Does anything beat buy & hold on Uniswap (UNI)?
Every setup we tested on Uniswap (UNI) — ranked out-of-sample, corrected for multiple testing, and forward-tracked in public from the day this page published. The honest answer is the headline.
No setup beat simply holding once tested honestly. We say so plainly.
Its best setup's out-of-sample profit factor (0.81) is below 1 — it lost money per trade on unseen data. Buy-and-hold benchmark: -38.3% CAGR over 8.0 years (+1.9% CAGR in the out-of-sample window).
Educational research from historical backtests — not investment advice. Past performance does not predict future results.
Uniswap: Nothing Beat Buy-and-Hold, and We Checked Everything
For Uniswap, we ran 544 indicator configurations through the same pipeline we apply to every asset, and none cleared the bar. This is a common outcome in crypto, where simply holding UNI produced a buy-and-hold CAGR of -38.3%% — alongside a maximum drawdown of -80.0%%, which is the price of admission. When the baseline compounds that hard, a timing rule has to be genuinely predictive, not just lucky during one bull run, to add anything. In a market that trades around the clock and moves violently, most rules here simply stepped out of moves that holding captured for free.
The best-looking candidate was Markov Regime (Confirmed) on the daily timeframe, posting an out-of-sample Sharpe of 1.56 against a multiple-testing hurdle of 2.29. That hurdle exists because picking the top result from 544 attempts manufactures apparent skill by construction. Only 17.8%% of setups beat holding at all, and the leader's edge did not hold up across 2.4 years of unseen data. Read this as evidence, not prophecy: crypto market structure shifts quickly, past performance does not predict future results, and a verdict of nothing today is a finding about history, not a forecast.
Every figure above is computed from our own backtests — nothing is estimated or invented. Hypothetical results; not investment advice.
The least-bad setups — shown with their failure numbers
Nothing here earned a verdict — these are the best of a losing field, published so you can see exactly how "best" still failed.
Markov Regime (Confirmed)
Mechanical rule (exactly as backtested): Stricter Markov-2 variant — long only when the regime signal is strongly bullish AND a learned-boundary 'hidden Markov' state agrees with the fixed +/-5% state. Signals are evaluated at daily-bar close, the position changes on the NEXT bar, 0.08% cost per side, long/flat only — no leverage, no shorting.
Out-of-sample (last ~30% of the window, never used to pick this setup): Sharpe 1.56 · alpha +32.1% · 26 trades over 2.4 yrs.
Markov Regime
Mechanical rule (exactly as backtested): Markov-2 hedge-fund regime model — long while the current state's transition probabilities favor bull over bear (P(bull next) > P(bear next)). Non-overlapping stickiness, strictly walk-forward. Trend-regime following, long/flat. Signals are evaluated at daily-bar close, the position changes on the NEXT bar, 0.08% cost per side, long/flat only — no leverage, no shorting.
Out-of-sample (last ~30% of the window, never used to pick this setup): Sharpe 1.44 · alpha +55.2% · 34 trades over 2.4 yrs.
Pivot Points (Standard)
Mechanical rule (exactly as backtested): Floor-trader pivots from the prior bar — buy near S1, exit at the pivot. Signals are evaluated at weekly-bar close, the position changes on the NEXT bar, 0.08% cost per side, long/flat only — no leverage, no shorting.
Out-of-sample (last ~30% of the window, never used to pick this setup): Sharpe 1.1 · alpha +22.1% · 13 trades over 1.7 yrs.
Since publication — including if it loses
The forward record is just getting started — the gap between the two is the honest score. Marked to market nightly from real prices, rules frozen at publication, as of 2025-04-17. Currently FLAT.
We tested 544 setups (indicator × parameters × timeframe) on Uniswap (UNI). Only setups with ≥30 trades qualify (326 did). Setups are ranked by out-of-sample Sharpe — the last ~30% of history, which standard-parameter rules never saw during selection. Because picking the best of 544 tries mines even the holdout, the VALIDATED verdict additionally requires the top setup’s OOS Sharpe to clear a selection hurdle of 2.29 (√(2 ln N)/√T) AND positive alpha in both windows. Of the eligible setups, 17.8% had positive out-of-sample alpha (median OOS Sharpe 0.55) — the table below is truncated, but this summary covers all of them. Full recipe: methodology · the engine’s contract lives in the repo as STRATEGY_METHODOLOGY.md.
