Pairs and stat arb: 28 Backtests
Data last updated
Reversal and pairs rules that bet two related prices, or a price and its own average, move back together.
2 of 28 strategies finished ahead of the benchmark on the raw number, and 0 had an unadjusted p-value under 0.05. None cleared the correction for trying 1,774 tests in total (a q-value below 0.10), so no result here is a confirmed edge. Scored 2026-10-31, using data up to 2025-03-15.
Strategies here tested on crypto, ETFs, US stocks. See the full family index, the full method and every result, or download the raw registry (JSON).
Results
| Rule | Market | Vs benchmark | Sample | p-value | Corrected q | Receipt |
|---|---|---|---|---|---|---|
| Low Turnover Short Term Reversal receipt Source: Medhat & Schmeling 2022 (short-term reversal and turnover) | US stocks | +3.0 pts/yr | 102 rebalances | 0.159 | 1.000 | On-site receipt |
| Sector Adjusted Short Term Reversal receipt Source: Da, Liu & Schaumburg 2014 | US stocks | +2.5 pts/yr | 102 rebalances | 0.211 | 1.000 | On-site receipt |
| Size-ladder diffusion (mega leads small), daily (wide US stocks) receipt Source: Lo & MacKinlay 1990 RFS; Chordia & Swaminathan 2000 JF; Hou 2007 RFS; replication of sweep150 s02-3 on 8.5 years | US stocks | −0.2 pts/yr | 2,136 rebalances | 1.000 | 1.000 | On-site receipt |
| VIX-conditioned 1-month reversal, equal weight otherwise (wide US stocks) receipt Source: Nagel 2012 RFS (evaporating liquidity: reversal returns rise with VIX) | US stocks | −0.4 pts/yr | 102 rebalances | 1.000 | 1.000 | On-site receipt |
| Short-term residual reversal (wide US stocks) receipt Source: Blitz, Huij, Lansdorp & Verbeek 2013 JFM (short-term residual reversal) | US stocks | −0.7 pts/yr | 102 rebalances | 1.000 | 1.000 | On-site receipt |
| Crash-week loser rebound, 20-session hold (wide US stocks) receipt Source: Daniel & Moskowitz 2016 JFE (losers rebound after market crashes); replication of sweep150 t15-2 with an equal-weight trigger instead of QQQ | US stocks | −1.1 pts/yr | 2,136 rebalances | 1.000 | 1.000 | On-site receipt Full test: Buy the Dip |
| Industry Peer Lead Lag receipt Source: Hou 2007 (industry lead-lag) | US stocks | −3.6 pts/yr | 443 rebalances | 1.000 | 1.000 | On-site receipt |
| Correlation-breakdown gated ETF pair re-convergence receipt Source: none published (Tickfloor research-desk test) | ETFs | −6.9 pts/yr | 633 rebalances | 1.000 | 1.000 | On-site receipt |
| Ungated ETF pair spread reversion (matched control for correlation-breakdown gate) receipt Source: none published (Tickfloor research-desk test) | ETFs | −7.0 pts/yr | 1,269 rebalances | 1.000 | 1.000 | On-site receipt |
| Cointegrated ETF pair (e.g. XLE/XOP, GLD/SLV) z-score reversion receipt Source: Gatev, Goetzmann, Rouwenhorst 2006 'Pairs Trading: Performance of a Relative-Value Arbitrage Rule' | ETFs | −11.3 pts/yr | 2,168 rebalances | 1.000 | 1.000 | On-site receipt |
| Sector ETF short-term reversal (1-week) receipt Source: Jegadeesh 1990 JF; Lehmann 1990 QJE | ETFs | −11.8 pts/yr | 1,052 rebalances | 1.000 | 1.000 | On-site receipt |
| RSI(2) mean reversion, SPY/QQQ (retest with BH2's failure checks) receipt Source: Connors & Alvarez 'Short Term Trading Strategies That Work' | ETFs | −12.2 pts/yr | 399 trades | 1.000 | 1.000 | On-site receipt |
| Sector ETF short-term reversal (1-week) (wide US stocks) receipt Source: Jegadeesh 1990 JF; Lehmann 1990 QJE | US stocks | −13.4 pts/yr | 442 rebalances | 1.000 | 1.000 | On-site receipt |
| Bollinger %B reversion on large-cap crypto (wide US stocks) receipt Source: John Bollinger's writings; widespread crypto retail use | US stocks | −14.2 pts/yr | 17,020 trades | 1.000 | 1.000 | On-site receipt |
| Detrended price rebound receipt Source: none published (Tickfloor research-desk test) | US stocks | −16.8 pts/yr | 638 trades | 1.000 | 1.000 | On-site receipt |
