24, 2015 5:33 pm

You're welcome. :happy

By the way, I noticed in another thread, there was a discussion on currencies potentially having negative correlation with stocks, though that can be unreliable and obviously you need to be on the right side of the trade. Eventually the discussion evolved to get to this post, which mentioned that "long-short currency based on PPP, yield and momentum can be of some use and can produce diversified return."
viewtopic.php?f=10&t=174263&newpost=2634109&start=50#p2633560

IIRC AQR defines value in currencies by what has high or low purchasing power. Carry is what has high yield (short what has low yield). Momentum is obvious. They skip defensive for currencies. So the currency part of QSPIX you can call styles or factors, but these are the old tricks found and known elsewhere. No guarantees about the future, but it's not really something "new" AQR came up with. Same with most of the other style/asset class combinations.

vencat
Posts: 200
Joined: Thu Sep 10, 2009 6:30 pm

Re: QSPIX - thoughts on interesting fund

Post by vencat » Thu Sep 24, 2015 5:55 pm

For most of us this is an inaccessible and arguably, unnecessary fund.
High minimums, tax inefficiency, high expense ratio and very short history settle it for me.
Maybe this is the holy grail of investing fund that was not available when Larry wrote his book: The Quest for Alpha: The Holy Grail of Investing .
I personally would watch and see how this performs over the next 5 years or so. If it truly delivers, there will be good, cheap copy cats. Hopefully the factors are persistent and pervasive then!
In the meantime, very satisfied with my investing with Vanguard for past 19 years with conventional global stocks and bonds albeit with small and value tilt. Have not needed high yield bonds or for that matter the insurance protection of commodities.
Another QSPIX enthusiast:

http://www.mutualfundobserver.com/2015/ ... mber-2015/

Venkat

User avatar
matjen
Posts: 1694
Joined: Sun Nov 20, 2011 11:30 pm

Re: QSPIX - thoughts on interesting fund

Post by matjen » Thu Sep 24, 2015 7:00 pm

vencat wrote:Another QSPIX enthusiast:

http://www.mutualfundobserver.com/2015/ ... mber-2015/

Venkat


Although Larry posted this earlier, I read it more closely this time and realized that it wasn't written by the usual guy at Mutual Fund Observer. This is written (and pretty heartily endorsed) by Samuel Lee. I'm pretty sure this must be Morningstar's Samuel Lee. Apparently he left Morningstar a little while ago. I know the "smart people like this fund" argument can annoy people...but he is pretty darn smart and I think has a history of not just liking everything that is shiny. :-)

How much capital should one dedicate to it?

Depends on how much you trust the strategy, the managers, and so on. I personally would invest up to 30% of my personal money in the fund (and may do so soon!), but that’s only because I have a high taste for unconventionality, decades of earnings ahead of me, high conviction in the strategy and people, and a pessimistic view of competing options (other alternatives as well as conventional stocks and bonds). Swedroe, on the other hand, says he has 3% of his portfolio in it.


http://www.morningstar.com/articles/aut ... l-lee.aspx
A man is rich in proportion to the number of things he can afford to let alone.

larryswedroe
Posts: 15598
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: QSPIX - thoughts on interesting fund

Post by larryswedroe » Thu Sep 24, 2015 8:40 pm

vencat
The important point is that QSPIX isn't seeking alpha, only beta of various factors across various asset classes.
There is no quest for alpha here.
Larry

grok87
Posts: 7575
Joined: Tue Feb 27, 2007 9:00 pm

Re: QSPIX - thoughts on interesting fund

Post by grok87 » Fri Sep 25, 2015 2:09 am

matjen wrote:
vencat wrote:Another QSPIX enthusiast:

http://www.mutualfundobserver.com/2015/ ... mber-2015/

Venkat


Although Larry posted this earlier, I read it more closely this time and realized that it wasn't written by the usual guy at Mutual Fund Observer. This is written (and pretty heartily endorsed) by Samuel Lee. I'm pretty sure this must be Morningstar's Samuel Lee. Apparently he left Morningstar a little while ago. I know the "smart people like this fund" argument can annoy people...but he is pretty darn smart and I think has a history of not just liking everything that is shiny. :-)

How much capital should one dedicate to it?

