Should I use Tick Data when back testing?

Hello,

I am currently develop trading strategies in Ninja Trader 7 (migrating to Ninja Trader 8) and back testing with Ninja Trader OHLC data points.

I would like high quality back test accuracy to have some faith in my back test results.

Would you recommend using purchased tick data or OHLC data points when back testing your strategies? The trading strategies I develop are for intra-day trading.

The strategies I create are for scalping or profit targets 7-10 ticks away from entry.

If yes, please refer some service where I can buy high quality tick data?

Thank You

 

It depends on a strategy. When back testing indicators, OHLC is fine as long as you planning to enter at the end of the bar. Otherwise some indicators might not calculate correctly. Chart will display a value, but it might not be a value when bar is forming in real time.

For full auto strategy, you need tick data for replay.

Kinetick is absolutely reliable, but they only provide 6 months worth of tick data. Most vendors where I bought data turned out garbage. Get it straight from exchange (that's what I did), pricey but garbage in garbage out.

 

Thank you RedDuke for the response,

What do you mean get tick data from the exchange? Can you please refer a link?

Thanks,

 

It depends what you're backtesting.

For some things it would be totally unnecessary and completely pointless.

 

Thanks Xela,

The strategies I create are for scalping (intraday strategies on 3-5 min bar chart) or profit targets 7-10 ticks away from entry.

I am concern that using OHCL give in correct results when entry and profit target is on the same bar.

Thanks

 

Try the anfutures.com tick data, the simplest and cheapest 15+ years history as I know. They offer ES and NQ futures only, but just enough to start testing...

 

%%
I dont ; but it takes all kinds to make a market. Very few systems day trade with profit; that is most likely why many traders/investors enjoy /use discretion. One system that has day traded with profit in the past,averages a trade every 2 days; i dont use it or day-trade or remember the average number of winning bars/candles per win.

 

Yes - makes sense: you'll certainly want tick data for backtesting that. And good luck with it.

 

Thank you so much MarkGroes

 

Tick data is just going to give you increased precision. It's not going to tell you if you could had actually traded at those prices - just imply it. Only the book data will tell you what is/was pragmatically tradable - either by hitting the bid or ask or assuming some known executable midpoint range that's product specific.

As an example of what I mean - imagine a product with 10 tick wide spreads (which totally exist). The tick data will show you where trades occurred and their size but that isn't going to actually confirm YOU could have made those trades without appropriate execution methods (resting orders, time in the queue, etc). Even if the next tick is only a tick away, the book could have been shifting around its "window" during that time. Best thing one could do here is assume a product specific amount of slippage and size and go from there.

Sadly market data for most products is both large, expensive, and hard to come by.