Optimisation! Share your experiences, please. - page 2

 
AndyGri:
Is it possible to adjust the parameters for so many trades and can the market change so immediately? What is the best way to do this?

The easiest check for "is there no curve fitting?" is to change one or two parameters by 5-10% and get poor test results. Have you checked?
 
Does anyone know why optimisation with the genetic algorithm is limited to 10,500 passes?
 
Apparently this is quite enough for the genetic algorithm:)
I have repeatedly noticed that there are parameters (profitable), which the genetic algorithm does not find. but it may be my fault, though...
 
Apparently this is quite enough for a genetic algorithm:)

Maybe so, but frontal optimization produces tens or hundreds of times more runs.
 
Some time ago in one thread on this forum an interesting idea about elimination of fitting was mentioned, namely: - Take the best parameters after optimization on one time interval and apply them to another time interval (not involved in the optimization)
 
 
AndyGri:
sashken:
57% modelling quality would not be enough:)
90% is just right, that's probably what's causing all the discrepancies.

Well, that's the maximum the tester gives. In principle, it's not the quality. In this particular EA, pending orders on extrema are working and closing by trailing stops or by orders. Thus, the quality does not have anything to do with the idea.

But if you don't have any ideas or principles, the 90% of it is the reason.
57% - it's not the tester that doesn't give us more data but you don't give it the quality historical data, the tester just signals that your data is rubbish and therefore the test results cannot be trusted.
Go here - https://www.mql5.com/ru/articles/mt4/tester - and read everything. There will be considerably fewer questions.
 
Exactly, the modelling quality can be raised to 99%, I saw it myself in a tester report(not my own though)
 
AndyGri:
sashken:
Why don't you brag? :) And post the code of the Expert Advisor?
Or at least tester reports. Maybe the picture will clear up:)

Strategy Tester Report
chas_GBP_TP_TSnorm_SLlowHigh

Symbol GBPUSD (Great Britain Pound vs US Dollar)
Period 1 Hour (H1) 2006.01.01 23:00 - 2007.03.21 00:00 (2006.01.01 - 2007.03.21)
Model All ticks (based on all smallest available periods with fractal interpolation of each tick)
Parameters porog=1; MAmor=2; MAtrend=24; risk=0.1; CandleBar=0.55; TP=120; TS=70; SL=54; TimeCH=10; VolP=1.5; t=0; porogSL=10; candle=11; candleEX=17; MAporog=17; DellOrd=23;
Bars in history 8494 Modelled ticks 1576481 Simulation quality 57.67%
Initial deposit 1000.00
Net profit 3745.64 Total profit 7681.59 Total loss -3935.95
Profitability 1.95 Expected payoff 25.14
Absolute drawdown 0.00 Maximum drawdown 384.68 (12.31%) Relative drawdown 12.31% (384.68)
Total trades 149 Short positions (% win) 60 (65.00%) Long positions (% win) 89 (75.28%)
Profitable trades (% of all) 106 (71.14%) Loss trades (% of all) 43 (28.86%)
Largest profitable trade 120.00 losing deal -263.31
Average profitable deal 72.47 losing transaction -91.53
Maximum number continuous wins (profit) 8 (547.11) Continuous losses (loss) 3 (-249.45)
Maximum Continuous Profit (number of wins) 635.18 (7) Continuous loss (number of losses) -263.31 (1)
Average continuous winnings 3 Continuous loss 1

It's all clear. Testing and optimisation on the clock, and real trading 2 times in 3 days, i.e. complete mismatch between the timeframe of the test and the real one. You should move the tests to daily chart, it's closer to the truth. Or, if it is inconvenient to output at the end of the American timeframe, then run the EA at a certain hour, but add it to the code:
...
//the time of advisor starting
extern int hour = 12;

...

int start() {
if (hour != TimeHour(Time[0])) return(0);
// EA code
...
}

Please note that the time you set in the hour variable corresponds to the time of your brokerage company, not the time on your computer. Therefore, it may be shifted.
 
AndyGri:
The sample is large - 160-200 market entries, and 2 weeks is not a long time for big changes. Is it possible to adjust the parameters for so many trades and can the market change so quickly? Can you tell me what to do?
160-200 trades - are you kidding? If you have 5-7 optimizable parameters, thousands of deals can easily fit the curve (And you have 16 external parameters for optimization!!! - Tens of thousands of trades will be fit to ANY curve if you have enough time and a more powerful computer! :o) Look for example here:
https://www.mql5.com/ru/forum/50458 solandr 18.03.06 20:11
This childhood I was engaged in a year ago. That system then successfully flushed in the real world (I reported about it in the same thread later, somewhere in May 2006). The idea of the system was conditionally taken from the ceiling (catching peaks in the noise). So you are not the first to step on the rake of fitting.
In my first post of this thread solandr 23.03.2007 15:43 I gave a link to Bakeev's suggestion about comparing results of optimization on martingale to understand how your algorithm can be fitted to a random curve (how viable your algorithm idea is). And then it's up to you to decide whether you want to do it or not.