control points

 
does higher time frame has more control points than lower time frame in optimization?
 
What "control points" ?
 
Alain Verleyen:
What "control points" ?
In optimization, the strategy tester
 
Dua Yong Rew: does higher time frame has more control points than lower time frame in optimization? - In optimization, the strategy tester

I don't know exactly what you are asking, but my advice and suggestion, is don't base your results on "control" points. Use tick testing even if they are the "emulated" kind. It gives much more realistic results than control points.

If however, you want to use it because your EA is "very slow" to test under "tick" testing/optimising, then it is a wake-up call to you that your code is highly inefficient and requires to be rewritten properly (for example, reading indicator values once per close of a bar and not on every tick).

 
Dua Yong Rew:
In optimization, the strategy tester

Ok. From what is said in this article, the answer is yes. But I really don't know how this testing mode is working exactly.

Strategy Tester: Modes of Modeling during Testing
Strategy Tester: Modes of Modeling during Testing
  • 2005.09.13
  • MetaQuotes Software Corp.
  • www.mql5.com
Many programs of technical analysis allow to test trading strategies on history data. In the most cases, the testing is conducted on already completed data without any attempts to model the trends within a price bar. It was made quickly, but not precisely
 
Alain Verleyen:

Ok. From what is said in this article, the answer is yes. But I really don't know how this testing mode is working exactly.

Hmmm, in simple words. Higher time frame has more control points and will be comparable to every ticks mode.

 of cos I fully understand every tick mode is still the most accurate 

 
Dua Yong Rew:

Hmmm, in simple words. Higher time frame has more control points and will be comparable to every ticks mode.

 of cos I fully understand every tick mode is still the most accurate 

I have a Breakout EA that works on the daily timeframe. Is it better to test it using control points or tick data? Which one will be more accurate?
 

Control points, is not even close to accurate, because it is not how the price went down-up (example) but on strict manner by OHLC from the lower timeframe you are testing. So, it happened (example) price on the lower timeframe in real first move went down then up, but when testing it shows first up then down, so it is more than not accurate, better test it on Open prices only. Why:

See the video. So that same test in real time it only fills the pocket to the brokers with more than 400 trades per symbol per day, and does not move from same as start. Not losing but not gaining too.

I mean, with little modifications, it can, but without..: it is wrong parameters input "by the test", but it needs to be modified. If you understand the difference. Better open prices, are more real. I've shown to you with example.

 
M.R.:

Control points, is not even close to accurate, because it is not how the price went down-up (example) but on strict manner by OHLC from the lower timeframe you are testing. So, it happened (example) price on the lower timeframe in real first move went down then up, but when testing it shows first up then down, so it is more than not accurate, better test it on Open prices only. Why:

See the video. So that same test in real time it only fills the pocket to the brokers with more than 400 trades per symbol per day, and does not move from same as start. Not losing but not gaining too.

I mean, with little modifications, it can, but without..: it is wrong parameters input "by the test", but it needs to be modified. If you understand the difference. Better open prices, are more real. I've shown to you with example.

I always thought open price is only one tick per bar
Control price is 4 ticks per bar
 
Dua Yong Rew:
I always thought open price is only one tick per bar
Control price is 4 ticks per bar

Yes, but not as it happens in real time, ie. in reality.

With 'Open prices' you are able to adjust your expert to check only once at open and all the functions only once, then to Sleep or wait for new bar to open, that is same as the test. Of course if it is not overoptimised with fixed parameters so it would not work as the test in next time (similar as AI-expers, 'Artificial Intelligence'). It is real on experts not counting entries on bar changes, but bigger trends.

'Control points' is good in situations if you have some 'Channel Expert' and it does not depend on 'bar' range movements, it is example, can be for big trends too.

But in cases where are calculated bars levels, example "price bigger than previous mid and vice versa", it is impossible with the previous 2 to get realistic tests for the future.

Control points: if You test on H4, it will take from H1, but if you test H1 it will take from M30,... ie. never from M1 or ticks.

Wow, I wrote a movie from text.

 

i need too in thise problam 

i write a exp that in openprice give mounthely 100% and in control give70% but in tick give -50% how returng tick mode to control mode ???? please help its so important for me