HEDGEFUNDIE's excellent adventure Part II: The next journey

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
langlands
Posts: 573
Joined: Wed Apr 03, 2019 10:05 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by langlands »

Uncorrelated wrote: Sun Aug 30, 2020 1:18 pm
hilink73 wrote: Sun Aug 30, 2020 1:10 pm
Uncorrelated wrote: Sat Aug 29, 2020 4:55 am

If you want to have as little arbitrary assumptions as possible, throwing away target volatility and HFEA is probably the first thing you should do. There are certain assumptions that lead to target volatility, those assumptions are false (see above). There are certain assumptions that lead to a strategy similar to HFEA, but it is very unlikely those assumptions are right for you.
Maybe I'm misunderstanding something (and I admit I'm not understanding half of what you are talking about re utility function, etc.) but I though TV is exactly what's writing on the box: how much volatility you want to accept (read: you're able to stomach)? I never saw it from the max return perspective...

I'm in for a TV of 25% but after this recent downturn I guess I'm able to stomach much more volatility (I've also got experience with dabbling in crypto trading, so volatility doesn't bother me so much (in the long run)).
I won't change my strategy or target volatility yet, though... PV says 80% UPRO/20% TMF for the next month.
The problem is that TV is not mean-variance optimal. That means there are other strategies that have the same risk as TV, but a higher expected return.

There are two conditions that need to be met before TV becomes optimal: recent volatility is the best possible estimate of future volatility, return is proportional to volatility. Both assumptions are clearly false. If you want to take less risk, TV or other market timing approaches are not the answer, the answer is to use a lower equity allocation.
Hi Uncorrelated,

To be more explicit about your discussion of target variance vs. target volatility, we are essentially talking about the difference between scaling your allocation by 1/σ^2 vs. 1/σ right? Assuming your return assumptions are fixed, mean variance gives that scaling by 1/σ^2 is theoretically correct as you noted. Scaling by 1/σ can be thought of as scaling by 1/σ^2 and then multiplying by sigma, hence your statement that target volatility assumes that returns are proportional to volatility, which you say is empirically shown to be false.

Ok. But many people doing this adventure are using constant percent allocations. The way I see it, target volatility allocation is the geometric mean of constant percent allocation and the 1/σ^2 derived from Merton's portfolio. In other words, it can't be that bad since it's moving in the right direction. Your criticism is essentially that it doesn't adjust the allocation enough to volatility movements.
hilink73
Posts: 476
Joined: Tue Sep 20, 2016 3:29 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by hilink73 »

Uncorrelated wrote: Sun Aug 30, 2020 1:18 pm
hilink73 wrote: Sun Aug 30, 2020 1:10 pm
Uncorrelated wrote: Sat Aug 29, 2020 4:55 am

If you want to have as little arbitrary assumptions as possible, throwing away target volatility and HFEA is probably the first thing you should do. There are certain assumptions that lead to target volatility, those assumptions are false (see above). There are certain assumptions that lead to a strategy similar to HFEA, but it is very unlikely those assumptions are right for you.
Maybe I'm misunderstanding something (and I admit I'm not understanding half of what you are talking about re utility function, etc.) but I though TV is exactly what's writing on the box: how much volatility you want to accept (read: you're able to stomach)? I never saw it from the max return perspective...

I'm in for a TV of 25% but after this recent downturn I guess I'm able to stomach much more volatility (I've also got experience with dabbling in crypto trading, so volatility doesn't bother me so much (in the long run)).
I won't change my strategy or target volatility yet, though... PV says 80% UPRO/20% TMF for the next month.
The problem is that TV is not mean-variance optimal. That means there are other strategies that have the same risk as TV, but a higher expected return.

There are two conditions that need to be met before TV becomes optimal: recent volatility is the best possible estimate of future volatility, return is proportional to volatility. Both assumptions are clearly false. If you want to take less risk, TV or other market timing approaches are not the answer, the answer is to use a lower equity allocation.
So, a fixed allocation as in the original adventure is better? But, which % allocation, then?
Semantics
Posts: 147
Joined: Tue Mar 10, 2020 1:42 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Semantics »

Uncorrelated wrote: Sat Aug 29, 2020 4:55 am
Semantics wrote: Fri Aug 28, 2020 8:34 pm Thanks for the suggestions, these ETFs look like they have good theoretical merit, however their performance has been poor lately and they likely won't have good results until small cap value starts to perform again. I've previously thought about leveraging USMV, which seems similar to your suggestions but excludes small cap, which is presumably why it has been performing better, but continue to stick to LETFs for now for simplicity.

But the interesting thing is simple 60/40 S&P 500/BTAL has still handily outperformed all of these in PV in terms of Sharpe (it has also beaten 60/40 S&P 500/BND). Maybe that won't continue, I don't know enough about the theoretical underpinnings. Is it just luck? Do they execute better in their BAB implementation? I'm thinking it's perhaps that this strategy captures a momentum effect due to only rebalancing and harvesting the BAB factor+cash monthly/quarterly, vs. 100% long low-vol stock capturing it in real time. In other words the periodic rebalancing allows intervals of increased exposure to the side that's performing well and decreased exposure to the underperforming side.
Higher sharpe does not imply it's better choice.

Since neither BTAL nor popular value ETF's have a statistically significant positive alpha and the difference between BAB expected return and value expected return is small, the only valid explanation is that it was luck. If the rebalancing interval mattered, then that was also pure luck. The momentum exposure of BTAL is positive but only if you don't correct for value exposure, so it's unlikely the momentum exposure will explain a significant part of future expected return.

If you look at the performance attribution tab in this very quick factor regression with BTAL and UPRO, you'll see that the market factor and BAB were responsible for about half of the return each (26 and 32 bps monthly). if you compare the performance to historical factor results, you'll see that market returns during this time period were approx 2.5x as large as the historical average and BAB between 2x and 4x as large.

It's clear why backtesting doesn't work: If you run a backtest, you implicitly assume that the mean return and correlations in the backtest period (in this case, 2011-2020) is a better estimate of future return than the BAB paper I linked (1926-2012). Clearly an absurd assumption.
I don't think that's absurd at all. Almost the entire period covered by the paper you linked corresponds to obsolete US monetary policy regimes. Frankly, I place more value on following the Fed than I do on academic dogma when making investment decisions. This whole thread is about a strategy that wouldn't have worked from 1960-1982 so using long-dated backtests over short-dated ones is already out the window.
I used stocks and t-bills, the correlation isn't relevant for this analysis. The question was whether it is possible to obtain alpha on the stock market with simple volatility techniques and the answer was no. Maybe it is possible to beat the bond market, but since we have not seen a bond bear market in 40 years it would not be wise to draw conclusions based on recent bond market data.
Isn't the formula for alpha literally a function of beta? I am not sure why the correlation is not relevant, unless you mean that it won't necessarily be negative in the future, but that's too theoretical - my goal is to make money in the real world and I think this trend is likely to continue for obvious psychological reasons.
User avatar
cos
Posts: 225
Joined: Fri Aug 23, 2019 7:34 pm
Location: Boston
Contact:

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by cos »

hilink73 wrote: Sun Aug 30, 2020 1:46 pm So, a fixed allocation as in the original adventure is better? But, which % allocation, then?
That depends on your market beliefs and your utility function.
User avatar
Uncorrelated
Posts: 1061
Joined: Sun Oct 13, 2019 3:16 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Uncorrelated »

langlands wrote: Sun Aug 30, 2020 1:22 pm
Uncorrelated wrote: Sun Aug 30, 2020 1:18 pm
hilink73 wrote: Sun Aug 30, 2020 1:10 pm
Uncorrelated wrote: Sat Aug 29, 2020 4:55 am

If you want to have as little arbitrary assumptions as possible, throwing away target volatility and HFEA is probably the first thing you should do. There are certain assumptions that lead to target volatility, those assumptions are false (see above). There are certain assumptions that lead to a strategy similar to HFEA, but it is very unlikely those assumptions are right for you.
Maybe I'm misunderstanding something (and I admit I'm not understanding half of what you are talking about re utility function, etc.) but I though TV is exactly what's writing on the box: how much volatility you want to accept (read: you're able to stomach)? I never saw it from the max return perspective...

I'm in for a TV of 25% but after this recent downturn I guess I'm able to stomach much more volatility (I've also got experience with dabbling in crypto trading, so volatility doesn't bother me so much (in the long run)).
I won't change my strategy or target volatility yet, though... PV says 80% UPRO/20% TMF for the next month.
The problem is that TV is not mean-variance optimal. That means there are other strategies that have the same risk as TV, but a higher expected return.

There are two conditions that need to be met before TV becomes optimal: recent volatility is the best possible estimate of future volatility, return is proportional to volatility. Both assumptions are clearly false. If you want to take less risk, TV or other market timing approaches are not the answer, the answer is to use a lower equity allocation.
Hi Uncorrelated,

To be more explicit about your discussion of target variance vs. target volatility, we are essentially talking about the difference between scaling your allocation by 1/σ^2 vs. 1/σ right? Assuming your return assumptions are fixed, mean variance gives that scaling by 1/σ^2 is theoretically correct as you noted. Scaling by 1/σ can be thought of as scaling by 1/σ^2 and then multiplying by sigma, hence your statement that target volatility assumes that returns are proportional to volatility, which you say is empirically shown to be false.

Ok. But many people doing this adventure are using constant percent allocations. The way I see it, target volatility allocation is the geometric mean of constant percent allocation and the 1/σ^2 derived from Merton's portfolio. In other words, it can't be that bad since it's moving in the right direction. Your criticism is essentially that it doesn't adjust the allocation enough to volatility movements.
A constant asset allocation is fine, that's the best possible allocation if you don't believe in market timing.

If you do believe in market timing, then the best approach is to forecast future return and volatility, and adjust your asset allocation according to Merton's portfolio problem (or other methods of maximizing expected utility).