| DeMarker exhaustion receipt Source: none published (Tickfloor research-desk test) | US stocks | −17.1 pts/yr | 201 trades | 1.000 | 1.000 | On-site receipt |
| Ultimate oscillator dip receipt Source: none published (Tickfloor research-desk test) | US stocks | −17.3 pts/yr | 86 trades | 1.000 | 1.000 | On-site receipt |
| Stochastic oversold cross receipt Source: none published (Tickfloor research-desk test) | US stocks | −17.5 pts/yr | 252 trades | 1.000 | 1.000 | On-site receipt Full test: Stochastic, CCI and Williams %R |
| Money flow index exhaustion receipt Source: none published (Tickfloor research-desk test) | US stocks | −17.7 pts/yr | 28 trades | 1.000 | 1.000 | On-site receipt |
| Wide down-bar recovery receipt Source: none published (Tickfloor research-desk test) | US stocks | −17.8 pts/yr | 46 trades | 1.000 | 1.000 | On-site receipt |
| Low-volume trend pullback receipt Source: none published (Tickfloor research-desk test) | US stocks | −18.1 pts/yr | 9 trades | 1.000 | 1.000 | On-site receipt |
| Three-bar island reversal receipt Source: none published (Tickfloor research-desk test) | US stocks | −18.1 pts/yr | 37 trades | 1.000 | 1.000 | On-site receipt |
| Chande momentum dip receipt Source: none published (Tickfloor research-desk test) | US stocks | −18.2 pts/yr | 68 trades | 1.000 | 1.000 | On-site receipt |
| Cross-sectional 1-day reversal, liquid large-cap stocks (wide US stocks) receipt Source: Jegadeesh 1990; Lo & MacKinlay 1990 RFS (short-term reversal); universe = harness's US_STOCKS_DAILY 40 large caps (S&P100 not available) | US stocks | −29.4 pts/yr | 2,135 rebalances | 1.000 | 1.000 | On-site receipt |
| Cross-sectional 1-day reversal, liquid large-cap stocks receipt Source: Jegadeesh 1990; Lo & MacKinlay 1990 RFS (short-term reversal); universe = harness's US_STOCKS_DAILY 40 large caps (S&P100 not available) | US stocks | −34.2 pts/yr | 2,135 rebalances | 1.000 | 1.000 | On-site receipt |
| Crypto Fear & Greed Index contrarian entries receipt Source: alternative.me Fear & Greed Index | crypto | −59.6 pts/yr | 72 trades | 1.000 | 1.000 | On-site receipt Full test: Fear & Greed Timing |
| Crypto 1-day reversal after a high-volume down day receipt Source: Nagel 2012 RFS (short-term reversal & liquidity provision), volume-conditioned | crypto | −66.4 pts/yr | 4,969 trades | 1.000 | 1.000 | On-site receipt |
| Bollinger %B reversion on large-cap crypto receipt Source: John Bollinger's writings; widespread crypto retail use | crypto | −81.2 pts/yr | 1,928 trades | 1.000 | 1.000 | On-site receipt Full test: Bollinger Band Bounce |
- Low Turnover Short Term Reversal
US stocks · +3.0 pts/yr · 102 rebalances · corrected q 1.000
Source, p-value and result file
- Source
- Medhat & Schmeling 2022 (short-term reversal and turnover)
- p-value
- 0.159
- Corrected q
- 1.000
- Receipt
- On-site receipt
- Sector Adjusted Short Term Reversal
US stocks · +2.5 pts/yr · 102 rebalances · corrected q 1.000
Source, p-value and result file
- Source
- Da, Liu & Schaumburg 2014
- p-value
- 0.211
- Corrected q
- 1.000
- Receipt
- On-site receipt
- Size-ladder diffusion (mega leads small), daily (wide US stocks)
US stocks · −0.2 pts/yr · 2,136 rebalances · corrected q 1.000
Source, p-value and result file
- Source
- Lo & MacKinlay 1990 RFS; Chordia & Swaminathan 2000 JF; Hou 2007 RFS; replication of sweep150 s02-3 on 8.5 years
- p-value
- 1.000
- Corrected q
- 1.000
- Receipt
- On-site receipt
- VIX-conditioned 1-month reversal, equal weight otherwise (wide US stocks)
US stocks · −0.4 pts/yr · 102 rebalances · corrected q 1.000
Source, p-value and result file
- Source
- Nagel 2012 RFS (evaporating liquidity: reversal returns rise with VIX)
- p-value
- 1.000
- Corrected q
- 1.000
- Receipt
- On-site receipt
- Short-term residual reversal (wide US stocks)