Depends on how much you trust the strategy, the managers, and so on. I personally would invest up to 30% of my personal money in the fund (and may do so soon!), but that’s only because I have a high taste for unconventionality, decades of earnings ahead of me, high conviction in the strategy and people, and a pessimistic view of competing options (other alternatives as well as conventional stocks and bonds). Swedroe, on the other hand, says he has 3% of his portfolio in it.


http://www.morningstar.com/articles/aut ... l-lee.aspx

30% is way too much imho. Again i realize many may consider me alarmist, but i don't think it is all that hard for this fund, which is 7x leveraged, to go to zero. I've got a small amount of money in the vanguard market neutral fund which is 2x leveraged.
"...people always live for ever when there is any annuity to be paid them"- Jane Austen

User avatar
backpacker
Posts: 1620
Joined: Mon Sep 22, 2014 2:17 pm

Re: QSPIX - thoughts on interesting fund

Post by backpacker » Fri Sep 25, 2015 6:19 am

larryswedroe wrote:grok
Further to why I think your concerns are overstated (though not misguided).
Remember AQR will scale MOM down as vol goes up and the vol crashes are at turn arounds which occur at very high vol So the leverage applied to MOM at least (where the big losses would have been) would have been scaled way down.

In fact there is good academic literature on scaling MOM strategies which has shown the big losses are cut dramatically without impacting returns (if memory serves/traveling today and tomorrow). This isn't an issue for long only MOM as that side doesn't have "crashes."


So, we need to not only believe in the relevant factors, we need AQR to predict future factor factor crashes so they can deleverage in time and keep the fund from blowing up?

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 » Fri Sep 25, 2015 7:46 am

grok87 wrote:
matjen wrote:
vencat wrote:Another QSPIX enthusiast:

http://www.mutualfundobserver.com/2015/ ... mber-2015/

Venkat


Although Larry posted this earlier, I read it more closely this time and realized that it wasn't written by the usual guy at Mutual Fund Observer. This is written (and pretty heartily endorsed) by Samuel Lee. I'm pretty sure this must be Morningstar's Samuel Lee. Apparently he left Morningstar a little while ago. I know the "smart people like this fund" argument can annoy people...but he is pretty darn smart and I think has a history of not just liking everything that is shiny. :-)

How much capital should one dedicate to it?

Depends on how much you trust the strategy, the managers, and so on. I personally would invest up to 30% of my personal money in the fund (and may do so soon!), but that’s only because I have a high taste for unconventionality, decades of earnings ahead of me, high conviction in the strategy and people, and a pessimistic view of competing options (other alternatives as well as conventional stocks and bonds). Swedroe, on the other hand, says he has 3% of his portfolio in it.


http://www.morningstar.com/articles/aut ... l-lee.aspx

30% is way too much imho. Again i realize many may consider me alarmist, but i don't think it is all that hard for this fund, which is 7x leveraged, to go to zero. I've got a small amount of money in the vanguard market neutral fund which is 2x leveraged.


It's an interesting proposition, whether you can create something that gives you the return of equities with the volatility of bonds/cash .. without risk creeping in somewhere else ... Even the best hedge funds have been vulnerable to market anomalies - where everything goes against the prediction (an acute, if improbable, risk ... as opposed to the more chronic day-to-day risk of equities)

My attitude towards alternatives has become: when you see something that works, jump on it ... If QSPIX works too well, some of that return's going to be arbitraged away; risk probably isn't ... I'd wonder would a trailing stop-loss be the way to invest? If it keeps working as advertised, you get all the benefits; if it doesn't, you risk jumping ship early, but it's allowed to invest in it in the first place .. That's how I treat funds like Third Point Offshore, and anything long/short equity
"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

larryswedroe
Posts: 15598
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: QSPIX - thoughts on interesting fund

Post by larryswedroe » Fri Sep 25, 2015 7:59 am

backpacker
First, it's not a question of "believing in the factors"--it's believing in the incredible amount of research that provides both the evidence that the factors exist, have been persistent and pervasive, and have very logical explanations for their existence and why they should persist. It's not like a religious belief which is based on "faith" but instead is based on facts.