The idea that TV is a compromise between a constant allocation and 1/σ^2 is absurd. First, the recent volatility is not the best forecast of future volatility. Second, the average between a constant allocation and 1/σ^2 is not 1/σ. Third even if we assume 1/σ is mv-optimal then using anything other than the risk-free asset as the out-of-market asset results in a inefficient allocation. Fourth because 1/σ does not have a known theoretical basis, it cannot be implemented in conjunction with lifecycle investing which results in a significant loss of lifetime utility (IIRC, not using lifecycle investing results in between 0.5% and 1% lower certainty equivalent return per year).

My criticism isn't that TV doesn't do enough, my criticism is that TV doesn't make any sense at all. If you want to time the market, throw away TV and start with something that actually makes sense.

As mentioned, I measured a large drop in risk-adjusted returns between in-sample and out-of-sample volatility managed strategies. Meaning, you should probably not trust any figures of in-sample tests of volatility managed strategies.
hilink73 wrote: Sun Aug 30, 2020 1:46 pm
Uncorrelated wrote: Sun Aug 30, 2020 1:18 pm
So, a fixed allocation as in the original adventure is better? But, which % allocation, then?
A fixed allocation is definitely better than a poor market timing allocation such as PV.

The optimal allocation depends on your personal risk aversion and assumptions about future return. The best allocation with some return assumption can be viewed in this topic: A mean variance framework for portfolio optimization. My return assumptions might not necessary match your return assumptions. For individuals with high risk aversion, 0% UPRO is optimal. For individuals with low risk aversion, 100% UPRO is optimal.
Semantics wrote: Sun Aug 30, 2020 2:07 pm I don't think that's absurd at all. Almost the entire period covered by the paper you linked corresponds to obsolete US monetary policy regimes. Frankly, I place more value on following the Fed than I do on academic dogma when making investment decisions. This whole thread is about a strategy that wouldn't have worked from 1960-1982 so using long-dated backtests over short-dated ones is already out the window.
We were talking about equity and BAB returns, not about treasuries. There is no evidence that equity and BAB returns are influenced by US monetary policy regimes.

Between 2010 and 2020, equity returned 12-14% annually (returns in excess of T-bills). Between 1930-ish and 2020, equity returned 7% annually. Which of these numbers do you think are the best estimates for future return? Do you think we can do better than these two estimates? If so, how? How about bonds?
Isn't the formula for alpha literally a function of beta? I am not sure why the correlation is not relevant, unless you mean that it won't necessarily be negative in the future, but that's too theoretical - my goal is to make money in the real world and I think this trend is likely to continue for obvious psychological reasons.
I think you misunderstood what I was trying to show. I tested various TV algorithms with stocks and t-bills and found that it was not possible to beat a constant allocation out-of-sample. These results indicate that varying the equity allocation with TV does not work. TV might work for bonds, but given the low evidence for TV in general and the fact volatility appears to be more predictable in stocks than in bonds, I seriously doubt it.

Alpha wasn't really the right word to use here. Even if a strategy delivers alpha, it doesn't mean it's good. I optimized for max utility rather than alpha.
langlands
Posts: 573
Joined: Wed Apr 03, 2019 10:05 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by langlands »

Uncorrelated wrote: Sun Aug 30, 2020 2:45 pm
langlands wrote: Sun Aug 30, 2020 1:22 pm
Uncorrelated wrote: Sun Aug 30, 2020 1:18 pm
hilink73 wrote: Sun Aug 30, 2020 1:10 pm
Uncorrelated wrote: Sat Aug 29, 2020 4:55 am

If you want to have as little arbitrary assumptions as possible, throwing away target volatility and HFEA is probably the first thing you should do. There are certain assumptions that lead to target volatility, those assumptions are false (see above). There are certain assumptions that lead to a strategy similar to HFEA, but it is very unlikely those assumptions are right for you.
Maybe I'm misunderstanding something (and I admit I'm not understanding half of what you are talking about re utility function, etc.) but I though TV is exactly what's writing on the box: how much volatility you want to accept (read: you're able to stomach)? I never saw it from the max return perspective...

I'm in for a TV of 25% but after this recent downturn I guess I'm able to stomach much more volatility (I've also got experience with dabbling in crypto trading, so volatility doesn't bother me so much (in the long run)).
I won't change my strategy or target volatility yet, though... PV says 80% UPRO/20% TMF for the next month.
The problem is that TV is not mean-variance optimal. That means there are other strategies that have the same risk as TV, but a higher expected return.

There are two conditions that need to be met before TV becomes optimal: recent volatility is the best possible estimate of future volatility, return is proportional to volatility. Both assumptions are clearly false. If you want to take less risk, TV or other market timing approaches are not the answer, the answer is to use a lower equity allocation.
Hi Uncorrelated,

To be more explicit about your discussion of target variance vs. target volatility, we are essentially talking about the difference between scaling your allocation by 1/σ^2 vs. 1/σ right? Assuming your return assumptions are fixed, mean variance gives that scaling by 1/σ^2 is theoretically correct as you noted. Scaling by 1/σ can be thought of as scaling by 1/σ^2 and then multiplying by sigma, hence your statement that target volatility assumes that returns are proportional to volatility, which you say is empirically shown to be false.

Ok. But many people doing this adventure are using constant percent allocations. The way I see it, target volatility allocation is the geometric mean of constant percent allocation and the 1/σ^2 derived from Merton's portfolio. In other words, it can't be that bad since it's moving in the right direction. Your criticism is essentially that it doesn't adjust the allocation enough to volatility movements.
A constant asset allocation is fine, that's the best possible allocation if you don't believe in market timing.

If you do believe in market timing, then the best approach is to forecast future return and volatility, and adjust your asset allocation according to Merton's portfolio problem (or other methods of maximizing expected utility).

The idea that TV is a compromise between a constant allocation and 1/σ^2 is absurd. First, the recent volatility is not the best forecast of future volatility. Second, the average between a constant allocation and 1/σ^2 is not 1/σ. Third even if we assume 1/σ is mv-optimal then using anything other than the risk-free asset as the out-of-market asset results in a inefficient allocation. Fourth because 1/σ does not have a known theoretical basis, it cannot be implemented in conjunction with lifecycle investing which results in a significant loss of lifetime utility (IIRC, not using lifecycle investing results in between 0.5% and 1% lower certainty equivalent return per year).

My criticism isn't that TV doesn't do enough, my criticism is that TV doesn't make any sense at all. If you want to time the market, throw away TV and start with something that actually makes sense.

As mentioned, I measured a large drop in risk-adjusted returns between in-sample and out-of-sample volatility managed strategies. Meaning, you should probably not trust any figures of in-sample tests of volatility managed strategies.
You don't have to use recent volatility as a measure of future volatility. First of all, because of volatility clustering, it's not that bad an estimate and is the baseline in any predictive model. Second of all, you can just use the VIX or any other measure of implied volatility. I have no idea why you would say that it's absurd to claim 1/σ is a compromise between 1 and 1/σ^2. I mean, your prediction of future stock volatility doubles. Your return expectations haven't changed. Merton tells you to decrease your stock allocation by a factor of 4. You decrease it by a factor of 2. That's absurd? It's simply a fact that applying TV will lead to an allocation in between constant and Merton's.

I've noticed a rather dogmatic tendency in your way of thinking where you feel the need to put any problem in a precise framework and then find the optimal solution within that framework. Coming from a math background, I can of course sympathize. However, finance is quite a bit messier than math or physics and there is always the danger of model error lurking in the background. When it comes to these murkier topics, I find it much more illuminating to get the "gist" of what each model is telling me rather than actually believing the numerical answer it spits out. So for example, the idea that when volatility increases, it makes sense to decrease allocation to that asset.
User avatar
Uncorrelated
Posts: 1061
Joined: Sun Oct 13, 2019 3:16 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Uncorrelated »

langlands wrote: Sun Aug 30, 2020 2:55 pm
Uncorrelated wrote: Sun Aug 30, 2020 2:45 pm
You don't have to use recent volatility as a measure of future volatility. First of all, because of volatility clustering, it's not that bad an estimate and is the baseline in any predictive model. Second of all, you can just use the VIX or any other measure of implied volatility. I have no idea why you would say that it's absurd to claim 1/σ is a compromise between 1 and 1/σ^2. I mean, your prediction of future stock volatility doubles. Your return expectations haven't changed. Merton tells you to decrease your stock allocation by a factor of 4. You decrease it by a factor of 2. That's absurd? It's simply a fact that applying TV will lead to an allocation in between constant and Merton's.

I've noticed a rather dogmatic tendency in your way of thinking where you feel the need to put any problem in a precise framework and then find the optimal solution within that framework. Coming from a math background, I can of course sympathize. However, finance is quite a bit messier than math or physics and there is always the danger of model error lurking in the background. When it comes to these murkier topics, I find it much more illuminating to get the "gist" of what each model is telling me rather than actually believing the numerical answer it spits out. So for example, the idea that when volatility increases, it makes sense to decrease allocation to that asset.
I'm aware of volatility clustering. According to my tests using either past volatility or VIX as an estimate for future volatility doesn't work well enough to beat the market. Only with a high quality volatility forecast (better than VIX), merton's portfolio theory and a trading strategy that limits turnover rate, I was able to obtain higher utility than a constant allocation (semi out-of-sample). Without all of these three measures in place I was unable to get any convincing results.

My way of dealing with the murkiness of finance is to demand extremely high quality evidence. Merton's portfolio model is very rigorous, the claim that volatility can be forecasted can be empirically tested. Combining these observations hopefully results in a robust strategy. Choosing a random set of rules and testing it on dubious data usually results in a poor strategy, I have seen that go wrong far too often, even when established researchers are involved.

If you think it's dogmatic to demand high quality evidence before individuals gamble their money away in the most competitive environment on earth, then I prefer to be dogmatic.
Sushmit
Posts: 6
Joined: Sun Apr 05, 2020 3:43 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Sushmit »

Hello Everyone,

I used to follow these post in 2019 and up until June of 2020. I had some cash which was 10% of my total worth which I saved in an investment account for future use. Cue the drop in March and I went all in on 33% UPRO; 33% TQQQ and 33% FNGU. I did so because the market had crashed and I wanted to wait for the bounce back before I buy TMF. As you can imagine, my position has increased almost 5 times and now I am nervous to hedge this without incurring large amount of taxes. I tried reading up on TMF/VXX or other negative beta funds to balance out my portfolio but I am a little lost, can someone help me hedge this based on our current understanding?

