...
About the cost of trading comment i dont think that makes much difference. From my limited experiments with reversing EAs i only test on narrow spreads (1 point EURUSD) for example and if you use anything over 20 point SL/TP etc the cost is tiny. Admittedly if you were trading live you would need to account for slippage etc so it would become an issue but backtests still show reversing not working.
I disagree about the transaction cost being tiny.
If you plot the trading cost (spread + commission) in terms of percentage of account; together with the equity curve; you will clearly see that the trading cost is the culprit.
attached an example of a random EURUSD strategy on hourly timeframe for whole year of 2011
- Buy @ 33% probability
- Sell @ 33% probability
- Flat @ 33% probability
- Stop loss at 50 pips
- Take profit at 50 pips
- Execution @ market (meaning we cross the spread)
- Spread used: 0.5 pips for EURUSD (this is the spread I use when trading with Interactive Brokers)
- Commission used: 0.2 pips per USD 100K (this is equivalent to 0.2 pips if the base currency is USD; otherwise you have to exchange the rates)
Chart Legend:
- Chart is in percentage of equity
- Green line: cumulative equity (in percentage of account)
- thin dashed red line: cumulative commission (in percentage of account)
- thin solid red line: cumulative spread (in percentage of account)
- thick solid red line: cumulative total cost (commission + spread) (in percentage of account)
Two charts attached:
chart 1: the random strategy
chart 2: inverse of the random strategy
You can clearly see from chart 1 that the loss in equity curve is majorly attributed to transaction cost
The inverse of the strategy; is much worse; and a consistent negative curve