Second, there is no need to believe AQR can predict when crashes happen. They don't do any predicting at all. They follow algorithmic rules based on the evidence that volatility predicts volatility. So as volatility rises they lower leverage and vice versa.

Vencat
I would add again that 7x leverage might/would be concern if it was solely say on beta, or one factor like value (which has volatility lower than beta, but not that much lower). The difference is the overall leverage includes levering short term interest rates which have very tiny volatility and currencies which have much less volatility than stocks. So you cannot look at the overall volatility without considering the leverage as applied to each part. Also you should not ignore that they are leveraging say both MOM and value, which are negatively correlated--so even if one crashes, then the other will tend to do very well at the same time. Same thing true for defensive, which tends to do well when markets crashing

IMO there is very little risk here of a very large loss, and your concern about it going to zero is way, way, way overstated. Once volatility rises the leverage drops preventing the very thing you are concerned about, and the factors don't all go down at same time--there's no logical reason to think that say the momentum factor exposures will all crash at the same time (meaning within MOM you have exposure to stocks, bonds, currencies and commodities).

Hope that helpful
Larry

grok87
Posts: 7575
Joined: Tue Feb 27, 2007 9:00 pm

Re: QSPIX - thoughts on interesting fund

Post by grok87 » Fri Sep 25, 2015 8:18 am

lack_ey wrote: (Possibly too late in some cases, though. In fact, this is one of the risks of the fund, that some other fund that's much more aggressively leveraged is taking similar positions but is forced to unwind, causing a chain reaction of similar funds unwinding as a result of the volatility targeting programs.)

Interesting point
"...people always live for ever when there is any annuity to be paid them"- Jane Austen

grok87
Posts: 7575
Joined: Tue Feb 27, 2007 9:00 pm

Re: QSPIX - thoughts on interesting fund

Post by grok87 » Fri Sep 25, 2015 8:24 am

backpacker wrote:
larryswedroe wrote:grok
Further to why I think your concerns are overstated (though not misguided).
Remember AQR will scale MOM down as vol goes up and the vol crashes are at turn arounds which occur at very high vol So the leverage applied to MOM at least (where the big losses would have been) would have been scaled way down.

In fact there is good academic literature on scaling MOM strategies which has shown the big losses are cut dramatically without impacting returns (if memory serves/traveling today and tomorrow). This isn't an issue for long only MOM as that side doesn't have "crashes."


So, we need to not only believe in the relevant factors, we need AQR to predict future factor factor crashes so they can deleverage in time and keep the fund from blowing up?

It reminds me a little of portfolio insurance and the 1987 crash
http://www.nytimes.com/2012/10/19/busin ... .html?_r=0
"...people always live for ever when there is any annuity to be paid them"- Jane Austen

User avatar
grap0013
Posts: 1867
Joined: Thu Mar 18, 2010 1:24 pm

Re: QSPIX - thoughts on interesting fund

Post by grap0013 » Fri Sep 25, 2015 8:40 am

nisiprius wrote:
Apologies to those who think I derailed the thread by bringing in another fund. I should have started a separate thread.


My apologies if I came off too harsh. If so, that was not my intention. As I've stated before, you and I have completely different investing styles but your contributions are invaluable to this forum. I'm happy this thread has turned the corner in a more respectful manner. :sharebeer
There are no guarantees, only probabilities.

User avatar
oneleaf
Posts: 2265
Joined: Mon Feb 19, 2007 5:48 pm

Re: QSPIX - thoughts on interesting fund

Post by oneleaf » Fri Sep 25, 2015 8:43 am

grok87 wrote:I've got a small amount of money in the vanguard market neutral fund which is 2x leveraged.


Were you able to find a way to get into the market neutral fund for less than the $250k minimum?

User avatar
backpacker
Posts: 1620
Joined: Mon Sep 22, 2014 2:17 pm

Re: QSPIX - thoughts on interesting fund

Post by backpacker » Fri Sep 25, 2015 10:26 am

larryswedroe wrote:There is no need to believe AQR can predict when crashes happen. They don't do any predicting at all. They follow algorithmic rules based on the evidence that volatility predicts volatility. So as volatility rises they lower leverage and vice versa.


