Coin Diversification & Fit
What it shows
A description of how coins behaved RELATIVE TO EACH OTHER: which pairs moved almost identically (that is, effectively duplicated the same behaviour), which moved independently, and which moved in opposite directions. The "Market" tab computes this over a pool of the largest coins; the "Your positions" tab (over the active positions in your portfolio). We compute from daily returns aligned by UTC date, over a 30- or 90-day window. This describes the past, not a forecast, not advice.
How to read it
Start with the tab and the window: "Market" describes a broad pool of coins, "Your positions" only what you hold open; the 30-day window shows the recent arrangement, 90 days a more durable picture. Pairs under "moved almost identically" are places where two different names historically described one behaviour; "independent" pairs are those whose daily returns did not travel together. For each row check the persistence label: "appeared only in part of the window" means the observation did not repeat across sub-periods and should not be treated as a fixed trait. If a coin is missing from the list, it did not have enough shared days of history; the widget says so outright instead of inventing a value. The whole view is a description of the past, containing no guidance on what to do about it, by design.
Parameters
- Association (r)
- A number from −1 to +1 describing how closely two coins' daily returns moved together. +1 = identically, 0 = independently, −1 = in opposite directions. We compute it from returns, not price levels, which means raw prices drift up with the market and inflate the result.
- Neutral reading: Values near zero mean the coins moved independently of each other in this window. That is an observation about the past, not a promise it will stay that way.
- Descriptive scale
- Instead of a raw number we show a label: "moved in opposite directions" (r ≤ −0.3), "independently" (up to 0.3), "partly together", "together", and "almost identically" (r ≥ 0.85). The "almost identically" band flags two positions that in practice described the same behaviour.
- Stable / variable
- We split the window into three shorter sub-periods and check whether the band comes out the same in each. "Held across sub-periods" = the behaviour was repeatable. "Appeared only in part of the window" = a one-off coincidence, not a lasting trait.
- Groups of 3–5 coins
- Sample sets with the lowest average mutual association. Selection is deterministic and skips stablecoins and wrapped tokens (their "independence" would be trivial); a group also cannot come from a single sector. An illustration of a historical pattern, not a proposed portfolio composition.
- How many distinct behaviours
- An approximation of the number of distinct behaviours in a basket: positions moving identically give 1, fully independent ones give as many as there are positions. A value of 2.3 across 6 positions says roughly two to three different movement patterns historically repeated in it.
- Data coverage
- Every result carries the computation date, the number of coins in the pool, the number of computed pairs and the median of shared days. A pair without enough shared days is SKIPPED. We never show a zero or a stand-in value for it. If only part of the pool cleared the threshold, the widget says outright that this is a slice of the market.
Example
Over a 90-day window the BTC · ETH pair comes out at r = +0.88 with the label "moved almost identically / held across sub-periods", which means historically these two positions described practically one behaviour across the whole window. In the same window XMR · SOL sits at r = +0.04 labelled "moved independently", but also "appeared only in part of the window", so in one of the three sub-periods the result fell into a different band. In the "Your positions" tab, with 6 active positions the distinct-behaviour measure reads 2.4, which means historically about two to three separate movement patterns repeated in the portfolio, not six.