P.S. I am 29 years old and have a long earning career ahead of me, I did not mind loosing the 10% of my total worth but now this is a substantial sum I would like to protect.

Edits : I am looking to add funds to balance this out to make the portfolio more robust. I can put upto 40% of the current value of this portfolio right away and wondering if it should go to TMF/VIXY etc.
Last edited by Sushmit on Sun Aug 30, 2020 9:54 pm, edited 1 time in total.
User avatar
Steve Reading
Posts: 2460
Joined: Fri Nov 16, 2018 10:20 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Steve Reading »

Uncorrelated wrote: Sun Aug 30, 2020 4:33 pm I'm aware of volatility clustering. According to my tests using either past volatility or VIX as an estimate for future volatility doesn't work well enough to beat the market.
I gotta say, I'm actually a little surprised that using implied future volatility along with the Merton equation, doesn't work a little better than just the Merton equation using some long-run volatility (say, 18%). I wouldn't use the VIX itself (that only encompasses the next 30 days I think) but maybe some more elegant implied volatility number, say the average of IV of the 3 month, 9 month and 15 month options ATM.

Any ways, that's neither here nor there.
"... so high a present discounted value of wealth, it is only prudent for him to put more into common stocks compared to his present tangible wealth, borrowing if necessary" - Paul Samuelson
User avatar
Meaty
Posts: 796
Joined: Mon Jul 22, 2013 7:35 pm
Location: Texas

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Meaty »

Sushmit wrote: Sun Aug 30, 2020 7:17 pm Hello Everyone,

I used to follow these post in 2019 and up until June of 2020. I had some cash which was 10% of my total worth which I saved in an investment account for future use. Cue the drop in March and I went all in on 33% UPRO; 33% TQQQ and 33% FNGU. I did so because the market had crashed and I wanted to wait for the bounce back before I buy TMF. As you can imagine, my position has increased almost 5 times and now I am nervous to hedge this without incurring large amount of taxes. I tried reading up on TMF/VXX or other negative beta funds to balance out my portfolio but I am a little lost, can someone help me hedge this based on our current understanding?

P.S. I am 29 years old and have a long earning career ahead of me, I did not mind loosing the 10% of my total worth but now this is a substantial sum I would like to protect.
Congrats on the good timing! Assuming this is in a taxable account, I think the only option would be to slowly transition your position to minimize the amount of realized gains. That said, you’d kick yourself if your correlated position crashed so if it were me I’d bite the bullet and just do it now
"Discipline equals Freedom" - Jocko Willink
langlands
Posts: 573
Joined: Wed Apr 03, 2019 10:05 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by langlands »

Sushmit wrote: Sun Aug 30, 2020 7:17 pm Hello Everyone,

I used to follow these post in 2019 and up until June of 2020. I had some cash which was 10% of my total worth which I saved in an investment account for future use. Cue the drop in March and I went all in on 33% UPRO; 33% TQQQ and 33% FNGU. I did so because the market had crashed and I wanted to wait for the bounce back before I buy TMF. As you can imagine, my position has increased almost 5 times and now I am nervous to hedge this without incurring large amount of taxes. I tried reading up on TMF/VXX or other negative beta funds to balance out my portfolio but I am a little lost, can someone help me hedge this based on our current understanding?

P.S. I am 29 years old and have a long earning career ahead of me, I did not mind loosing the 10% of my total worth but now this is a substantial sum I would like to protect.
It depends a bit on whether you wish to keep leverage in your portfolio permanently. If you don't, I believe there's no alternative to selling. Assuming you are a US investor, I suppose you are worried about short term capital gains. Two ways to temporarily hedge for a year to make your gains long term are puts and inverse etfs. I created a thread yesterday about this topic but haven't gotten any replies yet: viewtopic.php?f=10&t=324189. After a little more research however, my conclusion is that buying a put on SPY or QQQ will hedge your portfolio and does not count as an "offsetting position" for tax purposes. Importantly, since SPY isn't considered "substantially identical" to UPRO and QQQ isn't considered "substantially identical" to TQQQ for tax purposes, the long term/short term capital gains clock won't be reset when you buy the put. Similarly, buying an inverse etf does not count as an offsetting position. Basically, this is taking advantage of the IRS's lax attitude towards what ETFs count as "substantially identical." You'll probably want to read the relevant sections of https://www.irs.gov/publications/p550#e ... k100010629 for more details. It's not fun reading, but it contains important tax loss harvesting info for a non-vanilla investor (and your move in March certainly qualifies).

If you do wish to maintain leverage, you can do the math on whether your future contributions and selling some of your stocks that haven't appreciated as much can "dilute" your portfolio to keep the leverage to a level you're comfortable with.
Sushmit
Posts: 6
Joined: Sun Apr 05, 2020 3:43 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Sushmit »

Thanks for the prompt replies langland and Meaty! Let me clarify, I would like to continue my leveraged positions, however, as I focused on UPRO/TQQQ/FNGU; I dont have any etf/etn to balance it out. I was wondering what would be the recommended addition to my portfolio to balance it out; whether it is 55/45 with 45 TMF or something more appropriate that back tests better. I was initially planning to buy VIXY but realized it looses value to rollovers.

For additional information, I will be putting a part of my annual salary into this adventure which corresponds to 40% of current value of my holdings, and I can put that amount in right away to balance it.
tomphilly
Posts: 59
Joined: Wed Aug 05, 2020 11:29 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by tomphilly »

Sushmit wrote: Sun Aug 30, 2020 9:53 pm Thanks for the prompt replies langland and Meaty! Let me clarify, I would like to continue my leveraged positions, however, as I focused on UPRO/TQQQ/FNGU; I dont have any etf/etn to balance it out. I was wondering what would be the recommended addition to my portfolio to balance it out; whether it is 55/45 with 45 TMF or something more appropriate that back tests better. I was initially planning to buy VIXY but realized it looses value to rollovers.

For additional information, I will be putting a part of my annual salary into this adventure which corresponds to 40% of current value of my holdings, and I can put that amount in right away to balance it.
TMF remains the supreme volatility hedge to the leveraged equities, though a variety of alternatives have been discussed here but mostly shot down - they are cash, metals, BTAL, TYD and puts on TMV. I would start with 45% TMF - that is the classic strategy that is discussed at length here. You can always modify that after further research. Note that there is rising skepticism about TMF's future prospects due to rates being so low.
Last edited by tomphilly on Sun Aug 30, 2020 10:34 pm, edited 4 times in total.
Investing Lawyer
Posts: 9
Joined: Wed Apr 08, 2020 9:35 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Investing Lawyer »

Uncorrelated wrote: Sun Aug 30, 2020 4:21 am The optimal leverage depends on your risk aversion. If you have a logarithmic utility function, then the optimal leverage for equities is around 2x. If you are optimizing for max expected return, then there no upper bound for the optimal leverage. Most users don't have a log utility function, which makes optimizing for higher CAGR meaningless. (optimizing for CAGR is equivalent to optimizing for log wealth).
There is always an optimal leverage ratio for maxing CAGR, else one will go bust or have too large of a draw down to recover from. It's Sharpe divided by the standard dev again, i.e. (R(asset or portoflio)-Rf)/Variance, which implicitly assumes a continuous re balancing to maintain a constant ratio and a continuous distribution, so actual optimal leverage will be lower.
000
Posts: 2842
Joined: Thu Jul 23, 2020 12:04 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by 000 »

Haven't stopped by in a while. Is anyone bailing out of the HF EA at the current highs?
tomphilly
Posts: 59
Joined: Wed Aug 05, 2020 11:29 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by tomphilly »

000 wrote: Sun Aug 30, 2020 10:33 pm Haven't stopped by in a while. Is anyone bailing out of the HF EA at the current highs?
I'm nervous about an imminent correction but I learned my lesson sitting on the sideline before. My main concern is the SP500 is now 37% tech.
User avatar
willthrill81
Posts: 20980
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by willthrill81 »

000 wrote: Sun Aug 30, 2020 10:33 pm Haven't stopped by in a while. Is anyone bailing out of the HF EA at the current highs?
Stocks have historically spent much more time at or near market highs than crashing. If highs make someone skittish in the slightest, then any strategy remotely resembling this one is not appropriate for such an investor.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings
000
Posts: 2842
Joined: Thu Jul 23, 2020 12:04 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by 000 »

tomphilly wrote: Sun Aug 30, 2020 10:44 pm
000 wrote: Sun Aug 30, 2020 10:33 pm Haven't stopped by in a while. Is anyone bailing out of the HF EA at the current highs?
I'm nervous about an imminent correction but I learned my lesson sitting on the sideline before. My main concern is the SP500 is now 37% tech.
When I said "bailing out", I meant leaving the EA and going to an unleveraged allocation, not cash.

The tech concentration worries me too. It seems some major bad news (true or fake) concerning one of the big tech companies could really hurt.
000
Posts: 2842
Joined: Thu Jul 23, 2020 12:04 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by 000 »

willthrill81 wrote: Sun Aug 30, 2020 10:47 pm
000 wrote: Sun Aug 30, 2020 10:33 pm Haven't stopped by in a while. Is anyone bailing out of the HF EA at the current highs?
Stocks have historically spent much more time at or near market highs than crashing. If highs make someone skittish in the slightest, then any strategy remotely resembling this one is not appropriate for such an investor.
Better to bail at the highs than the lows. Concerning the HF EA in particular, one might decide that 3x LTT has already had much of its run.
Sushmit
Posts: 6
Joined: Sun Apr 05, 2020 3:43 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Sushmit »

tomphilly wrote: Sun Aug 30, 2020 10:27 pm
Sushmit wrote: Sun Aug 30, 2020 9:53 pm Thanks for the prompt replies langland and Meaty! Let me clarify, I would like to continue my leveraged positions, however, as I focused on UPRO/TQQQ/FNGU; I dont have any etf/etn to balance it out. I was wondering what would be the recommended addition to my portfolio to balance it out; whether it is 55/45 with 45 TMF or something more appropriate that back tests better. I was initially planning to buy VIXY but realized it looses value to rollovers.

