Forex automation & data availability

Hi,

I designed some systems for NT that do good on Kinetick's data on 15min charts. When I run them on IB's data they fail completely.

IB only provide data for the last 365 days.
So, it seems quite hard to design an FX strategy for IB that could be backtested on more than 365 days? Which makes it impossible to trade FX automated on IB.

 

Or is there something I am missing? Or just have to rely on larger timeframes where the data is more similar from different sources?

 

You should be able to compare your entry and exit points between IB and Kinetick for the last year.

I guess you have taken into account commissions and slippage, so:

- are the trades really different ?

- do you backtest renko bars, 3 line break and so on ? very difficult to backtest.

- do you use future information in your backtest that wouldn't be available in real time (data-snooping bias) ? Be careful with indicators like zig-zag.

- do you use very tight stops and targets so that they all happen within the same bar ?

 

Thanks for your answer.

- are the trades really different ?
-Yes

- do you backtest renko bars, 3 line break and so on ? very difficult to backtest.
-No but comparing ranges and open and close on bars on pairs like EURSEK.

- do you use future information in your backtest that wouldn't be available in real time (data-snooping bias) ? Be careful with indicators like zig-zag.
-not using indicators

- do you use very tight stops and targets so that they all happen within the same bar ?
-For this I paid my dues earlier

It seems the data and consecutively backtesting results are possibly more similar on more common pairs though.

 

See if you could generate signals using kenetick data and simulate execution using IB data... If the results are still wildly different, try comparing 15-minute bars on the last year's data... Are open, close, high and low different on most bars?

 

Thanks for the advice!

I did not succeed to generate signals on kinetick and execute on IB because the quotes are so different. The data on eg. EURSEK is on a 15m chart visibly different between Kinetick and IB.

However I made the systems less dependent on individual bars and more on more general calculations and the results started to be more similar.

 

Can you compare VWAP over short term bars over the two data sources? That should be more robust to individual trade prices than OHLC. If that is close but you still need OHLC for your strategy, maybe you can try your system on bars built out of short term VWAPS, like a 15 minute bar made out of 30 30-second VWAPS instead of individual ticks.

 

Thanks for the tip about VWAP.
If I focus more on futures I will test using it but now since I trade forex I cannot rely on it so much since its calculated from volume. Also using the combo of IB TWS + NT adding to the challenges of using volume based calculations.