Stock Selection - Baseline Returns by Holding Period

In order to fully evaluate a trading strategy, I need to develop a baseline with which I can compare it to and ensure the returns exceed the market returns.  The baseline performance is calculated for the period over which I will be backtesting my strategy results (backtest period).  The baseline period will begin on September 15, 1998 and end on September 15, 2009.  I will be using StockFinder 5.0 to perform the calculations.  I used all US common stocks for my research sample.

I wanted to incorporate some basic liquidity conditions into my tests to remove the sporadic behavior of illiquid and low priced stocks.  Each stock considered in the analysis met the following conditions:

• Closing price was greater than $1 (C > 1)
• Daily volume was greater than 100,000 shares (V > 1000)

I created a combination condition using the two criteria above.  I then averaged the return of all the stocks that passed the combination condition over several time periods: 1 day, 5 days, 10 days, 15 days, and 20 days.  I performed these calculations for each day during the backtest period generating a sample of 2768 returns for each time period.  Below is a chart listing the summary statistics for each return period sample:

Across the top of the chart are the length of time for each return (ex. average return of a stock over 20 days).  The upper section of the chart shows the percentile calculation for each sample.  The distributions are very similar for all return periods.  As expected, magnitude of the return increases with the length of the period.

The middle section shows the average of the samples falling in between the percentile limits (ex. average of all stocks between the 25th and 75th percentiles).  The averages help determine the robustness of the returns.  If all of the averages are similar, the returns are robust and not limited to a small portion of the sample.  Each sample is fairly consistent over each percentile range.

The bottom portion of the chart shows the win/loss characteristics of each sample.  The winning percentage increases the longer the stock is held.

Below is a graph that shows the average and median return for each period over the entire sample.  For example, the average return for a stock (meeting the liquidity criteria) over 20 days is 0.87%, and the median is 1.11%.

This graph can be used to help determine the period of time to hold a trade in order to overcome the effects of slippage and commissions.  You would need to trade with a significant amount of equity (without using leverage) to make a strategy profitable when holding a trade for only 1 day because the average return is only 0.07% (median is 0.11%).  Slippage and commission fees are much easier to overcome when a stock is held for 20 days because the average return is 0.87% (median is 1.11%), which means you would need much less equity to achieve the edge.  When developing a trading strategy, the returns can be compared to the values in the graph to determine if there is really trading edge or if you are just capturing the movement of the market.

Conclusions of the baseline analysis:

• Winning percentage increases as the length of the trade increases.
• Average and median return increases as the length of the trade increases.
• Less equity is required per trade to overcome slippage and commissions as the trade length increases.