For additional information, I will be putting a part of my annual salary into this adventure which corresponds to 40% of current value of my holdings, and I can put that amount in right away to balance it.
TMF remains the supreme volatility hedge to the leveraged equities, though a variety of alternatives have been discussed here but mostly shot down - they are cash, metals, BTAL, TYD and puts on TMV. I would start with 45% TMF - that is the classic strategy that is discussed at length here. You can always modify that after further research. Note that there is rising skepticism about TMF's future prospects due to rates being so low.
Thanks a lot tomphilly. I am still wondering what would be the best step for me to take. I will start buying tmf over the next few days and hope I get in before any major correction
User avatar
willthrill81
Posts: 20980
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by willthrill81 »

000 wrote: Sun Aug 30, 2020 10:50 pm
willthrill81 wrote: Sun Aug 30, 2020 10:47 pm
000 wrote: Sun Aug 30, 2020 10:33 pm Haven't stopped by in a while. Is anyone bailing out of the HF EA at the current highs?
Stocks have historically spent much more time at or near market highs than crashing. If highs make someone skittish in the slightest, then any strategy remotely resembling this one is not appropriate for such an investor.
Better to bail at the highs than the lows. Concerning the HF EA in particular, one might decide that 3x LTT has already had much of its run.
Best not to start such a strategy without being prepared to endure the wild ride.

The point of TMF is not its returns, especially since its expense ratio is 1.05%, which will greatly erode the pitiful current yield from LTT. Rather, the point is to provide a counter-balancing effect for 3x leveraged stocks (i.e. UPRO).
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings
tomphilly
Posts: 59
Joined: Wed Aug 05, 2020 11:29 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by tomphilly »

willthrill81 wrote: Sun Aug 30, 2020 11:08 pm The point of TMF is not its returns, especially since its expense ratio is 1.05%, which will greatly erode the pitiful current yield from LTT. Rather, the point is to provide a counter-balancing effect for 3x leveraged stocks (i.e. UPRO).
Are you still employing 80/20 UPRO/TMF at 25% target volatility? I decided to switch to a variation of this last week and bought the PV subscription for the TV signal.
Repege
Posts: 2
Joined: Sun Aug 30, 2020 11:20 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Repege »

willthrill81 wrote: Sun Aug 30, 2020 11:08 pm
000 wrote: Sun Aug 30, 2020 10:50 pm
willthrill81 wrote: Sun Aug 30, 2020 10:47 pm
000 wrote: Sun Aug 30, 2020 10:33 pm Haven't stopped by in a while. Is anyone bailing out of the HF EA at the current highs?
Stocks have historically spent much more time at or near market highs than crashing. If highs make someone skittish in the slightest, then any strategy remotely resembling this one is not appropriate for such an investor.
Better to bail at the highs than the lows. Concerning the HF EA in particular, one might decide that 3x LTT has already had much of its run.
Best not to start such a strategy without being prepared to endure the wild ride.

The point of TMF is not its returns, especially since its expense ratio is 1.05%, which will greatly erode the pitiful current yield from LTT. Rather, the point is to provide a counter-balancing effect for 3x leveraged stocks (i.e. UPRO).
I started with 60%UPRO/40%TMF, I swifted to 60%URPO/80%EDV recently. In consideration of 1.margin rates lower than 1.5% now. 2. EDV has much lower expense ratio and higher yield.
Ciel
Posts: 114
Joined: Thu Jun 27, 2013 11:14 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Ciel »

000 wrote: Sun Aug 30, 2020 10:33 pm Haven't stopped by in a while. Is anyone bailing out of the HF EA at the current highs?
I'm probably bailing this week. I went in with part of my Roth in April, and now I'm up about 40%. At the time I really did intend to go along with the intent of this -- committing a certain amount to the wild volatility ride no matter what. However with stock valuations what they are, and interest rates what they are, I'm simply more comfortable going 100% unleveraged total market instead. I'll be the first to admit that I have not done a deep dive into the nitty-gritty details of this strategy, and that's all the more reason I shouldn't be in it anyway. So, cold feet, not for me, etc. I can't be the only one.
User avatar
Uncorrelated
Posts: 1061
Joined: Sun Oct 13, 2019 3:16 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Uncorrelated »

Steve Reading wrote: Sun Aug 30, 2020 7:27 pm
Uncorrelated wrote: Sun Aug 30, 2020 4:33 pm I'm aware of volatility clustering. According to my tests using either past volatility or VIX as an estimate for future volatility doesn't work well enough to beat the market.
I gotta say, I'm actually a little surprised that using implied future volatility along with the Merton equation, doesn't work a little better than just the Merton equation using some long-run volatility (say, 18%). I wouldn't use the VIX itself (that only encompasses the next 30 days I think) but maybe some more elegant implied volatility number, say the average of IV of the 3 month, 9 month and 15 month options ATM.

Any ways, that's neither here nor there.
That surprised me as well. It took me a while to realize what is wrong: VIX is a measure of risk-neutral volatility (that means that it assumes options are priced by risk-neutral actors), but option traders are not risk neutral, this causes VIX to consistently under-estimate market volatility. Based on this explanation you might think that scaling VIX by some constant factor fixes this issue, but it turns out this doesn't work since the risk aversion of option traders changes quite substantially over time.

I was able to get much better results with a simple linear model. All tests that I have ran that include VIX as a feature to the model resulted in worse out-of-sample results. Shows that you should always test your hypothesis.
Investing Lawyer wrote: Sun Aug 30, 2020 10:30 pm
Uncorrelated wrote: Sun Aug 30, 2020 4:21 am The optimal leverage depends on your risk aversion. If you have a logarithmic utility function, then the optimal leverage for equities is around 2x. If you are optimizing for max expected return, then there no upper bound for the optimal leverage. Most users don't have a log utility function, which makes optimizing for higher CAGR meaningless. (optimizing for CAGR is equivalent to optimizing for log wealth).
There is always an optimal leverage ratio for maxing CAGR, else one will go bust or have too large of a draw down to recover from. It's Sharpe divided by the standard dev again, i.e. (R(asset or portoflio)-Rf)/Variance, which implicitly assumes a continuous re balancing to maintain a constant ratio and a continuous distribution, so actual optimal leverage will be lower.
Of course. But the question is why to optimize for CAGR to begin with. Unless you have very good reasons to assume you do indeed have a logarithmic utility function, the most likely answer is that you should not optimize for CAGR.

The only good reason I can come up with to have an asset allocation this aggressive is as part of a lifecycle investing strategy, But users who follow lifecycle investing only adopt such a strategy for a short while. In the context of lifecycle optimizing for γ = 1 (log utility) is no more meaningful than optimizing for γ = 1.01 or γ = 0.99.
User avatar
Meaty
Posts: 796
Joined: Mon Jul 22, 2013 7:35 pm
Location: Texas

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Meaty »

Ciel wrote: Mon Aug 31, 2020 1:44 am
000 wrote: Sun Aug 30, 2020 10:33 pm Haven't stopped by in a while. Is anyone bailing out of the HF EA at the current highs?
I'm probably bailing this week. I went in with part of my Roth in April, and now I'm up about 40%. At the time I really did intend to go along with the intent of this -- committing a certain amount to the wild volatility ride no matter what. However with stock valuations what they are, and interest rates what they are, I'm simply more comfortable going 100% unleveraged total market instead. I'll be the first to admit that I have not done a deep dive into the nitty-gritty details of this strategy, and that's all the more reason I shouldn't be in it anyway. So, cold feet, not for me, etc. I can't be the only one.
Aside from the volatility leverage creates, how does bailing really change anything? The classic portfolio pushed here (stock/bond indexing) is just as impacted by valuations and interest rates.
"Discipline equals Freedom" - Jocko Willink
Sushmit
Posts: 6
Joined: Sun Apr 05, 2020 3:43 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Sushmit »

tomphilly wrote: Sun Aug 30, 2020 11:39 pm
willthrill81 wrote: Sun Aug 30, 2020 11:08 pm The point of TMF is not its returns, especially since its expense ratio is 1.05%, which will greatly erode the pitiful current yield from LTT. Rather, the point is to provide a counter-balancing effect for 3x leveraged stocks (i.e. UPRO).
Are you still employing 80/20 UPRO/TMF at 25% target volatility? I decided to switch to a variation of this last week and bought the PV subscription for the TV signal.
What allocation is that ? Is there something I can read up on
tomphilly
Posts: 59
Joined: Wed Aug 05, 2020 11:29 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by tomphilly »

Sushmit wrote: Mon Aug 31, 2020 7:02 am
tomphilly wrote: Sun Aug 30, 2020 11:39 pm
willthrill81 wrote: Sun Aug 30, 2020 11:08 pm The point of TMF is not its returns, especially since its expense ratio is 1.05%, which will greatly erode the pitiful current yield from LTT. Rather, the point is to provide a counter-balancing effect for 3x leveraged stocks (i.e. UPRO).
Are you still employing 80/20 UPRO/TMF at 25% target volatility? I decided to switch to a variation of this last week and bought the PV subscription for the TV signal.
What allocation is that ? Is there something I can read up on
The strategy is similar to HFEA, except instead of quarterly re-balancing, you balance the ratio according to a volatility signal. In low volatility periods, your allocation will be your default allocation, e.g 80/20 UPRO/TMF, whilst in volatile periods, your ratio could go from anywhere between the default to the complete opposite - e.g 20/80 UPRO/TMF. The ratio to use is reported in portfolio visualizer when you have the paid subscription - alternatively if you're a smart math/finance person you could probably determine the signal yourself and save yourself the yearly subscription - regrettably I'm not smart enough.

