On TFH it states detecting price clusters is easy.

How would an easy approach work in Lite-C?

Had the same question when I read this article....

@JCL would you give some hints please?

what u mean "TFH"?

"The Financial Hacker", I'm assuming.

Yes, the finacial hacker.

https://financial-hacker.com/build-better-strategies-part-2-model-based-systems/

Still, clusters in price curves are real and can be easily identified in a histogram similar to the cycles spectogram.

Question is, can they be easily identified during live session?

Yes. Store the last N prices, then distribute them into several bins, each one with a fixed price range. The cluster is then the bin with the most prices.

Could anybody give a hint how to approach that (Lite-C only)?

I mean the buckets and counting?

I assume NumInRange(... is the easiest?

@JCL how much work would it be to have a sample code for that online in the manual?

My approach so far

vars Lows = series(priceLow());
vars Highs = series(priceHigh());
vars Maxs = series(priceHigh());
vars Mins = series(priceLow());



vars cls = series(NumInRange(Lows, Highs, Mins[0], Maxs[0], 24*7));

I am not sure NumInRange itself is suitable since it counts bars without considering their range.

Instead decide a bin width, let's say 10 pips. Make a loop and count bins by checking low and high of each bar. Now you have a histogram. Then you can check peaks, valleys or low count bins next to high count bins for SR zones. You need to quantify "low count" and "high count" for computer as well.