The Gateway to Algorithmic and Automated Trading

The accidental quant

Published in Automated Trader Magazine Issue 32 Q1 2014

Ernie Chan says it was not by design but by chance that he ended up in the market. As accidents go, it turned out to be a happy one. A theoretical physicist by training, he's also that rarest of traders who likes to talk openly about his ideas rather than keeping them secret. Ernie tells Automated Trader Editor Adam Cox about his research into mean reversion and momentum strategies and his journey from science to finance.

Adam: You've done a lot of work on the issue of whether to use mean reversion or trend-based strategies. Let's start with what your current research is telling you.

Ernie: Yes. In my first book I've been focusing on mean reversion strategies mostly and that actually has been my focus in my own trading as well as the strategies we run in our previous fund. And the reason is that I was sort of obsessed with Sharpe ratio. I was obsessed with generating steady profits and consistency in those profits. Now, that has a drawback. Mean reversion strategies do produce consistent profits and do produce high Sharpe ratios, but with a hidden downside. And the hidden downside is the presence of the fat tails and the outliers. Many people compare running mean reversion strategies with picking up pennies in front of a steamroller. Most of the time you get those pennies but once in a while you get run over. And that has been my experience. Other people might compare that with selling options. You gain premium most of the days but once in a while you get wipeouts.

Unless of course you impose various risk management techniques such as stop losses. But stop loss itself is problematic for mean reversion strategies because a) if you are trading stocks there's no stopping your loss during the overnight gap. You can only really effectively use stop loss for forex and futures, and even those markets you cannot apply stop loss during the weekends. So stop loss has somewhat limited benefits, unless you're trading intraday. Another problem with stop loss is that typically with a mean reversion strategy a stop loss contradicts your trading signal. So if the price goes lower, you're supposed to buy more instead of stopping out. So in principle a stop loss for a mean reversion strategy should only be placed somewhere such that in the back-test it was never breached. Once a stop loss level is breached in a mean reversion strategy, you should give up that strategy, you should never trade it again. That is a very unique stop loss that we can impose.

Cumulative Returns for a Momentum Strategy

Cumulative Returns for a Momentum Strategy

This figure shows the typical cumulative returns profile for a momentum strategy. The strategy performed superbly during the financial crisis of 2008-9, reaching a high watermark around April of 2009. But when the stock market started to recover, this strategy's performance also started a relentless decline. This is a phenomenon called "Momentum Crashes", and many commodity funds suffered a similar fate as this strategy.

Adam: In other words, the simple rule is that whatever your back test shows for a successful mean reversion strategy, the stop loss needs to be outside of that range so that you don't cancel your own strategy prematurely. You need to be prepared to lose that much.

Ernie: That is correct, yes. If you place the stop loss outside of the lowest price in a mean reversion back test, so that it will never be breached, and if in live trading it is breached, that practically invalidates your back test immediately. One should give up. Your back test says it should never be breached, so how come in live trading it's breached?

Adam: Much then depends on the extent of the back testing, so in your experience in these strategies, how big might a loss potentially be?

Ernie: Well, it very much depends on your own risk tolerance and your investors' risk tolerance. That's a pretty arbitrary number. Some people can tolerate 10% maximum drawdown, so the stop loss should be less than or equal to 10%. Other people can tolerate 50% maximum drawdown. Okay, that's great then we should place the stop loss at 50%. But again one should very much bear in mind that the stop loss is only effective only during market hours, so if one is holding a position overnight or over a weekend, all bets are off.

The remainder of this article is only available to Paid Subscribers

Click here to purchase a subscription to Automated Trader

  • Copyright © Automated Trader Ltd 2017 - Strategies | Compliance | Technology

click here to return to the top of the page