US stocks · −0.7 pts/yr · 102 rebalances · corrected q 1.000
Source, p-value and result file
- Source
- Blitz, Huij, Lansdorp & Verbeek 2013 JFM (short-term residual reversal)
- p-value
- 1.000
- Corrected q
- 1.000
- Receipt
- On-site receipt
- Crash-week loser rebound, 20-session hold (wide US stocks)
US stocks · −1.1 pts/yr · 2,136 rebalances · corrected q 1.000
Source, p-value and result file
- Source
- Daniel & Moskowitz 2016 JFE (losers rebound after market crashes); replication of sweep150 t15-2 with an equal-weight trigger instead of QQQ
- p-value
- 1.000
- Corrected q
- 1.000
- Receipt
- On-site receipt
- Industry Peer Lead Lag
US stocks · −3.6 pts/yr · 443 rebalances · corrected q 1.000
Source, p-value and result file
- Source
- Hou 2007 (industry lead-lag)
- p-value
- 1.000
- Corrected q
- 1.000
- Receipt
- On-site receipt
- Correlation-breakdown gated ETF pair re-convergence
ETFs · −6.9 pts/yr · 633 rebalances · corrected q 1.000
Source, p-value and result file
- Source
- none published (Tickfloor research-desk test)
- p-value
- 1.000
- Corrected q
- 1.000
- Receipt
- On-site receipt
- Ungated ETF pair spread reversion (matched control for correlation-breakdown gate)
ETFs · −7.0 pts/yr · 1,269 rebalances · corrected q 1.000
Source, p-value and result file
- Source
- none published (Tickfloor research-desk test)
- p-value
- 1.000
- Corrected q
- 1.000
- Receipt
- On-site receipt
- Cointegrated ETF pair (e.g. XLE/XOP, GLD/SLV) z-score reversion
ETFs · −11.3 pts/yr · 2,168 rebalances · corrected q 1.000
Source, p-value and result file
- Source
- Gatev, Goetzmann, Rouwenhorst 2006 'Pairs Trading: Performance of a Relative-Value Arbitrage Rule'
- p-value
- 1.000
- Corrected q
- 1.000
- Receipt
- On-site receipt
- Sector ETF short-term reversal (1-week)
ETFs · −11.8 pts/yr · 1,052 rebalances · corrected q 1.000
Source, p-value and result file
- Source
- Jegadeesh 1990 JF; Lehmann 1990 QJE
- p-value
- 1.000
- Corrected q
- 1.000
- Receipt
- On-site receipt
- RSI(2) mean reversion, SPY/QQQ (retest with BH2's failure checks)
ETFs · −12.2 pts/yr · 399 trades · corrected q 1.000
Source, p-value and result file
- Source
- Connors & Alvarez 'Short Term Trading Strategies That Work'
- p-value
- 1.000
- Corrected q
- 1.000
- Receipt
- On-site receipt
- Sector ETF short-term reversal (1-week) (wide US stocks)
US stocks · −13.4 pts/yr · 442 rebalances · corrected q 1.000
Source, p-value and result file
- Source
- Jegadeesh 1990 JF; Lehmann 1990 QJE
- p-value
- 1.000
- Corrected q
- 1.000
- Receipt
- On-site receipt
- Bollinger %B reversion on large-cap crypto (wide US stocks)
US stocks · −14.2 pts/yr · 17,020 trades · corrected q 1.000
Source, p-value and result file
- Source
- John Bollinger's writings; widespread crypto retail use
- p-value
- 1.000
- Corrected q
- 1.000
- Receipt
- On-site receipt
- Detrended price rebound
US stocks · −16.8 pts/yr · 638 trades · corrected q 1.000
Source, p-value and result file
- Source
- none published (Tickfloor research-desk test)
- p-value
- 1.000
- Corrected q
- 1.000
- Receipt
- On-site receipt
- DeMarker exhaustion
US stocks · −17.1 pts/yr · 201 trades · corrected q 1.000
Source, p-value and result file
- Source
- none published (Tickfloor research-desk test)
- p-value
- 1.000
- Corrected q
- 1.000
- Receipt
- On-site receipt
- Ultimate oscillator dip
US stocks · −17.3 pts/yr · 86 trades · corrected q 1.000
Source, p-value and result file
- Source
- none published (Tickfloor research-desk test)
- p-value
- 1.000
- Corrected q
- 1.000
- Receipt
- On-site receipt
- Stochastic oversold cross
US stocks · −17.5 pts/yr · 252 trades · corrected q 1.000
Full test: Stochastic, CCI and Williams %R
Source, p-value and result file
- Source
- none published (Tickfloor research-desk test)
- p-value
- 1.000
- Corrected q
- 1.000
- Receipt
