XJO quant ORB strategy

[CanOz]

Canoz, you've got enough data there to split it in half, optimize the first half and walk forward the second. What does that look like?

I suppose i could optimize the stops or the time series...?

 

[Gringotts Bank]

I suppose i could optimize the stops or the time series...?

The profit targets, the stops, the opening range itself (start/stop time), the days of the week, the time of exit, etc. Anything that improves the curve.

 

[StuDrefus]

Is an index the best type of market for a breakout strategy of this type? Don't equity indices tend to be fairly mean reverting (versus commodities, currencies etc)?

I only know of the ORB in the form presented in Toby Crabel's books from several decades back, but I'm pretty sure he relied on trading quite a diverse portfolio of commodity futures to smooth returns. Likewise with Larry Williams who used a not dissimilar 'volatility breakout' strategy (not based on the opening range).

It would be interesting to see this strategy tested over a portfolio.

 

[CanOz]

Is an index the best type of market for a breakout strategy of this type? Don't equity indices tend to be fairly mean reverting (versus commodities, currencies etc)?

I only know of the ORB in the form presented in Toby Crabel's books from several decades back, but I'm pretty sure he relied on trading quite a diverse portfolio of commodity futures to smooth returns. Likewise with Larry Williams who used a not dissimilar 'volatility breakout' strategy (not based on the opening range).

It would be interesting to see this strategy tested over a portfolio.

In my testing the only open range breakout that works now is on the HKFE index futures. I have another system that works very well fading an open range breakout under certain circumstances (filter). The former work well after a volatility contraction, where as the latter works well if the opening range is already equal to a certain proportion of the average daily range. It works on many different indices to differing degrees of success. I don't use a time series to define the range though.