Systematic Trading research and development, with a flavour of Trend Following
Au.Tra.Sy blog – Automated trading System header image 2

Drawdown Reduction

March 30th, 2011 · 13 Comments · Money Management

I was recently reading a blog post discussing “trading the equity curve” of a system. This usually entails adapting trading size based on whether the equity curve of the system is below or above its equity curve.

A similar concept is described in The Way of the Turtle, by ex-Turtle Curtis Faith:

The Turtles were instructed to decrease the size of the notional account by 20 percent each time we went down 10 percent of the original account.

The idea seems to make sense in a way: the further the notional account size is reduced during a drawdown, the lower the maximum drawdown amount figure should be. Similar to a racing driver “hitting the brakes” as and when their vehicle starts going off-course, to avoid going “into the ditch”.

In periods of prolonged losing periods, the result should be a lesser impact on the equity curve, and a lower drawdown figure.

Testing The Concept

Trading Blox has this functionality built-in, so it is easy to test. You can set parameters for this Drawdown Reduction technique: the Threshold dictates at which point the notional account size is reduced (10% drawdown in the example above) and the Amount dictates by how much the size is reduced (20% in the example above). The reduction is cumulative, meaning that every time a new 10% decrease is observed, a further reduction of trading size by 20% is applied.

I ran a 20-day Donchian breakout system with “classic” volatility-adjusted position sizing (risk per trade = 0.75% of equity = 2 ATR) as a starting point. The results obtained were as follows:

Performance Stats
CAGR
52.77%
Max DD
40.30%
MAR 1.31
Sharpe 1.22

 
The next run applied the Drawdown Reduction logic to the exact same system. As expected, the Max Drawdown figure does decrease by a fair amount, however note that the return also decreases by a greater amount, actually hurting the MAR ratio. The Sharpe ratio does not improve either:

Performance Stats
CAGR
43.44%
Max DD
35.80%
MAR 1.21
Sharpe 1.19

 
In hindsight, it simply looks that reducing the overall leverage of the original system might achieve the same results (reduction in both drawdown and return). Here are the results of the first system with reduced leverage (0.64%) so that the Max Drawdown amount matches that of the second system:

Performance Stats
CAGR
45.67%
Max DD
35.80%
MAR 1.28
Sharpe 1.23

 
Sharpe and MAR are closer to the first run results, with the CAGR being higher than in the second run, implying that a simple leverage reduction could be a better option.

Post Update: Of course, this is a single one-off test, from which you can hardly draw any conclusions, so do not go and dismiss it solely based on this post/test. A proper and more complete test with a large number of system variations – as kindly pointed out by Pumpernickel in the comments – would be a good start to evaluate the impact of the Drawdown Reduction technique. Something to play with in your own system development or to follow-up in a later post.

Managing the Unexpected

This comparison test benefits from hindsight. This money management technique might still bring a way to deal with extra-ordinary negative periods (where drawdowns would exceed the expected figures from the back-test). This way, the system could start trading with the “optimal” leverage derived from the back-test, with an extra safeguard (possibly with higher threshold triggers) only for cases when the system starts diverging substantially from the back-tested results (whether the system is “broken”, or experiences its worst period to date).

Another point to note is that the downside of reducing leverage during drawdowns usually increases the time required to get out of the drawdown.

Another possible use of this technique might be to use it to dynamically allocate to several systems in a suite: reduce the allocation of poorly-performing systems and shift the allocation to better performing systems (“starve the dogs and feed the stars”). This way, if a system starts becoming “broken”, its allocation is automatically decreased, which should reduce its (negative) impact on the overall suite performance.

 

Picture credits: Dirk Gently via flickr (CC)
Related Posts with Thumbnails

Tags:

13 Comments so far ↓

  • Alvantage

    In my opinion, this strategy of reducing trade size whenever equity drops will be of any use only if there is evidence that the strategy performance itself is persistent.

    On the other hand, an alternative strategy of increasing trade size whenever equity drops will be useful only if there is evidence that the strategy performance itself is anti-persistent (mean-reverting).

    Thus, we first need to establish whether the strategy itself is persistent.. or anti-persistent. :)

  • Pumpernickel

    Another blogger did a larger experiment. He tested four different trading systems with very different characteristics: some had very tight trailing stops, others none. Some had very tight initial stops (enter only when price breaks out of a very narrow or “pinched” Bollinger band etc), others had very loose initial stops. Some held trades for 150 days, others for 10 days. Some used profit-targets to “scale out” of part of the trade, others exited all-at-once.

    He put these systems into Trading Blox and used its “Drawdown Reduction Threshold” feature, as you did, to experiment with the Way Of The Turtle method of reducing per-trade risk when in a drawdown.

    This blogger stepped the parameters of each system over a reasonable range, getting about a thousand or so variants of each of the 4 trading systems. Then he plotted his results on a pair of scatter plots.

    Plot#1 was (MAR Ratio with Drawdown Reduction Disabled) versus (MAR Ratio with Drawdown Reduction Enabled). Plot #2 was (Sharpe Ratio w/DDR) versus (Sharpe Ratio without DDR).

    Some of the dots on his scatterplot did fall into the same zone as your single-system, single-set-of-parameters result. Some of the dots had (MAR-with) less than (MAR-without), and (Sharpe-with) less than (Sharpe-without). But only *some* of the dots on his scatterplot.

  • Jez Liberty

    @Alvantage: Quite true, which is why it would be nice to see how the Drawdown Reduction Technique works with various types of systems as pointed out by Pumpernickel.