- On-site receipt
- Money flow index exhaustion
US stocks · −17.7 pts/yr · 28 trades · corrected q 1.000
Source, p-value and result file
- Source
- none published (Tickfloor research-desk test)
- p-value
- 1.000
- Corrected q
- 1.000
- Receipt
- On-site receipt
- Wide down-bar recovery
US stocks · −17.8 pts/yr · 46 trades · corrected q 1.000
Source, p-value and result file
- Source
- none published (Tickfloor research-desk test)
- p-value
- 1.000
- Corrected q
- 1.000
- Receipt
- On-site receipt
- Low-volume trend pullback
US stocks · −18.1 pts/yr · 9 trades · corrected q 1.000
Source, p-value and result file
- Source
- none published (Tickfloor research-desk test)
- p-value
- 1.000
- Corrected q
- 1.000
- Receipt
- On-site receipt
- Three-bar island reversal
US stocks · −18.1 pts/yr · 37 trades · corrected q 1.000
Source, p-value and result file
- Source
- none published (Tickfloor research-desk test)
- p-value
- 1.000
- Corrected q
- 1.000
- Receipt
- On-site receipt
- Chande momentum dip
US stocks · −18.2 pts/yr · 68 trades · corrected q 1.000
Source, p-value and result file
- Source
- none published (Tickfloor research-desk test)
- p-value
- 1.000
- Corrected q
- 1.000
- Receipt
- On-site receipt
- Cross-sectional 1-day reversal, liquid large-cap stocks (wide US stocks)
US stocks · −29.4 pts/yr · 2,135 rebalances · corrected q 1.000
Source, p-value and result file
- Source
- Jegadeesh 1990; Lo & MacKinlay 1990 RFS (short-term reversal); universe = harness's US_STOCKS_DAILY 40 large caps (S&P100 not available)
- p-value
- 1.000
- Corrected q
- 1.000
- Receipt
- On-site receipt
- Cross-sectional 1-day reversal, liquid large-cap stocks
US stocks · −34.2 pts/yr · 2,135 rebalances · corrected q 1.000
Source, p-value and result file
- Source
- Jegadeesh 1990; Lo & MacKinlay 1990 RFS (short-term reversal); universe = harness's US_STOCKS_DAILY 40 large caps (S&P100 not available)
- p-value
- 1.000
- Corrected q
- 1.000
- Receipt
- On-site receipt
- Crypto Fear & Greed Index contrarian entries
crypto · −59.6 pts/yr · 72 trades · corrected q 1.000
Full test: Fear & Greed Timing
Source, p-value and result file
- Source
- alternative.me Fear & Greed Index
- p-value
- 1.000
- Corrected q
- 1.000
- Receipt
- On-site receipt
- Crypto 1-day reversal after a high-volume down day
crypto · −66.4 pts/yr · 4,969 trades · corrected q 1.000
Source, p-value and result file
- Source
- Nagel 2012 RFS (short-term reversal & liquidity provision), volume-conditioned
- p-value
- 1.000
- Corrected q
- 1.000
- Receipt
- On-site receipt
- Bollinger %B reversion on large-cap crypto
crypto · −81.2 pts/yr · 1,928 trades · corrected q 1.000
Full test: Bollinger Band Bounce
Source, p-value and result file
- Source
- John Bollinger's writings; widespread crypto retail use
- p-value
- 1.000
- Corrected q
- 1.000
- Receipt
- On-site receipt
How to read these results
"Vs benchmark" is the strategy's annualized return minus an equal-weight benchmark, rebalanced daily, on the same assets and days. Strategies are charged trading costs; the benchmark pays none. Ahead is not a confirmed edge: the p-value is the unadjusted chance of a gap this large by luck, and the corrected q allows for how many strategies were tried, because testing hundreds of ideas throws up lucky winners. Every strategy saw only data up to the same cutoff date. The receipt is the frozen result file for that rule.
"Sample" is trades for rules that enter and exit a position, and rebalances for portfolio rules. A rebalance is a scheduled date on which a portfolio rule re-ranks and held at least one position. It is the sample size for portfolio rules. A holdings change is a rebalance where what the portfolio held, or its weights, differed from the previous one. Only these cost money to trade. The two are different counts and are never swapped. Definitions.