When Monero jumps hard, it tends to give some back the next day — and when it crashes, it tends to bounce. We measured this over 8.7 years of daily prices.
The trade is the simplest possible expression of that: fade yesterday’s move — short after a pop, buy after a drop — on Hyperliquid, where XMR trades 24/7.
Status: PAPER-TRADE PHASE — validating live, no capital committed. Not investment advice.
Computed directly from the 3,172 daily closes in our own price feed — no model, just averages. Big drops have, on average, been followed by an up day; big pops by a down day. That asymmetry is the entire signal.
Monero barely follows Bitcoin anymore — over the last two years, Bitcoin explains only ~19% of its daily variance (R² 0.19). It lives in its own universe for a real structural reason: as a privacy coin it has been serially delisted from regulated exchanges (Binance 2024, Kraken-EEA 2024, dozens more), so its price-moving events are its own regulatory calendar, not crypto’s. Those shocks knock the price around, and a thin market with a stubborn base of privacy-demand holders tends to snap part-way back — which is exactly the statistical signature we measure: genuine 1-day mean reversion (autocorrelation −0.14, variance ratios significantly below 1).
| Measure | Value | What it means |
|---|---|---|
| Test window | 8.7 years | Nov 2017 → Jul 2026, every day included |
| Observations | 3,170 daily returns | no cherry-picked sub-period |
| t-statistic (net) | 3.07 | unlikely to be luck, not impossible |
| Sharpe ratio (net) | ~1.0 | return per unit of risk, after fees |
| Calendar years positive | 9 of 10 | 2025 was the losing year |
| Costs assumed | 9 bps per flip | real Hyperliquid taker fees, both sides |
| Venue check | replicates on Hyperliquid's own candles | 184 traded days — consistency check, not proof |
Same discipline as everything else on this site: every parameter cell tested is reported, chronological in-sample / out-of-sample split, costs charged on every position flip. The strategy was also the first specification we wrote down, not the best cell of a grid.
Monero daily close, 2017-11-09 → 2026-07-16, from our own price feed — the exact series the backtest ran on. The violence is the point: the big single-day spikes and crashes visible here are what the signal fades. Illustrative of the input data; this chart is not a P&L — no live track record exists yet.
PAPER-TRADE PHASE — validating live, no capital committed. This is a modest, decaying edge that cleared honest testing and real costs; it is not “proven” and we do not call it alpha. If live paper fills match the backtest over 30+ days, a small gated pilot follows — and if they don’t, this page will say so. See the process working at /leadlag and /track-record.
Previously tested and retired: W-1 (funding stress → bond funds) — withdrawn after a data-alignment error was found in our own re-check. Details.