Experiment 3

Experiment 3 (4/25 – 5/12)

The results of Experiment 2 indicated that the MACD 1/5 EA was better than the rest, and the results were statistically significant.  In addition, we saw some indications that the shorter timeframes were more profitable, and that the MACD and DMA versions were better than the MAFR version, although not all of these results were statistically significant.

Given the above, Experiment 3 focused on the MACD and DMA versions, both at the short timeframes, and in different operational modes as shown in the figure.

Where:

Mode A is the original operational mode previously studied.  The entry trigger is PN=0.2, and statistics are segregated by session and day of the week in addition to indicator states
Mode B is the same as A, but there is no day of week segregation.  Only the session data is kept separate
Mode C uses no segregation between day of week or session.  The only data bins are those defined by the indicator states.

Results

The following data was obtained during the duration of the experiment:

Cumulative pips during the test period

Average pips per trade

This qualitative metric is calculated as the ratio of the mean gain/loss to the p-value of that configuration as compared with the aggregate of the rest of the configurations.

Analysis

We tested four hypothesis as follow:

Ho1: The difference between any two configurations is due to randomness  (α=0.05)

We fail to reject the null hypothesis at the selected probability criterion.

Ho2: The difference between any one configuration and the aggregate of the rest is due to randomness  (α=0.05)

We fail to reject the null hypothesis at the selected probability criterion.

Ho3:  The difference between any two Versions (aggregate Timings) is due to randomness  (α=0.05)

We fail to reject the null hypothesis at the selected probability criterion.

Ho4: The difference between the two Timings (aggregate Versions) is due to randomness  (α=0.05)

We fail to reject the null hypothesis at the selected probability criterion.

Conclusions

None of the results were statistically significant at the selected probability criterion.  The Q-metric results point to the MACD-B variant as coming out on top in this experiment, but again, results are not statistically significant.

The next phase of the experiment will repeat this setup, however, we’ll enable multiple entries and each entry’s lot size will be determined such that 0.05% of the total account value is at risk.  This will allow the EAs to efficiently use the  variable TP and SL bands.  For instance, for larger PN ratios, the SL will be closer to the entry price, so in order to risk the same amount, a larget lot size is needed.  As a result, the EA bets larget lots as the confidence of a move increases.