Right now the volatility signal is risk on - meaning it's 80/20, pedal-to-the-metal time. During the calamitous period of April, for example, it was 22/78. Note that you could experience much greater drawdowns with this than with classic HFEA, so be aware of the risk before implementing it.
Texanbybirth
Posts: 1394
Joined: Tue Apr 14, 2015 12:07 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Texanbybirth »

tomphilly wrote: Mon Aug 31, 2020 8:35 am
Sushmit wrote: Mon Aug 31, 2020 7:02 am
tomphilly wrote: Sun Aug 30, 2020 11:39 pm
willthrill81 wrote: Sun Aug 30, 2020 11:08 pm The point of TMF is not its returns, especially since its expense ratio is 1.05%, which will greatly erode the pitiful current yield from LTT. Rather, the point is to provide a counter-balancing effect for 3x leveraged stocks (i.e. UPRO).
Are you still employing 80/20 UPRO/TMF at 25% target volatility? I decided to switch to a variation of this last week and bought the PV subscription for the TV signal.
What allocation is that ? Is there something I can read up on
The strategy is similar to HFEA, except instead of quarterly re-balancing, you balance the ratio according to a volatility signal. In low volatility periods, your allocation will be your default allocation, e.g 80/20 UPRO/TMF, whilst in volatile periods, your ratio could go from anywhere between the default to the complete opposite - e.g 20/80 UPRO/TMF. The ratio to use is reported in portfolio visualizer when you have the paid subscription - alternatively if you're a smart math/finance person you could probably determine the signal yourself and save yourself the yearly subscription - regrettably I'm not smart enough.

Right now the volatility signal is risk on - meaning it's 80/20, pedal-to-the-metal time. During the calamitous period of April, for example, it was 22/78. Note that you could experience much greater drawdowns with this than with classic HFEA, so be aware of the risk before implementing it.
There was a poster, danielc, trying to come up with a spreadsheet that would allow you to calculate the allocations monthly. I don't know what happened to him, or his spreadsheet, but if you really want to save the annual subscription it might be worth a little bit of time to search the forums/send him a PM.
“The strong cannot be brave. Only the weak can be brave; and yet again, in practice, only those who can be brave can be trusted, in time of doubt, to be strong.“ - GK Chesterton
User avatar
Steve Reading
Posts: 2460
Joined: Fri Nov 16, 2018 10:20 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Steve Reading »

Uncorrelated wrote: Mon Aug 31, 2020 1:52 am VIX is a measure of risk-neutral volatility (that means that it assumes options are priced by risk-neutral actors), but option traders are not risk neutral, this causes VIX to consistently under-estimate market volatility. Based on this explanation you might think that scaling VIX by some constant factor fixes this issue, but it turns out this doesn't work since the risk aversion of option traders changes quite substantially over time.
Gotcha. So you tried multiplying the VIX by some constant factor greater than 1.0 and used that as the future volatility estimate, and the results were inferior as well?
"... so high a present discounted value of wealth, it is only prudent for him to put more into common stocks compared to his present tangible wealth, borrowing if necessary" - Paul Samuelson
User avatar
willthrill81
Posts: 20980
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by willthrill81 »

tomphilly wrote: Sun Aug 30, 2020 11:39 pm
willthrill81 wrote: Sun Aug 30, 2020 11:08 pm The point of TMF is not its returns, especially since its expense ratio is 1.05%, which will greatly erode the pitiful current yield from LTT. Rather, the point is to provide a counter-balancing effect for 3x leveraged stocks (i.e. UPRO).
Are you still employing 80/20 UPRO/TMF at 25% target volatility? I decided to switch to a variation of this last week and bought the PV subscription for the TV signal.
Yes, although I'm slowly starting to incorporate other 3x leveraged stock ETFs than just UPRO, selecting the one with the highest returns over the 1 month look back period. Right now, I'm in FNGU, a 3x LCG ETF, which is up nearly 23% in August alone.
tomphilly wrote: Mon Aug 31, 2020 8:35 am
Sushmit wrote: Mon Aug 31, 2020 7:02 am
tomphilly wrote: Sun Aug 30, 2020 11:39 pm
willthrill81 wrote: Sun Aug 30, 2020 11:08 pm The point of TMF is not its returns, especially since its expense ratio is 1.05%, which will greatly erode the pitiful current yield from LTT. Rather, the point is to provide a counter-balancing effect for 3x leveraged stocks (i.e. UPRO).
Are you still employing 80/20 UPRO/TMF at 25% target volatility? I decided to switch to a variation of this last week and bought the PV subscription for the TV signal.
What allocation is that ? Is there something I can read up on
The strategy is similar to HFEA, except instead of quarterly re-balancing, you balance the ratio according to a volatility signal. In low volatility periods, your allocation will be your default allocation, e.g 80/20 UPRO/TMF, whilst in volatile periods, your ratio could go from anywhere between the default to the complete opposite - e.g 20/80 UPRO/TMF. The ratio to use is reported in portfolio visualizer when you have the paid subscription - alternatively if you're a smart math/finance person you could probably determine the signal yourself and save yourself the yearly subscription - regrettably I'm not smart enough.

Right now the volatility signal is risk on - meaning it's 80/20, pedal-to-the-metal time. During the calamitous period of April, for example, it was 22/78. Note that you could experience much greater drawdowns with this than with classic HFEA, so be aware of the risk before implementing it.
The drawdowns could be smaller though due to the ability to adjust the allocation according to volatility.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings
perfectuncertainty
Posts: 132
Joined: Sun Feb 04, 2018 7:44 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by perfectuncertainty »

Texanbybirth wrote: Mon Aug 31, 2020 8:59 am
tomphilly wrote: Mon Aug 31, 2020 8:35 am
Sushmit wrote: Mon Aug 31, 2020 7:02 am
tomphilly wrote: Sun Aug 30, 2020 11:39 pm
willthrill81 wrote: Sun Aug 30, 2020 11:08 pm The point of TMF is not its returns, especially since its expense ratio is 1.05%, which will greatly erode the pitiful current yield from LTT. Rather, the point is to provide a counter-balancing effect for 3x leveraged stocks (i.e. UPRO).
Are you still employing 80/20 UPRO/TMF at 25% target volatility? I decided to switch to a variation of this last week and bought the PV subscription for the TV signal.
What allocation is that ? Is there something I can read up on
The strategy is similar to HFEA, except instead of quarterly re-balancing, you balance the ratio according to a volatility signal. In low volatility periods, your allocation will be your default allocation, e.g 80/20 UPRO/TMF, whilst in volatile periods, your ratio could go from anywhere between the default to the complete opposite - e.g 20/80 UPRO/TMF. The ratio to use is reported in portfolio visualizer when you have the paid subscription - alternatively if you're a smart math/finance person you could probably determine the signal yourself and save yourself the yearly subscription - regrettably I'm not smart enough.

Right now the volatility signal is risk on - meaning it's 80/20, pedal-to-the-metal time. During the calamitous period of April, for example, it was 22/78. Note that you could experience much greater drawdowns with this than with classic HFEA, so be aware of the risk before implementing it.
There was a poster, danielc, trying to come up with a spreadsheet that would allow you to calculate the allocations monthly. I don't know what happened to him, or his spreadsheet, but if you really want to save the annual subscription it might be worth a little bit of time to search the forums/send him a PM.
I have a spreadsheet that does it.
Sushmit
Posts: 6
Joined: Sun Apr 05, 2020 3:43 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Sushmit »

perfectuncertainty wrote: Mon Aug 31, 2020 10:39 am
Texanbybirth wrote: Mon Aug 31, 2020 8:59 am
tomphilly wrote: Mon Aug 31, 2020 8:35 am
Sushmit wrote: Mon Aug 31, 2020 7:02 am
tomphilly wrote: Sun Aug 30, 2020 11:39 pm

Are you still employing 80/20 UPRO/TMF at 25% target volatility? I decided to switch to a variation of this last week and bought the PV subscription for the TV signal.
What allocation is that ? Is there something I can read up on
The strategy is similar to HFEA, except instead of quarterly re-balancing, you balance the ratio according to a volatility signal. In low volatility periods, your allocation will be your default allocation, e.g 80/20 UPRO/TMF, whilst in volatile periods, your ratio could go from anywhere between the default to the complete opposite - e.g 20/80 UPRO/TMF. The ratio to use is reported in portfolio visualizer when you have the paid subscription - alternatively if you're a smart math/finance person you could probably determine the signal yourself and save yourself the yearly subscription - regrettably I'm not smart enough.

Right now the volatility signal is risk on - meaning it's 80/20, pedal-to-the-metal time. During the calamitous period of April, for example, it was 22/78. Note that you could experience much greater drawdowns with this than with classic HFEA, so be aware of the risk before implementing it.
There was a poster, danielc, trying to come up with a spreadsheet that would allow you to calculate the allocations monthly. I don't know what happened to him, or his spreadsheet, but if you really want to save the annual subscription it might be worth a little bit of time to search the forums/send him a PM.
I have a spreadsheet that does it.
Can you share it please, I am curious about it too
langlands
Posts: 573
Joined: Wed Apr 03, 2019 10:05 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by langlands »

Uncorrelated wrote: Mon Aug 31, 2020 1:52 am
Steve Reading wrote: Sun Aug 30, 2020 7:27 pm
Uncorrelated wrote: Sun Aug 30, 2020 4:33 pm I'm aware of volatility clustering. According to my tests using either past volatility or VIX as an estimate for future volatility doesn't work well enough to beat the market.
I gotta say, I'm actually a little surprised that using implied future volatility along with the Merton equation, doesn't work a little better than just the Merton equation using some long-run volatility (say, 18%). I wouldn't use the VIX itself (that only encompasses the next 30 days I think) but maybe some more elegant implied volatility number, say the average of IV of the 3 month, 9 month and 15 month options ATM.

Any ways, that's neither here nor there.
That surprised me as well. It took me a while to realize what is wrong: VIX is a measure of risk-neutral volatility (that means that it assumes options are priced by risk-neutral actors), but option traders are not risk neutral, this causes VIX to consistently under-estimate market volatility. Based on this explanation you might think that scaling VIX by some constant factor fixes this issue, but it turns out this doesn't work since the risk aversion of option traders changes quite substantially over time.

