Cryptocurrency Relationships Revealed: Correlation Heatmaps
Compare how liquid cryptoasset returns move together now versus the prior window—and export a stable, citable view.
Fresh ·CryptoDigest first explored cryptoasset correlations in 2018. This edition rebuilds that research question with current data, transparent calculations, and downloadable results; it is not presented as a copy of the earlier article.
Choose 2–8 assets. Each main value is the current-window coefficient; the smaller value is its change from the immediately preceding equal-length window.
Pearson correlation of 1d log returns. The smaller number is the change from the immediately preceding window. Correlation describes co-movement, not causation.
| Asset | BTC | ETH | BNB | SOL | DOGE | LINK |
|---|---|---|---|---|---|---|
| BTC | 1.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
| ETH | 0.000 | 1.000 | 0.000 | 0.000 | 0.000 | 0.000 |
| BNB | 0.000 | 0.000 | 1.000 | 0.000 | 0.000 | 0.000 |
| SOL | 0.000 | 0.000 | 0.000 | 1.000 | 0.000 | 0.000 |
| DOGE | 0.000 | 0.000 | 0.000 | 0.000 | 1.000 | 0.000 |
| LINK | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 1.000 |
Accessible data table
The interactive matrix includes a screen-reader table containing every selected coefficient.
Key findings
- 01
BTC and ETH move most closely in this window
Their return correlation is 0.00. This describes co-movement, not a shared cause.
- 02
BTC / ETH changed most
The coefficient changed 0.00 from the previous equal window.
- 03
Correlation is regime-dependent
A coefficient can change when market-wide volatility, liquidity, or idiosyncratic information changes; one window should not be treated as permanent.
What the original resource established
The original CryptoDigest study asked a durable question: which major cryptocurrencies tended to move together, and did those relationships differ across periods? Its heatmap presentation was later cited in academic finance research.
This edition preserves that question, the matrix form, the distinction between stronger and weaker relationships, and the need to specify a sample period. It does not present reconstructed prose or unknown historic coefficients as recovered originals.
A reproducible, window-aware lab
The interactive version computes coefficients from log returns, exposes the observation interval and lookback, compares adjacent equal-length windows, and flags absolute changes of at least 0.20 for inspection.
Exports contain derived coefficients—not raw provider observations. This edition is calculated from the latest retained production candles.
Methodology
For each asset, price observations are ordered by UTC bucket and transformed to log returns: rₜ = ln(Pₜ / Pₜ₋₁). Pairwise Pearson coefficients are calculated on overlapping observations only.
The selected sample is divided into current and immediately preceding equal windows. A “meaningful break” is a descriptive review threshold when the absolute coefficient change is at least 0.20; it is not a calibrated statistical significance claim.
Quality rules
- Exclude stale and invalid prices before computing returns.
- Use internal asset identities rather than ticker text alone.
- Display the interval, window, observation time, method version, and limitations.
- Do not infer causality or future performance from correlation.
Limitations
- Pearson correlation summarizes linear co-movement and can miss nonlinear relationships.
- Results are sensitive to sampling interval, window length, missing observations, and market regime.
- Shared quote-currency and market-wide factors can inflate apparent relationships.
- Provider coverage, retention, and quality state can affect the resulting coefficients.
- Past correlation does not predict return or establish a causal mechanism.
Data freshness
Expected update cadence: daily for public research views; five minutes for current market state.
Data and reuse
Embed
Responsive figure with visible CryptoDigest attribution.
Sources and evidence
Primary and approved sources used for this edition.
| Source | Tier | Published | Supports |
|---|---|---|---|
| CryptoDigest derived market dataset (coinmarketcap)Normalized daily stored returns | 2 | Every displayed matrix value is computed from stored prices rather than generated prose. |
Cite this resource
Stable edition 2026.08
Version history
- 2026.08
Interactive lab, adjacent-window comparison, exports, embed, and explicit reconstructed-edition disclosure.
- Original
Original CryptoDigest correlation heatmap study; exact archive metadata remains under recovery.