Plain-English daily read on which Bybit perps are drifting apart, which pair spreads are stretched, and whether any of it has actually made money historically. Signal-only — no auto-trading.
Coins are moving more independently than usual. This is the friendliest backdrop for relative-value trades: when two historically-linked coins drift apart here, it's more likely to be a genuine divergence than market-wide noise.
Average pairwise 60d correlation: 0.439 (was 0.53 14 days ago, now falling) across 210 pairs.
NEARUSDT and 1000PEPEUSDT used to move together (60d corr +0.51) but that link just broke (14d corr +0.18). Over the last 7 days NEARUSDT moved +29.6% vs 1000PEPEUSDT's -6.4% — a 36.0pt gap. Historically the weaker leg (1000PEPEUSDT) is the one that tends to catch up.
ZECUSDT and SUIUSDT used to move together (60d corr +0.63) but that link just broke (14d corr +0.23). Over the last 7 days ZECUSDT moved +30.6% vs SUIUSDT's -1.5% — a 32.1pt gap. Historically the weaker leg (SUIUSDT) is the one that tends to catch up.
NEARUSDT and DOGEUSDT used to move together (60d corr +0.47) but that link just broke (14d corr +0.11). Over the last 7 days NEARUSDT moved +29.6% vs DOGEUSDT's -2.0% — a 31.6pt gap. Historically the weaker leg (DOGEUSDT) is the one that tends to catch up.
NEARUSDT and SUIUSDT used to move together (60d corr +0.59) but that link just broke (14d corr +0.08). Over the last 7 days NEARUSDT moved +29.6% vs SUIUSDT's -1.5% — a 31.1pt gap. Historically the weaker leg (SUIUSDT) is the one that tends to catch up.
NEARUSDT and WLDUSDT used to move together (60d corr +0.41) but that link just broke (14d corr -0.04). Over the last 7 days NEARUSDT moved +29.6% vs WLDUSDT's +4.9% — a 24.8pt gap. Historically the weaker leg (WLDUSDT) is the one that tends to catch up.
NEARUSDT and USELESSUSDT used to move together (60d corr +0.40) but that link just broke (14d corr +0.06). Over the last 7 days NEARUSDT moved +29.6% vs USELESSUSDT's +10.2% — a 19.4pt gap. Historically the weaker leg (USELESSUSDT) is the one that tends to catch up.
SNDKUSDT and XRPUSDT used to move together (60d corr +0.15) but that link just broke (14d corr -0.19). Over the last 7 days SNDKUSDT moved +13.4% vs XRPUSDT's -4.0% — a 17.4pt gap. Historically the weaker leg (XRPUSDT) is the one that tends to catch up.
SNDKUSDT and HYPEUSDT used to move together (60d corr +0.34) but that link just broke (14d corr -0.00). Over the last 7 days SNDKUSDT moved +13.4% vs HYPEUSDT's -3.8% — a 17.2pt gap. Historically the weaker leg (HYPEUSDT) is the one that tends to catch up.
The beta-adjusted spread between NEARUSDT and LINKUSDT is 3.4 standard deviations from its 30-day norm — stretched enough to flag. Playbook: fade NEARUSDT (the rich leg), lean toward LINKUSDT.
The beta-adjusted spread between NEARUSDT and ENAUSDT is 3.0 standard deviations from its 30-day norm — stretched enough to flag. Playbook: fade NEARUSDT (the rich leg), lean toward ENAUSDT.
The beta-adjusted spread between NEARUSDT and DOGEUSDT is 3.0 standard deviations from its 30-day norm — stretched enough to flag. Playbook: fade NEARUSDT (the rich leg), lean toward DOGEUSDT.
The beta-adjusted spread between NEARUSDT and 1000PEPEUSDT is 2.9 standard deviations from its 30-day norm — stretched enough to flag. Playbook: fade NEARUSDT (the rich leg), lean toward 1000PEPEUSDT.
The beta-adjusted spread between NEARUSDT and SUIUSDT is 2.9 standard deviations from its 30-day norm — stretched enough to flag. Playbook: fade NEARUSDT (the rich leg), lean toward SUIUSDT.
The beta-adjusted spread between NEARUSDT and ADAUSDT is 2.9 standard deviations from its 30-day norm — stretched enough to flag. Playbook: fade NEARUSDT (the rich leg), lean toward ADAUSDT.
The beta-adjusted spread between IOSTUSDT and SNDKUSDT is 2.7 standard deviations from its 30-day norm — stretched enough to flag. Playbook: fade IOSTUSDT (the rich leg), lean toward SNDKUSDT.
The beta-adjusted spread between NEARUSDT and TAOUSDT is 2.7 standard deviations from its 30-day norm — stretched enough to flag. Playbook: fade NEARUSDT (the rich leg), lean toward TAOUSDT.
Each hypothesis is walked backwards through history, its forward returns measured net of a realistic 0.84% round-trip cost, split into in-sample/out-of-sample halves, swept across a grid of parameters (a real edge should survive nearby settings — a CLUSTER; if only one exact setting works it's a SPIKE, i.e. overfit), and bootstrap-resampled 2,000+ times for a probability the edge is real. “Promotable” requires n≥30 signals, positive out-of-sample return, a CLUSTER verdict, P(edge>0)≥95%, and a clean point-in-time audit.
| Hypothesis | Signals | 7d net % | OOS % | Verdict | P(edge>0) | Promotable? |
|---|---|---|---|---|---|---|
| H1: Correlation breakdown → convergence bet | 26 | +1.33 | -4.96 | CLUSTER | 0.71 | no |
| H2: Correlation breakdown → persistence bet | 26 | +1.33 | -4.96 | CLUSTER | 0.71 | no |
| H3: Spread z-score reversion | 33 | +1.95 | +16.56 | SPIKE | 0.77 | no |
| H4: Beta deviation reversion | 0 | +0.00 | +0.00 | NO-EDGE | 0.00 | no |
| H5: Funding-stretched divergence | 1 | +1.15 | +1.15 | NO-EDGE | 0.00 | no |
| H6: OI-confirmed divergence | 26 | +1.33 | -1.37 | CLUSTER | 0.72 | no |
| H7: Volume-confirmed divergence | 4 | +4.09 | -2.07 | SPIKE | 0.68 | no |
| H8: Cross-sectional dispersion burst | 0 | +0.00 | +0.00 | SPIKE | 0.00 | no |
| H9: Sector/family reversion | 1 | +0.50 | +0.50 | NO-EDGE | 0.00 | no |
| H10: Stablecoin/peg proxy | 68 | -1.03 | -4.17 | NO-EDGE | 0.04 | no |
Beta-hedged pair trades, next-bar-open entry/exit, real fees + slippage, top-5 signals/day, max 5 concurrent, regime-gated. Survivorship-biased — the universe is today's top-N, so dead coins that would have hurt short legs are invisible.
| Total return | +50.64% |
| Sharpe (ann.) | 3.644 |
| Max drawdown | -18.26% |
| Win rate | 52.0% |
| Profit factor | 2.318 |
| Trades | 25 |