    @Pumpernickel: would you have the link to that larger test please? I agree that you can hardly draw any conclusions from this single test and I would be interested to see the results of this larger test – and so would the blog readers, probably. If you could provide the link, I could update the post for the benefit of other readers – thanks.

  • Jing

    Hi, Jez
    I think the beauty of trading equity curve can not be clearly seen from history backtesting because the systems we design/test are somewhat optimized/curve fitting to history data. The beauty of trading equity curve lies in the future. When we design a system, that system must work well in history data, or else we will not use it. But the future market will always change, that system may suddenly crash (such as the original turtle, or some moving average crossover due to the popularity of that strategy or market change), if we reduce the weight of a system when it fails to work in market, it can save us a lot of money. I am not sure whether I express it clearly, when a trading strategy stops working in future (due to that system is not robust, too many people use the same strategy, or market change and that strategy not suited to current market condition), trading equity curve can dramatically reduce the loss of a system which performs good in history but suddenly stops working in future. And this benefit can’t be easily seen from history backtesting.

  • Pretorian

    Hi Jez,
    I have been tinkering with this and other ideas to control risk: limiting total risk (TB has a blox f0r that), limiting correlations, limiting instrument exposure, etc. My conclusion is the same as yours: we can achieve the same risk control effect by lowering initial bet size with the added benefit of using less parameters so the system is more robust.

    Another conclusion is that over the long term the best we can do with a Long Term Trend Following system is achieve a MAR of around 1, which is what most of the Market Wizards you follow have. This realization has helped me to stop kidding myself and concentrate in robustness and dealing with drawdowns on a mental level. By the way, two of the wizards themselves told me personally that this kind of “tricks” don’t work. This might me better explained by one of Ed Seykota’s phrases:

    “If you don’t like drawdowns you can stop trading”.

    Thanks

  • Jing

    That means if we trade equity curve (maybe reduce size when in drawdown, or trade the moving average of equity curve, when equity curve cross below its MA, we stop trading it/reduce size and resume after equity curve cross up its MA), our future maximum drawdown can be reduced. So I see trading equity curve as a way to deal with robustness. It is actually a hedge against the sudden failure of a previously excellent system (in backtesting) but may fail in future trading.

  • RiskCog

    When I read about the Turtles I thought this “drawdown reduction” technique was dopey from a system design perspective, but maybe it has value from psychological perspective for sheltering the new traders.

    Dynamically allocating to different systems takes the same care as designing each system. I can’t imagine using a rule like a step reduction in exposure to an instrument by 20% after a 10% loss inside a system.

    One approach that appeals to me is to design systems with as low of expected drawdowns as possible, this way one knows if a system is mis-behaving before it has done much damage to the account value. With this type of goal drawdown reduction isn’t really something that is bolted onto the outside of a system.

  • Pumpernickel

    The “other blogger” is Future Jez. Someday you’ll review your old threads looking for research ideas. You’ll see this set of comments, which suggest applying Drawdown Reduction to a thousand parameterizations of several highly diverse systems. You’ll perform the suggested experiments and analyze the data, looking to see whether DR is (always, sometimes, never) beneficial. You’ll analyze whether certain system characteristics (presence or absence of an initial stop, tightness of trailing stop, average trade length, average trade duration, etc) are associated with beneficial DR. You’ll tinker with different ways to present the data (“visualization”), to find the clearest way to highlight insightful patterns in your 4000 simulation results.

  • Jez Liberty

    Haha, very good Pumpernickel…
    Very true and good suggestion. I actually recall quite a few comments on the blog, suggesting interesting follow-up research (some from you actually), and will indeed be a great source of inspiration for Future Jez, like yours on this post. Thanks!

    @RiskCog: I feel that, for a large part, reducing drawdowns is linked to reducing leverage (and returns) also. Of course, you do see systems with high MARs during back-testing or live trading for a while, but I have the feeling that these systems probably suffer from over-fitting and might not be so robust in the long term (as evidenced by the MAR ratio of TF Wizards being around 1 as pointed out by Pretorian and also illustrated with this “Where have all the Sharpe ratios over 1 gone?” plot ).
    As Jing was saying (and as I was trying to hint at in “Managing the Unexpected”), a method for reducing allocation of poorly-performing systems (whether with this Drawdown Reduction Technique or something more sophisticated like Ralph Vince LSPM or your own allocation optimisation algorithm) might be a sensible way of gradually phasing them out of the overall allocation.
    It also probably helps on a mental level to deal with unexpected losses.

    Otherwise, how would you handle one of your systems starting “mis-behaving”? I suppose you could decide to suddenly stop trading them when unhappy with their performance…

  • Nitin Gupta

    It reminds of one of the strategies by emilio tomasini wherein u dont trade a trading system if the smaller moving average of equity curve goes down below the larger moving average of equity curve.

  • Pumpernickel

    You might want to caution readers that the post in its current state is woefully incomplete. A good notification would be to insert this comment you made, before the very first sentence of the blog post: “You can hardly draw any conclusions from this single test.”

  • Jez Liberty

    Well, the post was not trying to draw any definitive answer as to whether the Drawdown Reduction improve a system, but rather present this money management logic and illustrate it with a quick test.
    But this is a good point: I have added an updated “warning” in the test results discussion part.

  • Demon

    One thing this method has a positive effect on is risk of ruin, the price for that is a deterioration in risk adjusted performance as it unsures you have the smallest bet-size on at the bottom of the curve and the largest bet-size at the top. You pays your money and you takes your choice.

Leave a Comment