EA Optimization, Backtesting & Analyzing

Thank you for your feedback!
It really helps a lot!

Do you or anyone else know any books to read or do you have any fields that you could point to that would be helpful in algo trading? Webpages, courses, or specific areas for me or anyone else reading here?

Would LOVE to have a mentor as I know how valuable it is! I had a mentor for 6 months when I took my trading course In London many years ago but they didn’t believe in EA’S so I didn’t look for them until finding this forum.
If you come up with something or find some sort of mentoring program that you think fit, pls make an announcement here or so everyone can get use of it or if you want to get in contact you can find me at (dohax (a) protonmail (dot) com
(I hope it’s ok if I share my email​ address here, Yannick)

I’ve only tested Ichu EA here on Quivofx witch is very advanced, can someone recommend any other EA to start learning optimization?
Was thinking of MA or BB, for lesser parameters?
Think I will go thru BB and analyze your and Walters conversations and check the SET files to understand the thinking behind it.
I will not create a high risk account (yet) that Walter has in that post more study the progress.

Overall I’m extremely grateful and thankful for all the insights I’ve gotten from reading and searching for “clues” on how I need to be thinking regarding to algo trading. I went thru some posts several times to see if I missed something :joy: If I understand how you guys think and do stuff I know what to search for and then start the learning process.

I’m doing a backtest now back to 2016.01 and would like to organize the data in excel, thinking of keeping the data of one TF and Pair so I don’t need to optimize again and again, someone got any ideas? This is mostly to get thru the pairs and to find the most profitable one(s) , then later fine tune it with more filters like ADX and day/time, short/long, and so on.

My first test period was EA’s on the 1M and 5M chart just to get a feel for how things work. I know there is a lot of noise on those TF’s. I only back tested 3 months and thought that I could backtest it monthly, was up 20% during three weeks and then something happened when i fiddled whit broker time settings and it went down 19% during three days before I paused the EA.
I was risking 2% on every trade, then added funds to the account and forgot to change the % Risk per trade and the losses compounded.
I’ve downloaded fresh tickdata thru Tickstory and will start over again with higher TF’s and longer backtesting to find a better “balance”

Lesson learnt 🙂

If i trade on lower TF’s and much higher probability setups. I will probably do some basic TA and draw trend lines, support and res levels and let it trade in between those areas, for semi automatic trading but that is another project when I add more computing power to my home. Will add 2 setups soon.

Downloaded Khan Academy (Android) for math studys and a friend will help me and point to were I need to focus and what could help whit my Algo trading 🙂

Again thank you for the encouragement, I will continue to update this post and anyone, feel free to add to it if you got some valuable info to share so others can adopt.

Hi Nikolaj, Thank you for your post! We have a general optimization thread here Feel free to add to it! I’m new to EA’s but have some trading experiance.

Hi all,
I would like to share with you my opinion on how to get profitable settings for ICHI (with all due respect to the really experienced guys here).
Believe it or not it is actually fairly easy:) – you just need to follow a proper protocol. So here is what I would suggest:
1. Choose a pair and time frame

Do you check timframes under 15M or only 1H and 4H? Anyone here going for Daily or Weekly?

2. Get decent data – the best way to go is 99% with variable spread. However, an easy (and still reliable way) is to open a DEMO account with Alpari and use their data – I`ve checked on several occasions 99% vs 90% Alpari over large number of trades and results were very similar (DD differed within 1%)

Mine is from Dukaskopy thru Tickstory (fullversion) I only for 99.9% date when I test my results in Quant Analizer check Monte Carlo and to look for Day/time filters

3. Turn true “Trade on new bar” – this way you can do your optimization on “Open prices only” – much faster and give similar results to tick data.
4. Optimize over a LARGE PERIOD (minimum 10 years – I prefer 15 years) – this way you know that your settings are NOT CURVE FITTED!!! That is very very important!

I usually go for 1000 trades history 3-5 years and then I check with out-data I’ll test the setting on non optizied period a other 3-5 period.

5. Choose settings (from the optimization results) with minimum DD and AT LEAST 200-250 trades – the more the better. If you have less trades for given settings be very careful with your Money management – the less trades you have for given set file the higher chance of the results being RANDOM!!! (Read “The Drunkard’s Walk: How Randomness Rules Our Lives” if you want to see the point here)

Thx, I usually go for Profit and PF but will try to focus more on DD because I hate losing, lol. Slow and steady usually wins the race

5. “Leave” time period for virtual forward test – so if you optimize lets say from Jan. 2000 to Jan 2015 and get your sweet set file with DD 4% and nice Profit factor – do a virtual forward test from Jan 2015 to Jan 2018 to see if your set file is still profitable. This way you have much higher chance to find reliable profi set!!!

Thx, I usullay optimize the closer period and do the backtest on the older period, maybe it’s better to turn it around as markets changes, thx

6. Optimize only for 1 setting at a time with fixed lot and equal TP and SL. Probably only exception is the very first thing that you optimize for – “Type of strategy”, Entry filter and Exit strategy – you optimize those together. TURN OFF all other filters (ADX) and SL functions (ATR, Candle trail, ect). Once you get your profi set start optimizing 1 setting at a time – you can start with TP and SL. After that ADX, then trailing SL functions, time filters, ect but only one thing at a time. DO NOT USE MANY FILTERS – YOUR RISK OF GETTING CURVE FIITED settings INCREASES the more filters you use!!! FILTER CAN NOT TURN UNPROFITABLE STRATEGY INTO A PROFITABLE ONE – it can only improve DD and PF of a PROFITABLE STRATEGY!!!

Thx, will try this.

7. Diversify your portfolio – get 10-20 profi sets on different pairs ideally with low correlation and dont risk more then 1 % of your account per trade (even less!). Dont be greedy – youll regret it when get into DD. IF you loose 50% of your account youll need 100% profit after that only to get back to the starting point!

Yes, the % Is very important, I had profitable setting for weeks then the losses compounded and weeks winning were lost in days (I added funds to the account and forgot to change % risk per trade

8. Get several accounts with different brokers – you minimize your risk if something happens to a broker and also you can compare the execution b/n the different brokers (you run the same EA with same set files on all of them) – this way is much easier to notice dodgy practices (SL hunting, ect).

Do you use ECN accounts? I’ve been checking Pepperstone and FXopen, they seems legit. any tips?

I hope that will help all those new traders that are struggling with finding profitable settings. Many of the above ideas can be used for optimization of any other EA:)!

Best regards

Nikolay[/quote]