"We don't predict future volatility. We have an algorithm that predicts future volatility and we do what it tells us." :oops:

Look. Certain patterns of past volatility have predicted future volatility in the past. Maybe those patterns will persist and maybe they won't. The point is just that this is yet another bit of complexity layered on top of the "simple" factor bets QSPIX is taking. Even if you're generally optimistic about factor investing (as I in fact am!), you might not be so optimistic about predicting future volatility.
Last edited by backpacker on Fri Sep 25, 2015 10:33 am, edited 1 time in total.

User avatar
backpacker
Posts: 1620
Joined: Mon Sep 22, 2014 2:17 pm

Re: QSPIX - thoughts on interesting fund

Post by backpacker » Fri Sep 25, 2015 10:30 am

grok87 wrote:
backpacker wrote:
larryswedroe wrote:grok
Further to why I think your concerns are overstated (though not misguided).
Remember AQR will scale MOM down as vol goes up and the vol crashes are at turn arounds which occur at very high vol So the leverage applied to MOM at least (where the big losses would have been) would have been scaled way down.

In fact there is good academic literature on scaling MOM strategies which has shown the big losses are cut dramatically without impacting returns (if memory serves/traveling today and tomorrow). This isn't an issue for long only MOM as that side doesn't have "crashes."


So, we need to not only believe in the relevant factors, we need AQR to predict future factor factor crashes so they can deleverage in time and keep the fund from blowing up?

It reminds me a little of portfolio insurance and the 1987 crash
http://www.nytimes.com/2012/10/19/busin ... .html?_r=0


That's a nice point. Yet one more reason to be skeptical of the "just trust the algorithms" line of thinking.

New York Times wrote:It was just a quarter-century ago that Wall Street was shaken to its core by the Oct. 19, 1987, stock market crash.

On one day, the Dow Jones industrial average lost 23 percent of its value. People wondered if that heralded a new Depression. A front page headline in The New York Times asked, “Does 1987 Equal 1929?”

It did not. The next recession, a mild one, was more than two years away.

What it did signify was the beginning of the destruction of markets by dumb computers. Or, to be fair to the computers, by computers programmed by fallible people and trusted by people who did not understand the computer programs’ limitations. As computers came in, human judgment went out.

lack_ey
Posts: 5715
Joined: Wed Nov 19, 2014 11:55 pm

Re: QSPIX - thoughts on interesting fund

Post by lack_ey » Fri Sep 25, 2015 10:53 am

These are two different points that are being made.

AQR is not predicting volatility in the sense of having to guess or forecast the future and do the tea leaf readings. It is an appropriate response to say that they're just following algorithmic rules about volatility targeting and that such measures may be able to help in many cases (though it seems nobody is disputing that).

On the other hand, we know that this won't help every time, and weird market anomalies do happen, we get flash crashes and the like, and other conditions, so it is right to point that out (though nobody's disputing that).

Wading past all that, you need to understand the limitations and what's going on.

backpacker wrote:[...]Certain patterns of past volatility have predicted future volatility in the past. Maybe those patterns will persist and maybe they won't. The point is just that this is yet another bit of complexity layered on top of the "simple" factor bets QSPIX is taking. Even if you're generally optimistic about factor investing (as I in fact am!), you might not be so optimistic about predicting future volatility.

But this is to me a strange reading of the situation. The extra complexity, as far as I can tell, is likely to help overall but not work every time. You're presenting it as if it's a bad thing. Even if you have very low confidence in it working, it should be closer to neutral, unless you really expect it to make the fund take on extra risk at precisely the wrong time, or that complexity in of itself is harmful (or potentially the chain reaction scenario, though again remember that the fund does trading relatively patiently and isn't as leveraged in risky assets as many others and probably could sidestep trouble).