I was able to get much better results with a simple linear model. All tests that I have ran that include VIX as a feature to the model resulted in worse out-of-sample results. Shows that you should always test your hypothesis.
If you are comparing constant percent allocation vs. target volatility or target variance, I think it's very important that you do all comparisons with daily rebalancing (if you did already, then great). Recall the numerous posts from people very happy about the fortuitous timing of their quarterly rebalancing this year along with a smaller number of other people who were less happy with their unfortunate timing. There's a lot of variance to the quarterly rebalancing back tests that I hope you took into account.

Notice as well that the quarterly rebalanced constant percent allocation that HEDGEFUNDIE uses has a very important ad-hoc volatility adjustment built into it. When volatility is decreasing, UPRO tends to be rising and when volatility is increasing, UPRO tends to be falling. If you are in between rebalancing periods, your UPRO/TMF portfolio is automatically following a volatility adjustment. In other words during the market crash in March if you aren't rebalancing UPRO, you are automatically following a target volatility type strategy since you are allowing your URPO allocation to decrease drastically (via market price) while volatility is high.

Edit: Target variance isn't really the right term (obviously targeting a constant volatility or variance is the same thing). By target variance, I just mean scaling by 1/vol^2 since target volatility is in essence scaling by 1/vol.
Last edited by langlands on Mon Aug 31, 2020 12:49 pm, edited 1 time in total.
perfectuncertainty
Posts: 132
Joined: Sun Feb 04, 2018 7:44 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by perfectuncertainty »

Sushmit wrote: Mon Aug 31, 2020 10:57 am
perfectuncertainty wrote: Mon Aug 31, 2020 10:39 am

I have a spreadsheet that does it.
Can you share it please, I am curious about it too
Link

I havent updated it - since I don't use TV. I use a differrent method using inverse volatility
tomphilly
Posts: 59
Joined: Wed Aug 05, 2020 11:29 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by tomphilly »

35% target volatility results in a noteworthy improvement in CAGR with no real difference in drawdown or Sharpe, at least since 2010.

Image

Note: I'm contributing 3k a month here.
Sushmit
Posts: 6
Joined: Sun Apr 05, 2020 3:43 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Sushmit »

perfectuncertainty wrote: Mon Aug 31, 2020 12:41 pm
Sushmit wrote: Mon Aug 31, 2020 10:57 am
perfectuncertainty wrote: Mon Aug 31, 2020 10:39 am

I have a spreadsheet that does it.
Can you share it please, I am curious about it too
Link

I havent updated it - since I don't use TV. I use a differrent method using inverse volatility
Thanks!! Is there some post where you detail your thought process I can read up on?
langlands
Posts: 573
Joined: Wed Apr 03, 2019 10:05 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by langlands »

tomphilly wrote: Mon Aug 31, 2020 8:35 am
Sushmit wrote: Mon Aug 31, 2020 7:02 am
tomphilly wrote: Sun Aug 30, 2020 11:39 pm
willthrill81 wrote: Sun Aug 30, 2020 11:08 pm The point of TMF is not its returns, especially since its expense ratio is 1.05%, which will greatly erode the pitiful current yield from LTT. Rather, the point is to provide a counter-balancing effect for 3x leveraged stocks (i.e. UPRO).
Are you still employing 80/20 UPRO/TMF at 25% target volatility? I decided to switch to a variation of this last week and bought the PV subscription for the TV signal.
What allocation is that ? Is there something I can read up on
The strategy is similar to HFEA, except instead of quarterly re-balancing, you balance the ratio according to a volatility signal. In low volatility periods, your allocation will be your default allocation, e.g 80/20 UPRO/TMF, whilst in volatile periods, your ratio could go from anywhere between the default to the complete opposite - e.g 20/80 UPRO/TMF. The ratio to use is reported in portfolio visualizer when you have the paid subscription - alternatively if you're a smart math/finance person you could probably determine the signal yourself and save yourself the yearly subscription - regrettably I'm not smart enough.

Right now the volatility signal is risk on - meaning it's 80/20, pedal-to-the-metal time. During the calamitous period of April, for example, it was 22/78. Note that you could experience much greater drawdowns with this than with classic HFEA, so be aware of the risk before implementing it.
Hi tomphilly,

After looking into more detail at what portfolio visualizer actually means by target volatility, I regret to say that you and I are both mistaken about what it does. Someone who already has experience doing this should chime in as well to make sure everyone is on the same page.

My original understanding of target volatility is that you want to adjust the ratios of the assets (as you said) so that during times of high volatility, you'd decrease the amount of UPRO relative to TMF. I think this makes sense.

Looking at this example run however shows that what's actually going on is that portfolio visualizer is keeping the ratio of assets the same and is only turning a single global knob to target the volatility of the overall portfolio. (Go to the Timing Periods tab where you can see the change in allocations). https://www.portfoliovisualizer.com/tes ... tion2_1=45

Uncorrelated, is your interpretation of target volatility mine and tomphilly's or portfolio visualizer's? Maybe we're all talking past each other and don't even know it.
tomphilly
Posts: 59
Joined: Wed Aug 05, 2020 11:29 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by tomphilly »

langlands wrote: Mon Aug 31, 2020 1:55 pm Hi tomphilly,

After looking into more detail at what portfolio visualizer actually means by target volatility, I regret to say that you and I are both mistaken about what it does. Someone who already has experience doing this should chime in as well to make sure everyone is on the same page.
Thanks, you're right, something is up with PV - I looked again and it's using "realized annualized volatility" which varies between different simulations. Your backtest was showing a 22.36% volatility whilst mine was 13.37% at the last signal date of 08/28/2020. I assumed this was a constant derived from SPY or VIX or somerthing. I vaguely remember there was a dropdown were you could select the signal asset, I used to put in SPY. Definitely need to investigate why this is.

By the way, unrelated to the volatility signal issue, a couple of differences between your and my backtests - I have trade execution at next close, and I have downside volatility only. The former, I believe, is beneficial because you are closely following the volatility signal guidance (rather than waiting until the end of the month), the latter, I don't fully understand, but it resulted in far better CAGR's.

Are you a PV subscriber? (can you see the current signal at the bottom of the timing period tab?)
perfectuncertainty
Posts: 132
Joined: Sun Feb 04, 2018 7:44 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by perfectuncertainty »

Sushmit wrote: Mon Aug 31, 2020 1:36 pm
perfectuncertainty wrote: Mon Aug 31, 2020 12:41 pm
Sushmit wrote: Mon Aug 31, 2020 10:57 am
perfectuncertainty wrote: Mon Aug 31, 2020 10:39 am

I have a spreadsheet that does it.
Can you share it please, I am curious about it too
Link

I havent updated it - since I don't use TV. I use a differrent method using inverse volatility
Thanks!! Is there some post where you detail your thought process I can read up on?
Read this thread
perfectuncertainty
Posts: 132
Joined: Sun Feb 04, 2018 7:44 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by perfectuncertainty »

tomphilly wrote: Mon Aug 31, 2020 2:23 pm
langlands wrote: Mon Aug 31, 2020 1:55 pm Hi tomphilly,

After looking into more detail at what portfolio visualizer actually means by target volatility, I regret to say that you and I are both mistaken about what it does. Someone who already has experience doing this should chime in as well to make sure everyone is on the same page.
Thanks, you're right, something is up with PV - I looked again and it's using "realized annualized volatility" which varies between different simulations. Your backtest was showing a 22.36% volatility whilst mine was 13.37% at the last signal date of 08/28/2020. I assumed this was a constant derived from SPY or VIX or somerthing. I vaguely remember there was a dropdown were you could select the signal asset, I used to put in SPY. Definitely need to investigate why this is.

By the way, unrelated to the volatility signal issue, a couple of differences between your and my backtests - I have trade execution at next close, and I have downside volatility only. The former, I believe, is beneficial because you are closely following the volatility signal guidance (rather than waiting until the end of the month), the latter, I don't fully understand, but it resulted in far better CAGR's.

Are you a PV subscriber? (can you see the current signal at the bottom of the timing period tab?)
It is standard to annualize the volatility even if the calculation is based on a limited time frame. SD of returns over the period x SQRT(252)
langlands
Posts: 573
Joined: Wed Apr 03, 2019 10:05 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by langlands »

tomphilly wrote: Mon Aug 31, 2020 2:23 pm
langlands wrote: Mon Aug 31, 2020 1:55 pm Hi tomphilly,

After looking into more detail at what portfolio visualizer actually means by target volatility, I regret to say that you and I are both mistaken about what it does. Someone who already has experience doing this should chime in as well to make sure everyone is on the same page.
Thanks, you're right, something is up with PV - I looked again and it's using "realized annualized volatility" which varies between different simulations. Your backtest was showing a 22.36% volatility whilst mine was 13.37% at the last signal date of 08/28/2020. I assumed this was a constant derived from SPY or VIX or somerthing. I vaguely remember there was a dropdown were you could select the signal asset, I used to put in SPY. Definitely need to investigate why this is.

By the way, unrelated to the volatility signal issue, a couple of differences between your and my backtests - I have trade execution at next close, and I have downside volatility only. The former, I believe, is beneficial because you are closely following the volatility signal guidance (rather than waiting until the end of the month), the latter, I don't fully understand, but it resulted in far better CAGR's.

Are you a PV subscriber? (can you see the current signal at the bottom of the timing period tab?)
No, I'm not a PV subscriber.

