Yes I follow ehlers work closely, and have even managed to beat some of his best indicators - and at present waiting for a verdict from ehlers himself on some of my work.

On filtering 5 second ticks, you might want to look into particle filtering. The book “Introduction to High Frequency Finance” provides examples - but it all depends on how much smoothing you want to achieve. I’d imagine changing the sampling rate i.e. to 10 seconds and doing some kind of (spline) interpolation between the points can help. Is this data very noisy?

Looks like you must have crossed paths with Ehlers at some point? You can google Ehlers and find discussion of his Z based algorithms mentioned in his ebooks (complete with code) in the 2003,4 time frame but discussion dried up and recent trading shops selling his algorithms for trading your account I hear are underwater.

Do you have recent successes, links, systems you can describe that are working in todays wip-saw markets? I’m especially interested in tick/5 second bar intraday algorithms??

Thanks, Curt

My approach here is from a purely signal processing perspective, where one attempts to deduce characteristics of a time series given the time series itself and nothing else. But your findings are interesting — maybe this points to an opportunity in the future to improve our models by combining the benefits of two different methods that do the same thing. Thanks for posting!

While its a popular method, I’ve also found that combining fixed interval time periods (ie 5 min interval) with inconsistent interval periods (ie 500 contract interval) yields interesting results… although I’m still trying to fully comprehend it all. Depending upon the setup used, the inconsistent method, if small enough compared to the fixed time method, can often predict very near-term moves, much like local predictability, which in high-frequency can be used to determine short-term trends. Just $0.02 but may be off the mark. Either way, very interesting post! Thanks!