Also keep in mind the baseline risk level. Let's say volatility hits out of the blue before the fund can react based on its algorithms. So forget the algorithms. The fund isn't really long in risky positions but hedged. Do you especially think that all of the net trades are going to be unfavorable, or enough to force it out? It could happen, but what's the chance? Probably higher than many simplified models would suggest, but how high?

vencat
Posts: 200
Joined: Thu Sep 10, 2009 6:30 pm

Re: QSPIX - thoughts on interesting fund

Post by vencat » Fri Sep 25, 2015 12:01 pm

Larry,

Sorry i got the alpha and beta mixed up.

More fuel for the enthusiasts (Robert T may have referenced this earlier):

https://grandstreetwm.wordpress.com/201 ... ger-part-1
https://grandstreetwm.wordpress.com/201 ... er-part-2/

Venkat

User avatar
backpacker
Posts: 1620
Joined: Mon Sep 22, 2014 2:17 pm

Re: QSPIX - thoughts on interesting fund

Post by backpacker » Fri Sep 25, 2015 5:16 pm

lack_ey wrote:
backpacker wrote:[...]Certain patterns of past volatility have predicted future volatility in the past. Maybe those patterns will persist and maybe they won't. The point is just that this is yet another bit of complexity layered on top of the "simple" factor bets QSPIX is taking. Even if you're generally optimistic about factor investing (as I in fact am!), you might not be so optimistic about predicting future volatility.

But this is to me a strange reading of the situation. The extra complexity, as far as I can tell, is likely to help overall but not work every time. You're presenting it as if it's a bad thing. Even if you have very low confidence in it working, it should be closer to neutral, unless you really expect it to make the fund take on extra risk at precisely the wrong time, or that complexity in of itself is harmful (or potentially the chain reaction scenario, though again remember that the fund does trading relatively patiently and isn't as leveraged in risky assets as many others and probably could sidestep trouble).

Also keep in mind the baseline risk level. Let's say volatility hits out of the blue before the fund can react based on its algorithms. So forget the algorithms. The fund isn't really long in risky positions but hedged. Do you especially think that all of the net trades are going to be unfavorable, or enough to force it out? It could happen, but what's the chance? Probably higher than many simplified models would suggest, but how high?


The question is how safe the fund's 7x leverage is. Larry's response to grok is that it's safe because the fund deleverages when the algorithm says that expected volatility is high. It could be that on the whole, 7x leverage + volatility management is better than 7x leverage. I have no idea.

What would be a bad situation, I think, is if the leverage the fund is using is safe only assuming that the volatility management works as expected. This because volatility management is the exactly the sort of thing (like portfolio insurance) that often doesn't work as expected. So the leverage better be safe without assuming that it does. At least that's my view.

Now, how safe 7x leverage is depends on how much leverage the fund is applying to what and when. I'm not sure that we know much about that.

Keep in mind too that the various factors tend to be correlated across asset classes. This gives us greater reason to think that momentum and value are real, but also means that using those strategies across different asset classes provides less protection than one might think.

lack_ey
Posts: 5715
Joined: Wed Nov 19, 2014 11:55 pm

Re: QSPIX - thoughts on interesting fund

Post by lack_ey » Fri Sep 25, 2015 5:37 pm

Here's Figure 3 corresponding to Figure 1 (style returns) I posted earlier.
Image

Maybe it's 5x sometimes, 9x other times. Look at Figure 1 again and imagine, for the times when there are losses, that the leverage is 9x rather than whatever it actually was.

The range isn't so great that 7x vs. whatever it is at the moment is likely to be the difference between living and dying. The volatility targeting is a feature but not a crutch, mostly. Certainly relying on something like that would be playing with fire. It's not really doing that.

Now, the only natural expectation is for live-money trading to encounter something worse than seen in just 23 years of simulated data (for reference, maximum simulated composite drawdown in the period was just 18.5%), but enough to really sink the fund? Maybe, but I don't think it's all that likely.

User avatar
oneleaf
Posts: 2265
Joined: Mon Feb 19, 2007 5:48 pm

Re: QSPIX - thoughts on interesting fund

Post by oneleaf » Fri Sep 25, 2015 5:45 pm

backpacker wrote:The question is how safe the fund's 7x leverage is. Larry's response to grok is that it's safe because the fund deleverages when the algorithm says that expected volatility is high. It could be that on the whole, 7x leverage + volatility management is better than 7x leverage. I have no idea.

