Trade copier was added to Smart Forex Tester Suite recently – inspired by positive results of our previous tests.
Trade copier is intended to use in live trading the winning trading signals achieved by our automated trader running on the TrueFX data feed.
We continued testing on GBPUSD, since this currency pair delivered good results in our previous testing. The reference test was using the inbuilt TrueFX data feed and inbuilt Autotrader strategy (version1).
We started both tests during London trading session on October 28, 2020.
We see that the result is overall positive, and in line with previous testing – though the win ratio was slightly lower that what we used to achieve.
The copier was placing the trades to the IG demo feed via the MT connector. We see the same number of trades but the results are worse.
Partly this was caused by requotes; some of which required up to 5 retries to complete the trade.
Below is a log snippet showing how the strategy attempted to close one BUY order. You can see it took nearly 1.5 s to close which resulted in the eventual closing price slippage of 3.7 pips.
2020.10.28 02:18:10 : elapsed 0 ms; order type: BUY; attempting to close at 1.29837
2020.10.28 02:18:10 : elapsed 390 ms; order closing failed for price: 1.29837 Error code: 138
2020.10.28 02:18:10 : elapsed 687 ms; order closing failed for price: 1.29822 Error code: 138
2020.10.28 02:18:11 : elapsed 1078 ms; order closing failed for price: 1.29811 Error code: 138
2020.10.28 02:18:11 : elapsed 1406 ms; order closing retries: 3
2020.10.28 02:18:11 : order close price: 1.298
On the one hand, the delay and pip loss from the requotes were not too big in absolute numbers. On the other hand, these delays confused the algorithm and spoiled next trade entries.
We repeated the copier test the next day. Our reference test was running non-stop. We started the Copier at about the same time into the London session as the day before.
Below is the reference test result. You can see that the performance was not ideal but still from the beginning of the Copier test it was profitable (~15 pips) and the total win ratio improved to 51%.
However, the Copier was less successful. And we again had a lot of retries – maximum number of retries was 8!
Bottom line. Our tests proved the copier idea is viable. We improved our MT connector and made it robust to trade errors. However, we will need to find a better feed to test the Copier on. Another lesson learned is that we need to make the automated trading strategy itself less dependent on such errors.