Targeted Volatility Analysis
One of the most overlooked areas in trading is the focus on keeping a constant bet size in order to ensure consistent risk-adjusted returns. Most traders fail to explicitly adjust for market volatility, which can have a catastrophic impact when you have the misfortune of betting the wrong way when the market is swinging violently. The use of leverage makes this even more dangerous. However this factor affects the buy and hold investor as well. Portfolio management is predicated on creating an asset allocation for investors that have a certain risk/return profile based on expected returns and volatility. Most passive portfolios keep constant exposure to market conditions, leaving them effectively overleveraged when volatility spikes in years like 2008.
So how do portfolio managers in the field deal with such issues? A friend of this blog, Jerry Wagner has a great deal of experience making these types of decisions as the president of Flexible Plan Investments. CSS was recently in attendance at the NAAIM Uncommon Knowledge 2010 Conference in Orlando where Jerry presented an interesting technique for position sizing/hedging using current and historic volatility. This is the first in a two part follow-up to the Secret Sauce Trading Seminar at in which one of our CSS researchers—David Abrams—also presented a paper. So what is Jerry’s secret sauce to address the issue of maintaining constant volatility for a portfolio? The equation is really very simple yet mathematically elegant:
You can incorporate this into your own trading with or without leverage. Here’s how to do it with leverage:
If (HR < 1) Position Size = 1 – HR
If (HR >1 and HR<2) Position Size= HR
If (HR> 2) Position Size = 2
Position size of 1 means full position, 2 means 100% leverage. So a hedge ratio (HR>1 and HR<2) means leverage of (HR-1). This has the benefit of continuously adjusting position size based on the ratio of the historic volatility to the current volatility. The numerator can be adjusted by a volatility multiplier to “target” a specified level of volatility and make your system more conservative or aggressive.
Let’s look at an example.
CAGR | St Dev (daily) | Sharpe | DVR | MaxDD | |
BUY HOLD | 7.29% | 1.25% | 36.66% | 0.01% | -55.19% |
TVA (no leverage) | 6.55% | 0.84% | 49.14% | 0.02% | -34.09% |
TVA (200% leverage) | 7.59% | 0.79% | 60.35% | 0.05% | -30.58% |
In the graph below you can see how a simple long only strategy using TVA and leverage to control position size can reduce the max drawdown in half and almost double the Sharpe ratio while also slightly increasing the CAGR.
The resulting equity curves are much more stable than a buy and hold allocation despite using leverage explicitly. In future posts we will look at how to incorporate volatility and other metrics into position sizing. These will be incorporated into our portfolio management tools for systems. Good work Jerry!
And what is HR (hedge ratio)? Is it total long portfolio divided by amount hedged?
Damian
The Hedge Ratio is the percentage of the portfolio being held in the investment used to hedge the position. We use cash, short index futures, inverse funds, bond funds or alternatives in our various strategies to fill this percentage. Thanks for asking. Jerry
David, did Jerry publish a paper at the NAAIM conference? If so what was the title? I thought I had downloaded all of them but cannot find a reference to his name.
Thanks
John
John/David,
Are these papers available for the public to download for people who did not attend?
Thanks,
Curtis
Curtis, here is a link
http://www.naaim.org/
I cannot find the link to this years papers anymore. If you Email them though they are very responsive.
All 17 of the papers from 2010 National Association of Active Investment Managers (NAAIM) competition for advances in active investment management are available free at http://naaim.org/wagneraward.aspx . It was quite an international competition this year with entries from the UK, Canada, the US and the winner from New Zealand.
The presentation David speaks of was from a PowerPoint I did at the annual meeting in Orlando. That PowerPoint plus all the others for the conference and past conferences are only provided to members. Membership information is available at http://naaim.org/memberclasses.aspx . Regional meetings are free and dates are regularly posted on the website or in the free newsletter. There is a charge (less for members) for the annual meetings and the CRAM Sessions (usually one on trading systems and one on marketing). The networking and exchange of information at the meetings is terrific!
As always David and staff did an amazing job synthesizing my much longer presentation. Thanks again. Jerry
thanks jerry for beating me to the punch, and thanks to all of you for the questions.
john, i believe you had a question before about the livermore and we will try to show a chart. curtis, thanks for the astute comments on the previous post set.
best
dv
Yes thanks David.
Nice analysis, have you used Statpro’s IT portfolio management software
Would it make sense to use ratio of sharpe’s (or DVR) instead of a ratio of standard deviations to create the hedge ratio ? For some strategies which do well in times of higher volatility in terms of risk-to-return, weighting the risk-to-return (DVR or sharpe) instead of the standard deviation would be better, in my humble opinion. I have been using that and have backtested that in a couple of my strategies and it has been pretty good.
Also for a multi-instrument trading, i presume the standard deviation of the equity curve would be chosen to calculate the hedge ratio, according to this blog post – is that correct ?
Thoughts ? Comments ?
Would it make sense to use ratio of sharpe’s (or DVR) instead of a ratio of standard deviations to create the hedge ratio ? For some strategies which do well in times of higher volatility in terms of risk-to-return, weighting the risk-to-return (DVR or sharpe) instead of the standard deviation would be better, in my humble opinion. I have been using that and have backtested that in a couple of my strategies and it has been pretty good.
Also for a multi-instrument trading, i presume the standard deviation of the equity curve would be chosen to calculate the hedge ratio, according to this blog post – is that correct ?
Thoughts ? Comments ?
David, I believe that there is a typo in the rules declaration i.e. “If (HR >1 and HR<2) Position Size= HR" should read "If Hedge Ratio 1 then Leverage = HR-1” (that’s lifted straight from Jerry’s paper). Having said that I must admit it seems to work a treat! Thanks for sharing.
David, I have read the paper and couldn’t eally understand exactly what the formula was for the short side. Do you have any idea?
Thanks
John
hi john, as a simplified means of applying the TVA, you can simply take the ratio of say 10-day volatility/ the 60 or 252 day average of 10-day volatility. lets call this the hedge ratio. your position size would then be 1/hedge ratio (max 200%). this would apply to longs and shorts. if you went short you would simply take the position size stated but obviously it would be -ve with a max of -200%.
best
dv
David, I applied this to the back-test of a system I have been working on for a while (polite cough) and using the SD of the unadjusted equity curve as the basis for the volatility calc. My CAGR went from 26.54 to 26.92 and the av. annualized max DD went from 9.42% to 9.07%. All positives but nothing earth shattering. However the SD of the annual returns went from 0.07 to 0.04 which is obviously a big change. Much smoother results. If I apply an additional factor (as per the paper) I can get much higher returns than the original strategy without the TVA adjustment.
Thanks
John