What would be a bad situation, I think, is if the leverage the fund is using is safe only assuming that the volatility management works as expected. This because volatility management is the exactly the sort of thing (like portfolio insurance) that often doesn't work as expected. So the leverage better be safe without assuming that it does. At least that's my view.

Now, how safe 7x leverage is depends on how much leverage the fund is applying to what and when. I'm not sure that we know much about that.

Keep in mind too that the various factors tend to be correlated across asset classes. This gives us greater reason to think that momentum and value are real, but also means that using those strategies across different asset classes provides less protection than one might think.


It is troubling to me that we would be depending on an algorithm to turn off leverage when expected volatility is high. I think we are getting to a point where quants/algorithms/rules-based methods are given too much of a pass. In this case, this is no different than market timing. Imagine having a leveraged stock portfolio that turns off the leverage when certain conditions are met (200 DMA or CAPE-based or whatever).

I get Larry's definition of passive. But there is a point at which the rules/parameters of the strategy become so complex that the combination of rules in and of itself becomes a matter of human judgement. I believe most human judgements can be reduced to some number of rules/parameters, when it comes to investing. A decision to pick one stock over another, or overweight one sector over another is not usually just a gut-feeling, but due to a very tangible reason that could become a "rule" and be argued to be passive. I think QSPIX's complexity in its use of leverage crosses that line from "we can still argue this is passive" to "c'mon, how can this be passive"?

larryswedroe
Posts: 15598
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: QSPIX - thoughts on interesting fund

Post by larryswedroe » Fri Sep 25, 2015 6:24 pm

backpacker
People are missing the point entirely.
The leverage say at 7:1 is misleading if you think of it as leverage of equity risks
You could be say 20:1 on three month bills and it would not be great risk because they don't move that much. Yet 5:1 on equities would be huge.
Since much of the leverage is on securities with low volatility and thus higher leverage the AVERAGE is misleading and makes no sense to use.
The issue is what is the fund's overall volatility. And given vol of 10% and say expected return of 7% it takes even a 3sd event to get a loss of say 20%, let alone blow up the fund causing it to go to zero. People simply don't understand IMO what they are talking about but make claims as if they do know.

And the fact is that volatility has predicted volatility quite well and this strategy has NOTHING to do with what happened with portfolio insurance in 1987 and just making that statement shows that one doesn't understand what is happening with this fund.

Also the fund deleverages when volatility is high as that predicts future high volatility. So as risk rises the fund gets less risky.

Larry

User avatar
backpacker
Posts: 1620
Joined: Mon Sep 22, 2014 2:17 pm

Re: QSPIX - thoughts on interesting fund

Post by backpacker » Fri Sep 25, 2015 7:15 pm

larryswedroe wrote:backpacker
People are missing the point entirely.
The leverage say at 7:1 is misleading if you think of it as leverage of equity risks.
You could be say 20:1 on three month bills and it would not be great risk because they don't move that much. Yet 5:1 on equities would be huge.
Since much of the leverage is on securities with low volatility and thus higher leverage the AVERAGE is misleading and makes no sense to use.
The issue is what is the fund's overall volatility. And given vol of 10% and say expected return of 7% it takes even a 3sd event to get a loss of say 20%, let alone blow up the fund causing it to go to zero. People simply don't understand IMO what they are talking about but make claims as if they do know.

And the fact is that volatility has predicted volatility quite well and this strategy has NOTHING to do with what happened with portfolio insurance in 1987 and just making that statement shows that one doesn't understand what is happening with this fund.

Also the fund deleverages when volatility is high as that predicts future high volatility. So as risk rises the fund gets less risky.


1) Everyone here understands that leveraging low-risk assets is different than leveraging high-risk assets. What we want to know is how much leverage AQR is using, and under what conditions, for which assets. Otherwise, how do we have any idea how to estimate potential losses?

