In effort to build up motivation and consistency I decided to create a Journal Thread to monitor my progress on my EA. I am a recently graduated Econometrician and I’ve been trading successfully for little over a year now. I have my real life trading strategies, but considering my background I was always fascinated by developing a small EA, that could provide some diversification and work as a support during potential downswings.
Since I am all about the data my eyes turned to scalping. Why? It’s the easiest to test, it generates a lot of data. I do believe that designing a Scalping EA is more difficult than other types (you have to account for the current volume, time of the day, volatility, trend power, existance of a rally etc.), but I firmly believe that once those are accounted for the ability to find out the quality of the algorithm is far superior.
For now I will focus purely on EURUSD, its an obvious choice, it ensures the lowest transaction costs. The average cost of trading it on the broker I use is 0.7 pip in prime hours (accounting both spread and commissions).
Initially I will keep the TP=SL as it simplifies the analysis and makes working with the data easier. (I might consider switching to TP=SL+2COST, just to make 50% winrate breakeven, but time will tell).
Alrights so thats it for the premise.
At this point the entry strategies that I am testing are:
First 1H data is used to identify the favorable trading hour.
Second 1M data is used to find the scalp entry.
I will keep you posted boys. If it works I see no problem sharing it, so I would appreciate any feedback.
PS. I prohibit Martin_K to write in this topic and would love if that could be respected! It is reasonable since its my journal right?
Hope you don’t mine me popping up bro. Looking forward to your journal and wishing you the best.
With all due respect for your system, have you ever considered dropping down the next level to tick charts for entry’s instead of the 1min. Personally, I live, love and breath the 70 tick chart.
Little table of winrates for referance. Assumption = avg cost 0,7 pip. First collums shows TP/SL in pips, second breakeven winrates, third winrates with avg pip profit of 1.
Hi mate
Firstly, you can’t actually prohibit me from doing anything. i think a better word may have been appropriate, like
"I would request that Martin does not… etc etc"
Secondly Yes, ok, i can respect that, it’s a reasonable Request
I’ll leave you guys to it
Be Cool
Okay, so I have decided to step it up a notch and I am going to base my SL/TP on the slope of the MA(100-200) in 1M chart. With a condition that if the function gives SL/TP of under 4 pips, trade does not occur. I think it could allow for more universal trading hours as it would ensure being profitable under different volatility conditions.
Eitherway I’ve decided to base the first model on an individually optimised MA Cross on 1M, with 3 additional conditions.
Measurement of local maximum difference of two MAs (right before the cross).
Measurement weather the slope of MA200 is not flat.
Detection system for rallies (based on increase in volume/value swings).
First would ensure optimal SL/TP placement and proper scalp trade
Second would work as a bottom level - ‘do not enter’.
Third would put no TP and Trailing Stop Loss, with closure when the volume goes down.
Thats the bottom layer for now, as in that one I see no place for tick data yet.
Then on the top level what I will do is use the software to input all possible combinations of MT4 indicators on 1H chart, with all combinations of variables to find the ones best correlated with the system. Gonna take a couple of hours of optimisation, but should show some good results…
I ran the basic model without any implementations what so ever, just as a guideline to where I am at. So just a simple optimisation of MA Cross, without the mentioned constraint. Here are the stats at this point.
I will call it alpha 0.1
Data from 1-1-2017 till now
Trades: 3756
Won: 2209 (58,81%)
Lost: 1547 (41,19%)
Spread: 1 (bigger than the actual)
Relative drawdown: 28%
Consecutive wins: 13
Consecutive loses: 9
It’s a good start, I will start implementing the constraints. Starting with 1H trend constraint => The trades are going to most likely half in count, but the drawdown is going to decrease.
BTW. Simulation was ran on trades of 2 lots with 10K account.
