So following on from my previous post, this post covers the results for an EA based on the standard HA bar strategy, i.e.:
1. If the HA bar colour changes from red to green then (1) set a canbuy flag, and (b) set the buy threshold to the high of the 1st green HA bar
2. If a subsequent HA bar breaks the buy threshold before the colour changes again, buy on open of the following bar (and close any still open sell position)
3. Set a fixed takeprofit to a value X between 10 and 200 pips (value X selected based on optimisation)
4. Set a stoploss at the lowest low of the last Y bars (value Y selected based on optimisation)
5. Trail the stop to the lowest low of the last Y bars (same value Y as above) if the new value is closer than the previous value
And vice versa for sells.
After checking that the EA executed all actions correctly (which it did), I started to test each market on both the Daily and H4 timeframes (Daily first & if results not showing positive expectancy in both backtest & backapp periods, then tested H4), optimising for the 2 variables X & Y.
I found 15 markets where the EA produced a positive expectancy for both the backtest and backapp periods. 2 examples are shown below - the USDCHF on the Daily timeframe and GBPUSD on the H4 timeframe:
USDCHF Dail, backtest results:
USDCHF Daily backapp period:
GBPUSD H4 backtest period:
GBPUSD H4 backapp period:
The quality of the equity curves for each of the different markets varied (some with higher drawdowns, other with lower drawdowns, etc), and the variables X & Y varied between markets, reflecting the different levels of volatility, imo.
So having identified 15 markets showing positive results during both backtest and backapp periods, I put the results into a porfolio on EA Analyser and the combined results were as follows (take note the sizing for each market was different and was selected so the maximum single loss was < 1% of the $10,000 account starting equity = $100, so typically the sizing was between 0.04 - 0.08 lots):
In % terms the total return during the combined 3.5 year backapp-backtest periods was 140%, with a max drawdown in % terms of 7.5% - not bad for an EA that took a couple of days to write.
So while the above EA looks pretty good - and I have put it into a demo account today to demo test for 1 month - I will now go back to the original HA / Smoothed Line EA and see if I can improve on the results from the basic EA - and post an update when I have one (maybe by this weekend).
