How far to backtest intraday?

My perspective is the one-minute chart, feeling for the trending day, long or short. And I do find testing over only six months to be reliable. But not three. And I find that optimized profit targets vary over those time scales.
 
Do you use the entire 6 months of data for in-sample testing, or do you reserve some of it for out-of-sample confirmation of in-sample results?
 
Quote from Retief:
... Something like a cockroach that eats crumbs and scuttles out of the way when larger animals are about...
More...
... or a baleen whale; a filter-feeder capable of long periods submerged and unnoticed, that trawls oceans devouring vast amounts of unseen, microscopic prey.
 
... or an eagle; the patient hunter circling high up and in wait for long periods, then swooping to tear flesh from its prey.
 
I think the serious point is that there are many different types of strategy that co-exist in the markets.

How long should you backtest a strategy?

Depends what sort of strategy you are dealing with.

If it is a cockroach strategy (i.e. based on an edge exploiting a persistent aspect of market structure) then a backtest over years of data ought to be able to validate this and give you extra confidence that it will work going forward.

If your strategy is an eagle strategy on the other hand (i.e. that successfully exploits some current, temporary aspect of market structure) then there is little point backtesting the strategy over a period of time that you know that the current situation didn’t exist.

[If they had been around (were they?), cockroaches might have done OK during the period of the dinosaurs. I expect eagles would have had more difficulty.]

The challenge with an eagle strategy is that you need to have a good understanding of why it happens to be working now, so that you can stop using the strategy when market structure changes; this is less of a problem with cockroach strategies ...
 
Quote from BrooksRimes:

For an intraday system such as based on 5 minute bars that generates a few trades per day, how far should it be backtested?

This is for futures (ES, NQ, TF, YM ZB).

Brooks
More...

As far as it takes to give you a couple of thousand trades. If you do not get that keep adding data. The right question to ask is how many trades and then determine how far.
 
You must log in or register to reply here.
Back
Top