A little scepticism is healthy, however, many of you know the feeling of being plagued non-stop by this awkward feeling.

My level of scepticism is always bobbing around unhealthy levels.

So it's nice when a guy like Ilya questions numbers and kicks the tyres a little.

An opportunity to treble check everything.

I reconciled the Lazy Backtesting IDE results for the Fractal Strategy with a manual Excel calculation and couldn't find anything grossly wrong.

Still, it differs from R's Hurst figure.

I know nothing about R.

In any case, this prompts the question.

How do you know when you are correct?

This is fundamental to financial analysis.

Firstly, the number of eyeballs and amount of man hours invested into checking numbers is important.

Being right is not democratic though - you need to understand the problem.

The Hurst statistic is 'estimated' against a sample of data chunked into ever smaller chunks.

Each day's estimate has an R2.

Using new tightened code (nothing majorly different) the average R2s are,

The '2^n' chunks are new.

The Sharpes from each configuration are,

The extremely high R2s give me a lot of confidence that the Hurst estimates are correct.

The broadly positive R2 and Sharpe correlation is a nice extra.

There could be bugs in other parts of the Lazy Backtesting IDE, no doubt, that's why I hope more sceptics get in touch.

Code is here.