As I always say, finding trading ideas is not the difficult part - there are literally hundreds of these ideas out in the public sphere on various blogs and books (for example, check out my new book “Algorithmic Trading: Winning Strategies and Their Rationale”). The difficult part is to improve and refine these ideas. This step is crucial because a published strategy is often well on its way to obsolescence. Its profitability is often on its way down due to increased competition or market structure changes. That doesn’t mean we should just abandon them though - a small tweak will frequently turn a strategy around and inject new life into it.
Let me give just one example. Some years ago, a finance professor1 from MIT and his collaborator published an extraordinarily simple strategy for trading profitably a large number of stocks in a “market-neutral” way. That is, this strategy is supposed to make money whether we are in a bull or bear market. The reason it can do this is because it shorts almost as many stocks as it buys, and in a bull market, the stocks it shorts go up a bit less than the stocks it buys, and vice versa in a bear market. The strategy in a nutshell: calculate the one-day close-to-close return of every stock in an index, and subtract the index return from its one-day return. Buy this stock if this relative return is less than zero, and short it if it is greater than zero, with the dollar amount (negatively) proportional to the relative return. The proportionality constant depends on the desired gross (long+short) market value of the resulting portfolio, and doesn’t affect the returns of this strategy. There is just one minor detail to note: the index return is calculated based on an equal-weighted average of all the component stocks’ returns, not based on the conventional market-capitalization-weighted average.
The authors of this strategy reported an astonishing Sharpe ratio of 4.5. Alarm bells should have gone off: why would someone publish such an excellent strategy instead of keeping it secret and trading it themselves? It turns out that there are many caveats with this specific backtest: no transaction costs were included, and the backtest was performed on a small-cap index which exacerbates the potential impact of transaction costs. When I backtested this strategy on the S&P 500 stock universe, with transaction costs included, the Sharpe ratio turns negative.
For many traders, this more careful backtest sounds the death knell of a strategy, and they will just drop it in disgust. But that’s not us. A simple tweak of the strategy by using close-to-open instead of close-to-close returns to compute the relative returns (and therefore entering a position at the open) will restore the Sharpe ratio to its former glory, even in the face of transaction costs2. Other tweaks are also possible that will reduce drawdown in certain market environments, making this quite a practical strategy with enduring profitability after all.
This example suggests that finding a trading idea and backtesting it properly is just the first step in developing an algorithmic trading system. More hard work is just ahead of us!
Ernest Chan is a hedge fund manager and the author of "Algorithmic Trading: Winning Strategies and Their Rationale" (Wiley, 2013) and “Quantitative Trading: How to Build Your Own Algorithmic Trading Business" (Wiley, 2009). Find out more about him at www.epchan.com.
1Khandani, Amir, and Andrew Lo. “What Happened to the Quants in August 2007?” Preprint, 2007. Available at http://web.mit.edu/alo/www/Papers/august07.pdf.
2 In truth, there are other problems associated with a backtest using open prices. I explore that in more details in my workshop on Mean Reversion Strategies.
The following article is from one of our external contributors. It does not represent the opinion of Benzinga and has not been edited.