Correlation, measured.

What you get

• Correlation of log returns between an asset and a benchmark, over a window you set
• Beta, so you can separate "moves with it" from "moves more than it"
• Lead/lag diagnostics: whether one series tends to move a bar before the other
• Spread statistics with a hedge ratio, z-score and half-life for mean-reversion work
• The number of aligned observations, reported alongside every figure

Why this is usually paid for

Correlation matrices are a standard feature of paid analytics products, usually as a coloured grid updated on someone else’s schedule over a window you cannot change.

The computation is trivial once both series are in the browser. What is worth paying attention to is the window and the alignment, which is why those are yours to set and are reported back with the answer.

How to use it

1. 1

   Pick a pair and a benchmark

   Most crypto questions are really "is this moving on its own, or is it moving because BTC is". Use BTC as the benchmark for that, and something else when the question is different.

2. 2

   Choose the window on purpose

   A correlation is a statement about a period, not about two assets. Three weeks of hourly bars and six months of daily bars can disagree completely, and both can be correct. Say which one you used.

3. 3

   Read beta next to correlation

   Correlation says how reliably two things move together; beta says how far. A 0.9 correlation with a 0.4 beta is an asset that follows faithfully and barely moves — a very different position from a 0.9 correlation with a 1.8 beta.

4. 4

   Check the observation count before believing anything

   A high correlation over 40 aligned bars is noise with a decimal point. The observation count is reported for exactly this reason; treat it as part of the result rather than as a footnote.

Questions

Also free