In this article we review a trading strategy developed with DLPAL and specifically the out-of-sample and forward performance using code generated for the Quantopian platform. The strategy outperforms the benchmark and year-to-date return is higher by about a factor of 2.
The DIAT4S3 strategy was developed in April 2013 with in-sample DIA data from inception (01/28/1998) to 12/31/2009. The out-of-sample where the strategy was evaluated covered the period from 01/04/2010 to 03/28/2013.
Although DLPAL has significantly many more functions and capabilities than the earlier version of the software that was used to develop the DIAT4S3 strategy, it still offers the old search capability for compatibility purposes.
Below is the search workspace for this strategy.
Below are the old results from the in-sample.
The above 10 strategies, 8 long and 2 short, were used in a combined strategy that was tested out-of-sample and validated across a portfolio of 12 ETFs.
On August 18 of this year, the DIAT4S3 strategy joined the Quantopian contest.
Below is the performance of the strategy using DLPAL generated code for the Quantopian platform. Backtest is highly accurate since exits are determined based on minute data.
It may be seen that the DIAT4S3 strategy has outperformed the benchmark (SPY) by about 37% after the end of the in-sample period. Sharpe is 1.29 and beta is reasonably low at 0.25.
Below is year-to-date performance:
The performance of the DIAT4S3 strategy is about double that of the benchmark. Beta at 0.15 is low and Sharpe is 2.33.
Note that the DLPAL search algo is improved and the search results in 11 strategies, 8 long and 3 short. Out-of-sample performance is improved, as well as, year-do-date performance shown below:
There is significant outperformance of the benchmark buy and hold total return. Beta is close to 0 and Sharpe is 2.73.
The above results can be checked with anyone with a copy of DLPAL because there is no stochasticity in the output of the program.
If you have any questions or comments, happy to connect on Twitter: @priceactionlab