your analysis is based on this, correct?

https://en.wikipedia.org/wiki/Hurst_exponent

if yes, how do you use it as a reverting indicator since, by definition, it is a trend-measuring tool?

...I had worked a bit with it in the past, it seemed to be helpful for higher periodicities which makes sense given the 'noisiness' of high frequency data.

Well, first, you should not place limitations on certain analysis methods, as being only this....or only that. Simply by applying process of elimination.

If something has no trend....or no trend strength....then what is the probable condition? Also combining frequencies is a crucial part of any process. Never analyze from perspectives of singularity.

Yes, it is based off the Hurst Exponent. However we use proprietary calculations/permutations/methods, etc. Mixing up the calculation, but also the data it analyzes. This is just a general guide/analysis method....one of many tools to place in your arsenal. We never use any single indicator, analysis method, study, research, etc to construct a strategy. We do not develop HFT strategies, nor do we use HF data. We have a few scalping strats, which mainly focuses on price, volume, and order book/depth deviations/volatility/spread/extremes....the majority of executions are reverting. ES approx 85% Reverting and 15% momentum.

Mixing up the Hurst calculation, but also the data it analyzes, will shine some light into some different areas. Dont simply apply to High and Low prices. Implement Price AND Volume. Separate your data, into:

Supportive Upward movements & volume

Supportive Downward movement & volume

Then apply your analysis to each, individually. Then compare.

So you mix up the Hurst calculations and then you mix up the Data could you expand on these statements or is it too proprietary?

Hurst exponent, as a measure of roughness, is not designed to show you any trend reversal... it is actually not designed to show a specific trend at all but just how random/volatile a timeseries is but that's a different discussion...

Of course, if you look at a graph you can make a number of different thoughts, but, personally speaking, I wouldn't use RSI to predict a trend.

Cheers,

Tasos

"Hurst exponent, as a measure of roughness, is not designed to show you any trend reversal... it is actually not designed to show a specific trend at all but just how random/volatile a timeseries is but that's a different discussion..."

YES,....Correct. First, Hurst xpo, or variations of hurst, or the concept in general, is just a general form/method of analysis, and should not be solely used as a foundation for a concept.....as with any "Indicator" it has limitations. Its just one of many "Tools" that should be in your toolbox.

There are some "Reversal" probability signals that can be extracted with variations of hurst, especially when paired with other analysis methods and when Hurst is applied to, then compare Upward Supporting Data vs Downward Supporting data. Since it measures a form of volatility, and if volatility changes with either data category, this can be perceived to give indication of change fo underlying structural support for direction or sentiment/bias.

So while the actual Hurst EXP method may not be a "Trend" or "Reversal" identifier, proprietary forms of Hurst, may be more insightful into these areas of analysis. However, even so, should not be used on its own.