2) AQR is targeting 10% volatility, yes. We all know that. How confident we should be that the fund's realized volatility will only be 10%? Who knows. That's the point. Complex strategies involving leverage have a nasty habit of working. Until they don't.

User avatar
backpacker
Posts: 1620
Joined: Mon Sep 22, 2014 2:17 pm

Re: QSPIX - thoughts on interesting fund

Post by backpacker » Fri Sep 25, 2015 7:22 pm

lack_ey wrote:Here's Figure 3 corresponding to Figure 1 (style returns) I posted earlier.
Image

Maybe it's 5x sometimes, 9x other times. Look at Figure 1 again and imagine, for the times when there are losses, that the leverage is 9x rather than whatever it actually was.


Well, suppose that the fund was 5x leverage during the great 2007-09 factor crash of -50%. If it's 9x during the next factor crash of -50%, the fund gets more or less wiped out. And no one said that factor crashes can't be bigger than -50%!

larryswedroe
Posts: 15598
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: QSPIX - thoughts on interesting fund

Post by larryswedroe » Fri Sep 25, 2015 7:34 pm

Backpacker, the fund doesn't work the way you think about it, and you cannot simply multiply leverage by a loss as it depends as I have tried to explain on where the leverage is. For example if it had say 20x the leverage on ST bonds and low leverage on equity factors during a crash then the losses might be small. Also remember in equity crashes momentum and defensive tend to do well. So you simply cannot multiply some leverage figure.

Note in this recent mini crash the fund is actually well up.

IMO a very good idea of the potential for losses is to look at the backtesting and the worst case was if memory serves was about 18%, which is about a 2.5 SD event. So want to look at a 3sd or even 4sd event you can then estimate losses.

Note again the factors tend to have no correlation and MOM and value are negatively correlated. That makes it hard to see situation with massive losses, or massive gains.

If the fund is "too complex" for you and you have too great concerns because you don't have enough confidence in the historical results, that is fine, don't invest. I do have enough confidence to use the fund as an effective diversifier and a relatively small allocation

Larry

lack_ey
Posts: 5715
Joined: Wed Nov 19, 2014 11:55 pm

Re: QSPIX - thoughts on interesting fund

Post by lack_ey » Fri Sep 25, 2015 7:53 pm

I'm not sure which parts of the discussion and notes you've read, backpacker, but the risk targets for different asset classes and styles are listed. The aggregate total of all of the positions is the 10% (or so) expected volatility, which as they say will not be the actual realized volatility over any given period of time.

The equity market neutral fund (QMNIX) was 2x / -1.8x at the end of June, before all the recent volatility (VIX was about 12-16 in the middle months of the year). The fund says it targets 6% standard deviation over the long run—though prospectus says range of 4-9%—for reference.

My rough guess is that QSPIX might be in a similar range of total equity exposure through stocks and equity indices when stocks are placid, if it's usually has about 55% of the total risk in that category and the whole thing has the 10% target. That's a bit higher than some earlier guesses I had.

The bigger deal is that the fund is not investing with a single strategy across stocks or any asset class; it is unlikely for trades based on different styles to all go bad simultaneously.


For what it's worth, in finance in general I get scared when anybody mentions "events" in terms of standard deviations much larger than 1, certainly 2. Before you can say if something is 2.5, 1.8, or 4.9, you better have a clear idea of the distribution, which we don't have. It may be worth reviewing the skew and kurtosis seen in the simulated backtest, though I personally wouldn't have much confidence in those in describing the future. Real-life finance distributions tend not to look like any kind of model past a very low number of SD, even more so when leverage is involved. Even if there is some legitimacy or useful information conveyed, I don't want to even think of it too often in these terms.

User avatar
LadyGeek
Site Admin
Posts: 41105
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: QSPIX - thoughts on interesting fund

Post by LadyGeek » Fri Sep 25, 2015 8:03 pm

Could someone please clarify how leverage is used for QSPIX? I'm looking the definition in the wiki: Leverage, but discussions on use of volatility as a multiplier (in this thread) has me confused.

According to the wiki: Leverage = debt/equity

For example, the volatility (statistical spread of returns) of (most) bonds is much lower than high-risk equities. Therefore, bonds have less leverage than high-risk equities.