Top 20 of 326 eligible setups
Ranked by out-of-sample Sharpe. Full + out-of-sample columns, costs included. Hypothetical.
| # | Setup | TF | Total ret | Sharpe | Max DD | Win | Trades | α vs B&H | OOS Sharpe | OOS α | OOS trades |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Markov Regime (Confirmed) | Daily | +931.3% | 0.61 | -80.0% | 45.1% | 51 | +72.3% | 1.56 | +32.1% | 26 |
| 2 | Markov Regime | Daily | >+999% | 0.97 | -92.3% | 49.4% | 85 | +179.5% | 1.44 | +55.2% | 34 |
| 3 | Pivot Points (Standard) | Weekly | >+999% | 0.78 | -93.6% | 61.5% | 39 | +115.3% | 1.1 | +22.1% | 13 |
| 4 | Detrended Price Osc. | Daily | >+999% | 0.35 | -99.9% | 42.5% | 214 | +208.7% | 0.9 | +218.8% | 67 |
| 5 | MA Envelope | Daily | >+999% | 0.36 | -93.4% | 67.3% | 110 | >+999% | 0.89 | +23.6% | 23 |
| 6 | Stochastic RSI | Daily | +57.9% | 0.95 | -99.6% | 53.4% | 58 | +44.2% | 0.82 | +166.1% | 21 |
| 7 | Ehlers Cyber Cycle | Daily | -100.0% | 0.49 | -100.0% | 35.3% | 167 | -52.4% | 0.8 | +141.2% | 48 |
| 8 | Connors RSI-2 | Daily | >+999% | 1.49 | -90.7% | 65.3% | 101 | +994.5% | 0.79 | +16.7% | 28 |
| 9 | Historical Volatility Regime | Daily | -99.9% | 0.0 | -100.0% | 44.4% | 54 | -18.2% | 0.74 | +31.0% | 16 |
| 10 | Stochastic Momentum Index | Daily | >+999% | 0.35 | -100.0% | 63.2% | 38 | +79.6% | 0.73 | +14.7% | 11 |
| 11 | Center of Gravity | Daily | -100.0% | -0.11 | -100.0% | 40.0% | 260 | -58.1% | 0.72 | +97.9% | 73 |
| 12 | Derivative Oscillator | Daily | -100.0% | -0.1 | -100.0% | 42.1% | 140 | -57.3% | 0.71 | +99.8% | 41 |
| 13 | VuManChu Cipher B | Daily | -100.0% | 0.04 | -100.0% | 32.5% | 80 | -29.5% | 0.71 | +91.2% | 31 |
| 14 | VWAP Bands | Daily | >+999% | 1.19 | -100.0% | 68.3% | 41 | +118.4% | 0.71 | +12.6% | 10 |
| 15 | Fibonacci Pivots | Weekly | >+999% | 0.98 | -92.6% | 60.9% | 46 | +197.4% | 0.7 | +13.5% | 14 |
| 16 | Hammer | Daily | >+999% | 0.35 | -91.1% | 46.8% | 62 | +193.1% | 0.69 | +106.3% | 20 |
| 17 | Stochastic Fast (5,3) | Daily | -100.0% | 0.16 | -100.0% | 40.4% | 265 | -58.9% | 0.69 | +21.8% | 78 |
| 18 | Bollinger 10 (x1.5) Break | Daily | -100.0% | -0.09 | -100.0% | 42.1% | 114 | -30.7% | 0.68 | +87.8% | 43 |
| 19 | Fisher Transform | Daily | -100.0% | 0.35 | -100.0% | 36.4% | 195 | -55.5% | 0.68 | +17.2% | 55 |
| 20 | Relative Vigor Index | Daily | -100.0% | 0.35 | -100.0% | 43.5% | 168 | -42.7% | 0.67 | +70.8% | 44 |
Hypothetical backtests with 0.08%/side costs. Not investment advice — see the full disclaimer.
These are historical backtests of mechanical rules. They are educational research, not investment advice, not a recommendation, and not tailored to you. Educational information only — not investment advice. Hypothetical backtested results; past performance does not guarantee future results. Trading involves risk of loss.