On October 13, 2008, Ajay Jani spoke at the New York Region Meeting.
Ajay focuses on emerging markets, and has done so for more than twenty years. In this talk, he explained that he thinks it’s important to have a system with well-designed rules and a position sizing algorithm. The concepts apply to any tradable security in any market.
His research began with the thought that for capital constrained traders (which is everyone except a central government banker), it would be useful to have systematic tools that help “time the model.” This would help the trader understand when they are in synch with the market and should use a larger position size.
His first attempt to do this was to place a moving average on the equity curve of the system. The equity curve is a measure of how much money an investor would have in their account by following all system buy and sell signals. Unfortunately, a simple 200-day moving average on the equity curve actually hurt performance with lower profits and higher drawdowns. The drawdown is a measure of how much equity a trader loses in their account from the peak.
The next step in his search for a tool was a more sensitive moving average. The 50-day moving average offered visually appealing results. However, the returns were even lower and drawdowns were even higher. Seeing the results, Ajay wondered if this was a productive research area. He went back to asking what his objective was, and he realized he wanted buy the low, or add money right before the system turns around and the drawdown comes to an end.
He reasoned that buying lows in a system means buying after performance has been relatively poor. To test the idea, he set the buy signal to be those times when the 60-day rate of change (ROC) of the equity was down by at least 5%. The results were promising, however they suffered from a look ahead bias. A dynamic approach to the problem would address this concern.
To make the 60-day ROC a dynamic tool, Ajay applied percentile ranks to the data. He collected the first year’s worth of data and began the ranking process. Each day offered a new data point, and he expanded his rankings to include all the available data. This allows the system to learn what good and bad performance is from the experience of the actual results.
When trailing sixty returns are in the lowest quintile, expectancy is probably pretty high and that is a good time to add money to the system. Specifically, he adds money when performance is in the lowest 10% of performance ranks. On average, that happens twice a year, and this process adds significant value to the system performance.
He noted that this idea is similar to card counting in Vegas. Blackjack players look for an edge, and popular wisdom is that an edge exists when the deck still has a lot of tens left to be dealt. They increase their bet size then, even though there is no guarantee that they’ll win. Although there is no guarantee, the blackjack player at least has improved their expectancy, which is the mathematical chance of profits.
In this video, Ajay offers the unique insight that using the equity curve to evaluate system performance can add value to a trading system. The lookback period of the ROC can be varied depending upon factors such as risk tolerance and transaction costs. Other indicators like the MACD or Bollinger Bands can also be applied to the equity curve. Of course, this entire process begins with the assumption that you have developed a system with a positive expectancy.