Holding ten different tokens is not diversification if they all move together. This scorer looks at how your capital is spread across asset classes and how correlated those classes are, then condenses it into a single 0-100 diversification score with concrete concentration warnings.
How it works
Two factors drive the score. First, concentration is measured with the Herfindahl-Hirschman Index (HHI), the sum of squared allocation weights:
HHI = Σ (weightᵢ)² (weights as fractions, e.g. 0.25)
effective positions = 1 / HHISecond, the tool estimates the average pairwise correlation of your allocation using a fixed correlation matrix between classes, where stablecoins are treated as near-uncorrelated and the volatile risk-on classes (L1, L2, DeFi, gaming) are moderately to highly correlated. The final score rewards low HHI and low average correlation:
score = 100 × (evenness factor) × (1 − avg correlation factor)Why ten tokens can still mean poor diversification
Consider a portfolio that is 50% in BTC, 30% in ETH, 10% in SOL, and 10% in AVAX. That is four assets and four names — it looks diversified. But all four are layer-1 coins with high pairwise correlations, especially during market stress. The HHI is 0.25 + 0.09 + 0.01 + 0.01 = 0.36, giving roughly 2.8 effective independent positions. The average correlation penalty is high because the L1 class dominates. The score reflects the real risk: a broad market sell-off hits all four simultaneously.
Adding a 10% stablecoin allocation and redistributing reduces both the HHI and the average correlation, improving the score even though you have “more crypto” broadly.
Asset classes explained
| Class | Description | Typical correlation with other crypto |
|---|---|---|
| L1 | Base-layer chains (BTC, ETH, SOL) | High with each other |
| L2 | Layer-2 scaling networks (ARB, OP, MATIC) | High with L1 |
| DeFi | Decentralised finance protocols (AAVE, UNI) | Moderate to high with L1 |
| Stablecoin | Pegged tokens (USDC, DAI, USDT) | Near zero |
| Infrastructure | Oracles, data layers (LINK, FIL) | Moderate |
| Gaming | Play-to-earn and NFT-adjacent tokens | Moderate to high in risk-on; low in bear |
Limitations of any static diversification model
The correlation matrix used here is representative but fixed. In practice, correlations in crypto markets are highly unstable. During calm periods, L2 tokens may move somewhat independently of BTC. During sharp sell-offs, nearly every crypto asset converges toward correlation 1.0 as leveraged positions unwind simultaneously. True diversification against systemic crypto risk requires exposure outside the crypto asset class entirely — traditional equities, bonds, commodities, or cash.
Use the score as a planning tool to reduce unnecessary concentration, not as a guarantee of protection during market crises.