I'm a bit behind in providing an update on my EA portfolio demo test. Below is the summary table providing an update to include last week's results (i.e. it excludes results for this week):
A few comments:
The demo testing exercise has proven very valuable in highlighting some issues with the EA's that are not evident in backtesting. The differences experienced are due to the fact that the backtesting environment is based on fixed historical data, whereas in demo testing (and presumably in live trading) the EA only executes instructions if there is a tick in the relevant market at the time that the EA checks he market.
So I've had 2 main issues:
1. In some EA's I check for some data at 5 mins past midday and 5 mins past midnight. During the 1st hour or so of server time the amount of ticks occurring can be sporadic, so sometimes the instructions to obtain data can't be executed - and hence later on in the day the EA is using old data that is incorrect.
2. Similarly, on some EA's I am using the Daily chart and the EA executes instructions at 0:00:00 server time when markets are temporarily closed for a minute or so (apparently this applies to IC Markets, not necessarily other providers) - and hence the EA cannot executre any instruction.
So for the RMD EA (this is the one which I posted using the HA model based on Rick's suggestion) in the table above which is applied to 15 markets, 10 use the daily timeframe so the results shown are from 5 markets only. I fixed this last week and so last week's results include results from the 10 daily markets (which had a negative impact of -100 pips last week - great timing), but running the EA through the backtester over the previous weeks shows the EA has missed trades netting 183 pips)
Similarly for the DB EA which is applied to 25 markets, 14 are Daily and so the results shown until last week are from 11 markets only. Running the EA through the backtester for the above period shows that 780 pips were missed because trades on the daily markets were not executed.
So my conclusion is while it is good policy (imo) to demo test any strategy to be traded, when using EA's its essential to pick up potential issues that may arise in live market action that are not seen in backtesting.
In terms of the results:
The BRI EA which looked like a star at the beginning has not performed well over the last 3 weeks, and is now in negative return territory, and a larger drawdown than seen in backtesting. I have to analyse this one in detail to understand why its not performing to expectation, but at this stage I will probably not start trading it live when I start the others, based on the above results, until I understand what the issue is (if any) or get more demo data.
As for the other 5, they are all performing well in terms of generating a return (despite the bugs mentioned above) and what I like is that it is not just 1 EA generating all the return, they are all contributing a double digit % contribution to the total pips earned.
One experiment I have decided to try on the DB EA is a money management tip I picked up from Tradeview - I noticed this EA is generally running with a profit balance on open trades, but as the trades gradually close the profit booked is less than the open profits would indicate. In this scenario, an idea worth trying is if the open profit balance equals a pre-defined amount (for example 2.5% of the equity balance or $250 on a $10K account balance, which I am trying this week) then the EA closes all open trades and books that profit, and then starts opening trades again. Using this approach has netted me just under 900 pips only from this EA so far this week (better than any of the previous weeks shown above so far).
