Not sure if I should devote a webinar to this, but considering how many people have asked about walk-forward applications in R, I wrote a blog post per request of one of my readers to do it with asset management data.
While I don't use quantstrat's walk-forward functionality, when you boil it down, a walk-forward test is simply a change of parameters at each rebalancing period according to some objective function (EG: every month, pick the best parameter set that maximizes cumulative return over the past month, for instance).
So here's the link to such a post.
A Walk-Forward Attempt on FAA | QuantStrat TradeR (
So what I did:
I ran about 100 different permutations in parallel of the algorithm that generates my returns (there are three posts so far on this topic--Flexible Asset Allocation), and each month, I'd pick the one (or equal-weight the ones) that had the highest return over the past time period.
I'm not sure if this is a bit more ad-hoc than what some readers may like, but if you start off with an initial equity stake in quantstrat, once you extract your returns (see some of my demos in the analytics section for where I do this), then you can certainly do walk-forward analysis on trading strategies, albeit without the per-trade statistics if you go this route.