@FxMasterGuru I may be asking the wrong question, But this test you made in a Demo account right?! Is there any difference from a real account for Strategy Tester? Is the TickStory the best data? I never heard about.. .thanks for sharing this with us
ironpips posted: @FxMasterGuru I may be asking the wrong question, But this test you made in a Demo account right?! Is there any difference from a real account for Strategy Tester? Is the TickStory the best data? I never heard about.. .thanks for sharing this with us
Well, TickStory is the only reliable, relatively low cost and easy to use backtester software I know. It uses Dukascopy's free price data.
Whether you do a TickStory backtest on a real or a demo account it does not matter as TickStory uses Duscascopy's M1 historical data and makes ''intelligent guesses'' about likely intra-candle price movement. So for tick-scalper EAs it is less accurate, but for strategies with an average win and loss > 10 pips, it is reasonably accurate, but still LESS than a DEMO forward test.
Of course, by recording on MT4 each and every tick the market produces and running an EA backtest on these self-recorded data would be more accurate than Dusascopy's M1 data, but running a backtest on EACH AND EVERY TICK the market had produces would extend the running time of a backtest by 100x or 1000x which would unrealistic. Running the Inertia Trader backtest on 2016's M1 data took 14+ hours. With real recorded tick data it would have taken 14 months at least, but even to be able to do that I should have recorded each and every tick on my MT4 the EURUSD market had produced since January of 2016... Which I have not done anyway.
AND FINALLY: I do not take neither DEMO nor BACKTEST results seriously. They are only good to provide an APPROXIMATE IDEA if an EA might be good or if it was outright bad. Since I do not trust DEMO results more than backtest results, I went ahead and started a REAL LIVE ACCOUNT FORWARD TEST, i.e. the VERY BEST TEST one can do on an EA.
Here is a 6-year backtest on the M15 version. In comparison with M5:
1. Profit Factor is significantly lower (1.68). 2. Expected payoff is slightly lower (10,250). 3. Relative drawdown is slightly worse (-49.6%). 4. Total net profit is only marginally better (8,775,160).
In summary: M5 has performed clearly better in this 6-year backtest, so I see no reason to run M15 alone or with M5.
fxMasterGuru, thanks for these backtests. very interesting that the M5 seems to be performing better than M15. Would it be possible for Vendor to tick the box that allows comments to show in history (or magic numbers), I believe the timeframes are contained here. Then we can do a customer analysis and decide based on at least the live data for this year, what the return would be for just M5, M15 and both together, and also the DD for each scenario. Cheers