Why is there such a huge discrepancy between "Real Ticks" modelling and "Every Tick" modelling?

New comment
 

I was testing an EA that a developer put on the codebase for trading Gold (based on Keltner channel and EMAs). 


Here is how the backtest looked on "Real Ticks" modelling:

1


Here is how the backtest looked on "Every Tick" modelling for the exact same setting, same leverage, and same time range (same everything except different modelling):

2


I'm guessing that the "Real ticks" modelling is a true reflection of reality?  But why such a drastically different result? 


maybe there is no floating spread or slippage applied to "Every Tick" if it is not based on broker? 

 
Don't use Every Tick, ever, it's completely useless now Real Ticks are available. "Every ticks" are simulated, calculated ticks, obviously the used algorithm is not good enough to be close to the reality.
 

Alain Verleyen #:
Don't use Every Tick, ever, it's completely useless now Real Ticks are available. "Every ticks" are simulated, calculated ticks, obviously the used algorithm is not good enough to be close to the reality.

Thanks, I'm using real ticks modelling now that I know, reality is reality, the other modelling seems to be displaying a result of fantasy

 
Conor Mcnamara #:

Thanks, I'm using real ticks modelling now that I know, reality is reality, the other modelling seems to be displaying a result of fantasy

1 Minute OHLC is very useful to get faster results. If and only if your strategy AND code is adapted to it, otherwise it's misleading.
 
Alain Verleyen # :
1 Minute OHLC is very useful to get faster results. If and only if your strategy AND code is adapted to it, otherwise it's misleading.

Good morning
Could you be more specific in your adapted code remark, please.

 
Gerard Willia G J B M Dinh Sy #:

Good morning
Could you be more specific in your adapted code remark, please.

Not really sorry. I don't want to share any code, in our world it's unfortunately how it works if you want to keep your intellectual property. And I don't even want to share the detailed ideas for the same reason.

The general idea is : 1 Minute OHLC is 4 ticks by minute which of course increase significantly the speed of backtests and optimizations. But for this mode to give suitable results (close to what real ticks will give), you need to have a series of adjustments to your EA. Just one example : the spread, 1 Minute OHLC will use spread from 1M data, which is the MINIMUM spread of that 1M period, it can be significantly different from real ticks. So, if your strategy allow it, you need to take that into account, otherwise your 1M OHLC results will be significantly overestimated. It's possible to reduce this difference seriously, by adjusting the strategy as running live, close to what is happening with 1 Minute OHLC (not entirely of course).

Some strategies are simply not suitable for 1M OHLC, for example if you are relying on ticks for your trading conditions.

 
Merci 
New comment