I hope everyone can objectively evaluate my MT4 backtest.(EURUSD 15 years backtest)

To add comments, please log in or register
laozhang713
15
laozhang713  
The first time I learned to write EA, I wrote a good EA, but I was ridiculed in various QQ groups, so I thought about asking you to be objectively evaluated.

1. Suppose I didn't cheat.

2. Suppose I have no past optimization.


Based on the above two conditions (because there is no clearest in my heart -, -), Trouble God gave the evaluation back report.


Backtesting time: 2003-2019 (I broke the position in 2002, I don't know if it is because of the problem of the first year of the euro, but to be honest, it exploded.)

Initial capital: 10,000 usd

Backtest use time period: M1 (purely for higher precision)

Spread: 10 (standard households, no handling fee, if you use a 0-point ECN account, you will make more money in good years, but the dull year-end fee will cause the net value of the principal to continue to retrace, so the standard household is finally selected)

EA type: band trend (99% order closed on the same day, so basically do not consider overnight fees, interest, inventory, etc.)

Risk management: 1, 2, 3, 4, 5, 6 (in fact, only manages the size of the position)


The net value curve of risk 1 is relatively smooth, but only 10%+ return in 10 years, so it was PASS.

Risk 6 is top, because the risk 7 back test is froze in 15 years.

Size per lot: Calculated according to the risk value and the net value of the account, not a fixed lot.

My own summary: I feel that the risk is controllable in the whole, and there are a few years of plain and faint, and there is no loss. In 2006-07, when I lost the least, I only had more than 2,000, and then I got up again. I only used it for 15 years. Risk 3 has lost 50%+, but in 2009, 10 years was a profit of 450%+, and in 18 years it was nearly 200%. The only 15 years is a heart, but I am more pleased that it is at least from 03 to 19 years. I didn't explode, I think this is more important. With this, I feel that I can sleep at night.


[Risk coefficient 3] 2003 to 2019 (actually in September 2018, the number of hands is capped, my program does not limit the number of hands, can not open a single, and the latter is also profitable)

https://img.alicdn.com/imgextra/i3/4020594036/O1CN01y3afUw1fgX53wT9kJ_!!4020594036.jpg


Because the number of LOTS is calculated on the basis of the net value ratio, the later curve is too radical, which has caused the judgment of the previous period. Therefore, I changed it to 1 fixed lots and made another evaluation (it is purely to see the previous net value curve)

https://img.alicdn.com/imgextra/i2/4020594036/O1CN019Jm7OA1fgX51RT3av_!!4020594036.png


Part of the order:

https://img.alicdn.com/imgextra/i3/4020594036/O1CN01tXmZJW1fgX57rITrS_!!4020594036.png

https://img.alicdn.com/imgextra/i1/4020594036/O1CN01f8Qz1f1fgX53wSp04_!!4020594036.png

https://img.alicdn.com/imgextra/i4/4020594036/O1CN01lSV4q81fgX567FX5p_!!4020594036.png


I hope that I can objectively evaluate EA from the perspective of reporting.

iRick
233
iRick  
How does you back test look with a spread of 30 or 50, slippage can easily be 20/30/40 on closing a trade
Chris70
405
Chris70  

You are aware that all the linked screenshots are in chinese and not everybody can understand chinese?

However, 69% drawdown under optimized conditions of a backtest is a 100% guaranteed total loss in real life.

Greg Pawlak
192
Greg Pawlak  
Hi, your quality test is n/a, so results backtest can be fortuity. Most important you have to get quality test between 90-99%, than you can accept results as good. Regards Greg
To add comments, please log in or register