Perhaps PV is calculating the volatility incorrectly (which would be an additional problem), but my main point was that target volatility won't adjust your UPRO/TMF percentage from 80/20 to 20/80 in high volatility times as you wrote. Instead, it will adjust your 80/20 UPRO/TMF portfolio to something like 40/10/50 UPRO/TMF/CASH. Now that I think about it, it's not that big a deal since the the difference between TMF and cash I would say is less than the difference between TMF and UPRO. But it's certainly still significant.
tomphilly
Posts: 59
Joined: Wed Aug 05, 2020 11:29 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by tomphilly »

langlands wrote: Mon Aug 31, 2020 2:48 pm No, I'm not a PV subscriber.
Ok, when you're a PV subscriber you will see the current required allocation at the latest signal date, at the very bottom of the Timing Periods tab.
langlands wrote: Mon Aug 31, 2020 2:48 pm Perhaps PV is calculating the volatility incorrectly (which would be an additional problem), but my main point was that target volatility won't adjust your UPRO/TMF percentage from 80/20 to 20/80 in high volatility times as you wrote. Instead, it will adjust your 80/20 UPRO/TMF portfolio to something like 40/10/50 UPRO/TMF/CASH. Now that I think about it, it's not that big a deal since the the difference between TMF and cash I would say is less than the difference between TMF and UPRO. But it's certainly still significant.
You have cash in your simulation because you specified it as the "out of market asset". I specified TMF as the out of market asset, no cash is involved. This keeps the strategy more like HFEA in a sense.

Here's a PV - check out the timing periods - you will see in April it is 22.42% UPRO, 77.58% TMF.
Last edited by tomphilly on Mon Aug 31, 2020 3:15 pm, edited 1 time in total.
User avatar
Uncorrelated
Posts: 1061
Joined: Sun Oct 13, 2019 3:16 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Uncorrelated »

Steve Reading wrote: Mon Aug 31, 2020 9:53 am
Uncorrelated wrote: Mon Aug 31, 2020 1:52 am VIX is a measure of risk-neutral volatility (that means that it assumes options are priced by risk-neutral actors), but option traders are not risk neutral, this causes VIX to consistently under-estimate market volatility. Based on this explanation you might think that scaling VIX by some constant factor fixes this issue, but it turns out this doesn't work since the risk aversion of option traders changes quite substantially over time.
Gotcha. So you tried multiplying the VIX by some constant factor greater than 1.0 and used that as the future volatility estimate, and the results were inferior as well?
Exactly. I used a linear model (pred_vol = a + b * vix) to predict future volatility, and that results in worse out-of-sample utility. I also tried scaling the predicted volatility by various amounts, but a scaling factor of 1 gave the best results indicating that my linear model was already the best possible. I should point out that using VIX actually results in fairly decent in-sample performance, but there is a big drop in performance between in-sample and out-of-sample.

I also tried using VIX as a feature in other models, but that consistently resulted in worse out-of-sample results. This indicates the information present in VIX isn't all that good.

With out-of-sample I mean that I used k-fold cross validation to split the data in k different parts. Then all k parts except 1 are used to determine the parameters, the performance is assessed on the last part. The process is then repeated k times. This method prevents some (but not all!) forms of overfitting.
langlands wrote: Mon Aug 31, 2020 11:45 am
Uncorrelated wrote: Mon Aug 31, 2020 1:52 am
If you are comparing constant percent allocation vs. target volatility or target variance, I think it's very important that you do all comparisons with daily rebalancing (if you did already, then great). Recall the numerous posts from people very happy about the fortuitous timing of their quarterly rebalancing this year along with a smaller number of other people who were less happy with their unfortunate timing. There's a lot of variance to the quarterly rebalancing back tests that I hope you took into account.

Notice as well that the quarterly rebalanced constant percent allocation that HEDGEFUNDIE uses has a very important ad-hoc volatility adjustment built into it. When volatility is decreasing, UPRO tends to be rising and when volatility is increasing, UPRO tends to be falling. If you are in between rebalancing periods, your UPRO/TMF portfolio is automatically following a volatility adjustment. In other words during the market crash in March if you aren't rebalancing UPRO, you are automatically following a target volatility type strategy since you are allowing your URPO allocation to decrease drastically (via market price) while volatility is high.

Edit: Target variance isn't really the right term (obviously targeting a constant volatility or variance is the same thing). By target variance, I just mean scaling by 1/vol^2 since target volatility is in essence scaling by 1/vol.
That was my method, indeed. I used daily rebalancing, k-fold cross validation for in-sample/out-of-sample splits and calculated the expected utility similar to the manipulation proof measure of performance you see around sometimes. I also assessed the performance with various holding periods, but that didn't change the conclusions.

With my best model, I also attempted to create an actual trading strategy. But this turned out to be extremely difficult: daily rebalancing results in the highest utility before transaction costs, but this causes excessive turnover and will never survive transaction costs. With longer rebalance intervals the expected utility is less impressive but the transaction costs are also lower. I tried to use a multi-period optimization algorithm from a paper but had poor results. My best model for balancing rebalance interval and transaction costs uses machine learning (reinforcement learning) to determine the right rebalance intervals. Combining all these idea's results in around 20% higher certainty equivalent return out-of-sample, about the same you can expect from a simple value tilt.

My claim is that if I have to pull out all these tricks to get a half-decent out-of-sample performance, then the average user certainly has no chance to beat the market with market timing strategies.

I don't believe HEDGEFUNDIE' rebalancing idea's have any merit. Pick your target allocation, rebalance when the current allocation deviates too far from the target allocation. No need to perform any weird tricks. If you believe weird tricks improve the expected utility, the rebalancing algorithm is not the right place to implement them.
langlands wrote: Mon Aug 31, 2020 1:55 pm
Uncorrelated, is your interpretation of target volatility mine and tomphilly's or portfolio visualizer's? Maybe we're all talking past each other and don't even know it.
I'm aware of multiple variants of 1/σ:

1. Keep the stock volatility constant, allocate the rest to the risk-free asset. This is the variant I tested and you see most often in papers.
2. Keep the stock volatility constant, allocate the rest to TMF.
3. Keep the overall portfolio volatility constant by scaling the entire portfolio allocation. Allocate the rest to the risk-free asset.
4. Adjust the expected volatility of stocks and bonds independently, then run a mean-variance analysis and select the portfolio with the target variance.

I think variant 1 is best suited to test the overall theory as it allows you to isolate the volatility clustering effects in the selected asset class. Most users seem to use variant 2 or 3. Variant 2 can result in an asset allocation of 20/80 which is... really dumb and definitely does not succeed at keeping the volatility constant. Variant 3 appears to assume volatility clustering works similar in stocks and bonds (Which may be true, but I haven't seen any evidence). Variant 4 is one that I have not seen used anywhere, but appears to have the strongest theoretical basis (as as far as 1/σ strategies go, this one really isn't that bad. A long time ago before I discovered utility functions, I thought this was the best possible approach). I believe PV can simulate strategy 1, 2 and 3 by selecting the appropriate option as the out-of-market asset, but not 4.
User avatar
Steve Reading
Posts: 2460
Joined: Fri Nov 16, 2018 10:20 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Steve Reading »

Uncorrelated wrote: Mon Aug 31, 2020 3:15 pm
Steve Reading wrote: Mon Aug 31, 2020 9:53 am
Uncorrelated wrote: Mon Aug 31, 2020 1:52 am VIX is a measure of risk-neutral volatility (that means that it assumes options are priced by risk-neutral actors), but option traders are not risk neutral, this causes VIX to consistently under-estimate market volatility. Based on this explanation you might think that scaling VIX by some constant factor fixes this issue, but it turns out this doesn't work since the risk aversion of option traders changes quite substantially over time.
Gotcha. So you tried multiplying the VIX by some constant factor greater than 1.0 and used that as the future volatility estimate, and the results were inferior as well?
Exactly. I used a linear model (pred_vol = a + b * vix) to predict future volatility, and that results in worse out-of-sample utility. I also tried scaling the predicted volatility by various amounts, but a scaling factor of 1 gave the best results indicating that my linear model was already the best possible.
Ok but just to clarify, you only used b>1.0 above? And when you scaled pred_vol, was it by multiplying it by a number greater than 1.0?
"... so high a present discounted value of wealth, it is only prudent for him to put more into common stocks compared to his present tangible wealth, borrowing if necessary" - Paul Samuelson
000
Posts: 2842
Joined: Thu Jul 23, 2020 12:04 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by 000 »

willthrill81 wrote: Sun Aug 30, 2020 11:08 pm Best not to start such a strategy without being prepared to endure the wild ride.

The point of TMF is not its returns, especially since its expense ratio is 1.05%, which will greatly erode the pitiful current yield from LTT. Rather, the point is to provide a counter-balancing effect for 3x leveraged stocks (i.e. UPRO).
I think the concern is that as interest rates become lower, both long term bonds and stocks become increasingly interest rate sensitive and thus the counter-balancing effect is reduced. Similarly, a US credit risk scare could hurt both components.
tomphilly
Posts: 59
Joined: Wed Aug 05, 2020 11:29 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by tomphilly »

Uncorrelated wrote: Mon Aug 31, 2020 3:15 pm I'm aware of multiple variants of 1/σ:

1. Keep the stock volatility constant, allocate the rest to the risk-free asset. This is the variant I tested and you see most often in papers.
2. Keep the stock volatility constant, allocate the rest to TMF.
3. Keep the overall portfolio volatility constant by scaling the entire portfolio allocation. Allocate the rest to the risk-free asset.
4. Adjust the expected volatility of stocks and bonds independently, then run a mean-variance analysis and select the portfolio with the target variance.

