Backtest reliability
I have developed an EA
If I try it in August, September 2019 I get excellent results.
If I try it on December 2019, January 2020 I lose.
I can't change this!
My EA works on ticks and the number of them is similar between August and January.
One difference is that August works up to H1, while January drops to M1.
Are there any differences in the test between August and January? Can I trust what I see?
Thank you
- Use custom input to optimizer
- 10points 3.mq4
- Tick Data vs 1 Min Candles (MT4)
Greg Pawlak:
Hi, first of all, the last 6 months are too short data to check the profitability of your EA, What time interval does your EA use? Secondly, the differences in tests can be the result of different market volatility and your EA apparently can't handle it. Regards Greg
Hi, first of all, the last 6 months are too short data to check the profitability of your EA, What time interval does your EA use? Secondly, the differences in tests can be the result of different market volatility and your EA apparently can't handle it. Regards Greg
Thanks Greg.
My EA works on individual ticks and on some key levels.
Nothing should change between H1 and M5 or M1.
If I go back to 1 year, it doesn't seem to work at all, I don't understand.
I don't consider volatility, I don't know how.
I'm sure there are differences between August and January, but I don't understand which ones.
Thank you very much
ok so you use the tick chart, what is the quality of the historical test? what data do you use to perform this test?
Greg Pawlak:
ok so you use the tick chart, what is the quality of the historical test? what data do you use to perform this test?
Forgive me, I'm a beginner, how do I choose which data to use? Which is correct to use?
ok so you use the tick chart, what is the quality of the historical test? what data do you use to perform this test?
Your EA works well on on a specific period and on a specific asset, it's simple, your EA will never work well in the future.
You need to always backtest at least 1-2 years and on multiple assets in order to include different market scenarios if you want to have a good point of view of your EA.
Any EA can be profitable for a specific time range in the past but what about the future?
If
you are beginner, why you take for the most hard kind strategies on forex?. Start trade on higher time frame like H1
Thanks for your valuable advice, but I still haven't figured out if the August test shows the reality of that month, or there are differences.
The need for a universal assisting tool for the strategy testers is apparent , a common ground ,a standard if you will where a newcomer is guided and informed about what they see . Even better , a one button solution "Test" ,no parameters ,forced real ticks , slippage simulation ,delays simulation ,spread widening , news simulation etc
Offcourse even then there will be pitfalls as Marco will soon point out.
