Discovering Patterns that Have Been Profitable Across All or Most of Dow-30 Stocks

Developing the capability of discovering price patterns that have been profitable in an in-sample of a group of securities, like for example the Dow-30 stocks, was an ambitious project I started last year in the form of an addition to the set of functions of Price Action Lab software. I have already presented some results for ETFs. In this post I present results for all Dow-30 stocks.

I call common patterns those that form in a security that belongs to a pre-defined group and are profitable across all or most of the securities in the group. This special category of patterns may be identified using a unique feature of the search function of Price Action Lab. Recently, I added an option that allows specifying the minimum percentage of the securities in the group that must show profitable results for each common pattern (currently in beta testing). For example, if that option is set to 80%, each common pattern must be profitable at least accross 24 stocks out of the 30 that compose the DJIA index.

Search Example

This example is for long-only stock trades and for common patterns with 5% profit target and stop-loss in the Dow-30 stocks with win rate of at least 66% and profit factor equal to at least 1.50. Each common pattern must be profitable in at least 75% of the Dow-30 stocks and the minimum profit factor for that is set at 1.01. The minimum win rate for short patterns is set at 99 to exclude them. Here is the workspace setup:

The in-sample period is from 01/04/1999 to 12/31/2009. Thirty search lines were created, one for each Dow-30 stock and the “Find Common Patterns Only” option was checked.  This test was for a regular (not-extended) search.

In-sample results

The in-sample results above (sorted by ticker name) indicate that the program found 15 patterns, out of which 5 were distinct for 12 stocks. Thus, the program in regular mode was able to identify 5 patterns that have been profitable in at least 75% of the Dow-30 stocks. This can be confirmed by clicking on a pattern line and using the “Back-Test Portfolio” tool. These are the results for the first pattern (highlighted) on the list:

It may be seen that this pattern was profitable in 23 out of the 30 stocks (76.67% win rate). Also, the number of trades for each stock are shown along with other statistics. Thus, although the trade sample size in the original results is small for each pattern, this is not relevant because what matters is the portfolio backtest sample size as shown below:

The portfolio backtest above, determined using the “Portfolio Backtest” tool, shows that the portfolio profit factor for each pattern in the results is well above 1.00, indicating high profitability, as expected. The performance of this system, when applied to the 12 distinct stocks, can be determined using the “Test Patterns” tool, after selecting the relevant directory with the in-sample data files:

The results are shown below:

By trading ALL signals in the in-sample for the 12 stocks, the result was a profit factor of 2.56 and a win rate equal to 72.42%.

Out-of-sample test

The results for the out-of-sample test, from 01/04/2010 to 09/20/2012, can be obtained using again the “Test Patterns” tool, after selecting the directory with the out-of-sample historical data files:

The results indicate a winning, high profitability system in the out-of-sample:

In the out-of-sample, the 5 distinct patterns generated 53 trades for the 12 distinct stocks at a 64.04% win rate with a profit factor equal to 1.96.

How to trade the system

The easiest way is by adding the system to Price Action Lab System Tracking and updating the data files on a daily basis:

One could also generate code for the 5 patterns and then use it to implement the system in some other platform. Then, the program will indicate when new signals arrive for next open and for which security.

I will be forward testing this system for about 6 months and then I will possibly update it by running the workspace again.

Disclosure: no relevant position at the time of this post.


This entry was posted in Price Action Strategies, Strategy Synthesis, Trading Strategies and tagged , . Bookmark the permalink.