Very Fast BackTesting

[Pablo]

Hello guys,

from yesterday night til today morning I could backtest data within 10 seconds.
But not it takes me like 40minutes...

What happened? Is there a way to backtest like in speed of light again? That was absolutely amazing.

Please, help me.

[Eleni Anna Branou]

Pablo Jaguanharo Carvalho Pinheiro:
Hello guys,

from yesterday night til today morning I could backtest data within 10 seconds.
But not it takes me like 40minutes...

What happened? Is there a way to backtest like in speed of light again? That was absolutely amazing.

Please, help me.

You probably changed the Modelling Quality you use for your backtests.

Every tick and every tick based on real ticks take longer, 1 minute OHLC and open prices only are much faster.

Another reason could be the symbols you are testing, a much used symbol with already downloaded history backtests much faster than a new one that you need to download its history on the strategy tester.

[Pablo]

Thanks a lot Eleni.
Im glad for your response, but anyway it was a lot faster.
I realized that not all my processor cores are being used, only one from 8.
Do you know how to make full use of my cpu capacity?
Thanks again.

[Pablo]

Oh, it has to do with modelling as you said.
Fantastic! I tried this "only price opening" and it goes really fast.
But how does that affect the quality of my backtesting? are those data reliable?

[Sergey Golubev]

Pablo Jaguanharo Carvalho Pinheiro #:
Oh, it has to do with modelling as you said.
Fantastic! I tried this "only price opening" and it goes really fast.
But how does that affect the quality of my backtesting? are those data reliable?

This is the article about it:
Testing trading strategies on real ticks

and my explanation about how I understand it: post #255

[Eleni Anna Branou]

Pablo Jaguanharo Carvalho Pinheiro #:
Oh, it has to do with modelling as you said.
Fantastic! I tried this "only price opening" and it goes really fast.
But how does that affect the quality of my backtesting? are those data reliable?

If you are testing a long term strategy with a few trades and more than 10-20 pips of TP and SL, your results will be quite trustworthy.

But if you are testing a scalping strategy with many trades that go for a couple of pips each time, they won't be.

[Pablo]

Thank you folks!
Thanks Eleni and Sergey for the responses.
my best regards.

[Olisa Ufondu]

Eleni Anna Branou #:

You probably changed the Modelling Quality you use for your backtests.

Every tick and every tick based on real ticks take longer, 1 minute OHLC and open prices only are much faster.

Another reason could be the symbols you are testing, a much used symbol with already downloaded history backtests much faster than a new one that you need to download its history on the strategy tester.

Thank you Elena. This was super helpful!!