How does volatility as discussed here apply to the debt/equity ratio?

Update: See next post for clarification of volatility.
To some, the glass is half full. To others, the glass is half empty. To an engineer, it's twice the size it needs to be.

lack_ey
Posts: 5715
Joined: Wed Nov 19, 2014 11:55 pm

Re: QSPIX - thoughts on interesting fund

Post by lack_ey » Fri Sep 25, 2015 8:30 pm

LadyGeek wrote:Could someone please clarify how leverage is used for QSPIX? I'm looking the definition in the wiki: Leverage, but discussions on use of volatility as a multiplier (in this thread) has me confused.

According to the wiki: Leverage = debt/equity

For example, the volatility (statistical spread of returns) of short term bonds is much lower than high-risk equities. Therefore, bonds have less leverage than high-risk equities.

How does volatility as discussed here apply to the debt/equity ratio?

The fund is primarily leveraged through the use of financial derivatives like futures and swaps, not so much actually going out and borrowing money and having debt on the books, like for a margin trading account. That doesn't make it any safer, but makes describing it in terms of "debt" as maybe misleading. The point is that the amount of underlying securities "controlled" or effectively invested in, both on the long and short sides, is greater than the amount of money invested in the fund.

In your example, I think you mean to say that (most) bonds have lower volatility (not leverage) than equities?

The fund has a volatility targeting scheme by which the positions are scaled up and down a little bit as market conditions change, in order to have all the net positions result in overall expected volatility in the 10% annual standard deviation range. This requires the amount of leverage to be adjusted over time. When underlying markets are less volatile, this makes the fund up the leverage effectively by taking on more futures contracts (and so on) per dollar invested. When underlying markets are more volatile, the opposite happens, and leverage is lowered.

In other words, it is the fund's strategy that calls for the leverage to be changed. The purpose of the leverage is to increase the fund's risk level and style (factor) exposures. This is how it has achieved 21% cumulative returns in a little less than 2 years, despite not being net invested in stocks or other asset classes. It is also how it's undoubtedly going to lose some or a lot of money at times even when other asset classes are going up (or down). How much leverage depends on the particulars of the market at the moment.

On the other hand, something like a 2x S&P 500 fund like ProShares Ultra S&P 500 (SSO) resets and maintains the amount of leverage every day to be the same. That way you get roughly 2x the daily return of the S&P 500.

User avatar
LadyGeek
Site Admin
Posts: 41105
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: QSPIX - thoughts on interesting fund

Post by LadyGeek » Fri Sep 25, 2015 8:48 pm

^^^ Thanks, that explains exactly what I was missing. I updated my post.

lack_ey wrote:On the other hand, something like a 2x S&P 500 fund like ProShares Ultra S&P 500 (SSO) resets and maintains the amount of leverage every day to be the same. That way you get roughly 2x the daily return of the S&P 500.

To be clear, this is a very different type of investment and not intended to be used in the same context. Actually, just don't use them in long-term retirement portfolios. See: Inverse and leveraged ETFs If anyone has questions, start a new thread to discuss.
To some, the glass is half full. To others, the glass is half empty. To an engineer, it's twice the size it needs to be.

lack_ey
Posts: 5715
Joined: Wed Nov 19, 2014 11:55 pm

Re: QSPIX - thoughts on interesting fund

Post by lack_ey » Fri Sep 25, 2015 9:06 pm

Right, for any given allocation, the more leverage at any given time, the higher the volatility. To try to keep volatility more consistent (primarily), the leverage is deliberately adaptive for this fund.

In theory you could do volatility targeting for any portfolio. For (long-only) stocks and bonds, this would essentially look like some kind of mechanical market timing scheme for shifting AA. If you had a target of 12% annual standard deviation this might be something like 80% stocks/20% bonds at some times, but 60% stocks/40% bonds at others, for example. I don't know if it would be much different over the long run compared with a static allocation. Volatility targeting is generally more seen for long/short funds, where you're scared of spikes in volatility, which could really kill the short positions if the assets zoom back up, and also because such funds are frequently leveraged too.

Post Reply