I think variant 1 is best suited to test the overall theory as it allows you to isolate the volatility clustering effects in the selected asset class. Most users seem to use variant 2 or 3. Variant 2 can result in an asset allocation of 20/80 which is... really dumb and definitely does not succeed at keeping the volatility constant. Variant 3 appears to assume volatility clustering works similar in stocks and bonds (Which may be true, but I haven't seen any evidence). Variant 4 is one that I have not seen used anywhere, but appears to have the strongest theoretical basis (as as far as 1/σ strategies go, this one really isn't that bad. A long time ago before I discovered utility functions, I thought this was the best possible approach). I believe PV can simulate strategy 1, 2 and 3 by selecting the appropriate option as the out-of-market asset, but not 4.
You're right, following the target ratio all the way to 20/80 UPRO/TMF is dumb - the return in April was 10%, versus 22% for HFEA. I don't see a way to cap this in PV's target volatility models, but in practice I guess I wouldn't exceed 45% TMF, even if the signal is telling me to go further.
isubrama
Posts: 29
Joined: Sun Apr 12, 2020 12:27 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by isubrama »

willthrill81 wrote: Mon Aug 31, 2020 10:18 am
tomphilly wrote: Sun Aug 30, 2020 11:39 pm
willthrill81 wrote: Sun Aug 30, 2020 11:08 pm The point of TMF is not its returns, especially since its expense ratio is 1.05%, which will greatly erode the pitiful current yield from LTT. Rather, the point is to provide a counter-balancing effect for 3x leveraged stocks (i.e. UPRO).
Are you still employing 80/20 UPRO/TMF at 25% target volatility? I decided to switch to a variation of this last week and bought the PV subscription for the TV signal.
Yes, although I'm slowly starting to incorporate other 3x leveraged stock ETFs than just UPRO, selecting the one with the highest returns over the 1 month look back period. Right now, I'm in FNGU, a 3x LCG ETF, which is up nearly 23% in August alone.
tomphilly wrote: Mon Aug 31, 2020 8:35 am
Sushmit wrote: Mon Aug 31, 2020 7:02 am
tomphilly wrote: Sun Aug 30, 2020 11:39 pm
willthrill81 wrote: Sun Aug 30, 2020 11:08 pm The point of TMF is not its returns, especially since its expense ratio is 1.05%, which will greatly erode the pitiful current yield from LTT. Rather, the point is to provide a counter-balancing effect for 3x leveraged stocks (i.e. UPRO).
Are you still employing 80/20 UPRO/TMF at 25% target volatility? I decided to switch to a variation of this last week and bought the PV subscription for the TV signal.
What allocation is that ? Is there something I can read up on
The strategy is similar to HFEA, except instead of quarterly re-balancing, you balance the ratio according to a volatility signal. In low volatility periods, your allocation will be your default allocation, e.g 80/20 UPRO/TMF, whilst in volatile periods, your ratio could go from anywhere between the default to the complete opposite - e.g 20/80 UPRO/TMF. The ratio to use is reported in portfolio visualizer when you have the paid subscription - alternatively if you're a smart math/finance person you could probably determine the signal yourself and save yourself the yearly subscription - regrettably I'm not smart enough.

Right now the volatility signal is risk on - meaning it's 80/20, pedal-to-the-metal time. During the calamitous period of April, for example, it was 22/78. Note that you could experience much greater drawdowns with this than with classic HFEA, so be aware of the risk before implementing it.
The drawdowns could be smaller though due to the ability to adjust the allocation according to volatility.
FNGU is ETN, not ETF.
tomphilly
Posts: 59
Joined: Wed Aug 05, 2020 11:29 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by tomphilly »

willthrill81 wrote: Mon Aug 31, 2020 10:18 am
tomphilly wrote: Sun Aug 30, 2020 11:39 pm Are you still employing 80/20 UPRO/TMF at 25% target volatility? I decided to switch to a variation of this last week and bought the PV subscription for the TV signal.
Yes, although I'm slowly starting to incorporate other 3x leveraged stock ETFs than just UPRO, selecting the one with the highest returns over the 1 month look back period. Right now, I'm in FNGU, a 3x LCG ETF, which is up nearly 23% in August alone.
Interesting, thanks. I have been toiling with the idea of adding lagging sector 3xETFs like FAS. Another curious ticker is RSP, though you'd have buy calls to get leverage. RSP is an equal weight S&P500 - my concern is SPY has lost a lot of its diversity, and RSP might fare better in an impending tech correction versus SPY.

Does anyone use options in this strategy as an alternative to 3x ETFs? I remember a few pages back there was a suggestion of buying puts on inverse leveraged ETFs like SQQQ and TMV to eliminate volatility decay. I ran a small experiment doing this. What I still don't quite understand is how to pick a strike price and expiration that would roughly match a 3x ETF in volatility, or when to close the positions and reopen them at new expirations.
Investing Lawyer
Posts: 9
Joined: Wed Apr 08, 2020 9:35 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Investing Lawyer »

Uncorrelated wrote: Mon Aug 31, 2020 3:15 pm
Steve Reading wrote: Mon Aug 31, 2020 9:53 am
Uncorrelated wrote: Mon Aug 31, 2020 1:52 am VIX is a measure of risk-neutral volatility (that means that it assumes options are priced by risk-neutral actors), but option traders are not risk neutral, this causes VIX to consistently under-estimate market volatility. Based on this explanation you might think that scaling VIX by some constant factor fixes this issue, but it turns out this doesn't work since the risk aversion of option traders changes quite substantially over time.
Gotcha. So you tried multiplying the VIX by some constant factor greater than 1.0 and used that as the future volatility estimate, and the results were inferior as well?
Exactly. I used a linear model (pred_vol = a + b * vix) to predict future volatility, and that results in worse out-of-sample utility. I also tried scaling the predicted volatility by various amounts, but a scaling factor of 1 gave the best results indicating that my linear model was already the best possible. I should point out that using VIX actually results in fairly decent in-sample performance, but there is a big drop in performance between in-sample and out-of-sample.

I also tried using VIX as a feature in other models, but that consistently resulted in worse out-of-sample results. This indicates the information present in VIX isn't all that good.

With out-of-sample I mean that I used k-fold cross validation to split the data in k different parts. Then all k parts except 1 are used to determine the parameters, the performance is assessed on the last part. The process is then repeated k times. This method prevents some (but not all!) forms of overfitting.
langlands wrote: Mon Aug 31, 2020 11:45 am
Uncorrelated wrote: Mon Aug 31, 2020 1:52 am
If you are comparing constant percent allocation vs. target volatility or target variance, I think it's very important that you do all comparisons with daily rebalancing (if you did already, then great). Recall the numerous posts from people very happy about the fortuitous timing of their quarterly rebalancing this year along with a smaller number of other people who were less happy with their unfortunate timing. There's a lot of variance to the quarterly rebalancing back tests that I hope you took into account.

Notice as well that the quarterly rebalanced constant percent allocation that HEDGEFUNDIE uses has a very important ad-hoc volatility adjustment built into it. When volatility is decreasing, UPRO tends to be rising and when volatility is increasing, UPRO tends to be falling. If you are in between rebalancing periods, your UPRO/TMF portfolio is automatically following a volatility adjustment. In other words during the market crash in March if you aren't rebalancing UPRO, you are automatically following a target volatility type strategy since you are allowing your URPO allocation to decrease drastically (via market price) while volatility is high.

Edit: Target variance isn't really the right term (obviously targeting a constant volatility or variance is the same thing). By target variance, I just mean scaling by 1/vol^2 since target volatility is in essence scaling by 1/vol.
That was my method, indeed. I used daily rebalancing, k-fold cross validation for in-sample/out-of-sample splits and calculated the expected utility similar to the manipulation proof measure of performance you see around sometimes. I also assessed the performance with various holding periods, but that didn't change the conclusions.

With my best model, I also attempted to create an actual trading strategy. But this turned out to be extremely difficult: daily rebalancing results in the highest utility before transaction costs, but this causes excessive turnover and will never survive transaction costs. With longer rebalance intervals the expected utility is less impressive but the transaction costs are also lower. I tried to use a multi-period optimization algorithm from a paper but had poor results. My best model for balancing rebalance interval and transaction costs uses machine learning (reinforcement learning) to determine the right rebalance intervals. Combining all these idea's results in around 20% higher certainty equivalent return out-of-sample, about the same you can expect from a simple value tilt.

My claim is that if I have to pull out all these tricks to get a half-decent out-of-sample performance, then the average user certainly has no chance to beat the market with market timing strategies.

I don't believe HEDGEFUNDIE' rebalancing idea's have any merit. Pick your target allocation, rebalance when the current allocation deviates too far from the target allocation. No need to perform any weird tricks. If you believe weird tricks improve the expected utility, the rebalancing algorithm is not the right place to implement them.
langlands wrote: Mon Aug 31, 2020 1:55 pm
Uncorrelated, is your interpretation of target volatility mine and tomphilly's or portfolio visualizer's? Maybe we're all talking past each other and don't even know it.
I'm aware of multiple variants of 1/σ:

1. Keep the stock volatility constant, allocate the rest to the risk-free asset. This is the variant I tested and you see most often in papers.
2. Keep the stock volatility constant, allocate the rest to TMF.
3. Keep the overall portfolio volatility constant by scaling the entire portfolio allocation. Allocate the rest to the risk-free asset.
4. Adjust the expected volatility of stocks and bonds independently, then run a mean-variance analysis and select the portfolio with the target variance.

I think variant 1 is best suited to test the overall theory as it allows you to isolate the volatility clustering effects in the selected asset class. Most users seem to use variant 2 or 3. Variant 2 can result in an asset allocation of 20/80 which is... really dumb and definitely does not succeed at keeping the volatility constant. Variant 3 appears to assume volatility clustering works similar in stocks and bonds (Which may be true, but I haven't seen any evidence). Variant 4 is one that I have not seen used anywhere, but appears to have the strongest theoretical basis (as as far as 1/σ strategies go, this one really isn't that bad. A long time ago before I discovered utility functions, I thought this was the best possible approach). I believe PV can simulate strategy 1, 2 and 3 by selecting the appropriate option as the out-of-market asset, but not 4.
If you want to predict volatility I would suggest using a GARCH model on sp500 returns. Personally, I've only seen GARCH modeling done on daily returns, but they've looked really good (haven't got around to it myself using longer dated sequences). Unfortunately with the typical GARCH(1,1) model you only get a prediction one period ahead. A linear model of VIX to predict the future volatility is not going to work. Least squares estimates are likely not consistent due to the absurd autocorrelation of the VIX and therefore, the residuals as well. If I wanted to model the VIX itself, I would consider only using an ARIMA model. I imagine if you run a least squares with the past value to predict the future and an ARIMA(1,0,1) you would get different estimates of the coefficients for the parameter for the lagged value. (Vix would likely need to be differenced before modeling so both models would be bad)
Post Reply