Tickfloor

RSI(2) Mean Reversion: does it work?

Last updated

In-sample, not a promise. Every figure we measured here comes from a backtest on past data.

Did RSI(2) Mean Reversion beat the benchmark in Tickfloor's test?

No. We tested RSI(2) Mean Reversion on 40 US stocks, 1 crypto pair, 2016 to 2025. After costs, it returned 14.9 to 43.2 percentage points a year less than the benchmark. None held up once we allowed for the many other rules we tested. This is a historical test of the rule as described, not a prediction or advice.

Connors' RSI(2) rule, buy when the 2-period RSI drops to an extreme low and sell on the bounce, is repeated across trading blogs as a proven pullback edge.

The rule we tested. The 2‑period RSI dropping to an extreme oversold level to enter long, exiting on a bounce back above the midline, Connors' published version of the rule, not a looser or tighter threshold.

Receipt: parameters, data, window and result file

Connors RSI(2) mean reversion

Run id
trader-books-connors-rsi2, run 15 September 2026
Frozen parameters
timeframe 1d; not recorded
Data source
vendor not recorded in the result file; data fingerprint sha256 fb40d60216b63ab2...
Instruments and coverage
40 assets: AAPL, NVDA, AMZN, GOOGL, META, TSLA, JPM, V, UNH, XOM, JNJ, COST and 28 more (full list in the result file)
Date window
15 September 2016 to 14 March 2025 (discovery cutoff 15 March 2025)
Sample
5,937 completed trades over 15 September 2016 to 14 March 2025 (8.5 years)
Cost rule
0.05% per side; charged on every change in position/weight; benchmark pays none; no funding/borrow on shorts
Fill rule
not recorded
Benchmark
equal-weight long of the same assets, rebalanced daily across assets trading that day (no costs)
Run record (public)
research/hunt2/ledger.jsonl, line with id trader-books-connors-rsi2
Full result file
not published yet

Connors RSI(2) pullback on BTC daily

Run id
rsi2_btc, run 6 October 2026
Frozen parameters
timeframe 1d; rsiPeriod=2, entryBelow=10, trendWindow=200, exitWindow=5
Data source
vendor not recorded in the result file; data fingerprint sha256 56e180ab31f32f27...
Instruments and coverage
1 asset: BTCUSDT
Date window
18 August 2017 to 14 March 2025 (discovery cutoff 15 March 2025)
Sample
60 completed trades over 18 August 2017 to 14 March 2025 (7.6 years)
Cost rule
0.10% per side; charged on every change in position/weight; benchmark pays none; no funding/borrow on shorts
Fill rule
not recorded
Benchmark
equal-weight long of the same assets, rebalanced daily across assets trading that day (no costs)
Run record (public)
research/hunt2/ledger.jsonl, line with id rsi2_btc
Full result file
not published yet

Connors cumulative RSI(2): buy when 2-day sum of RSI(2) < 35 above SMA200, exit on close > SMA5, US stocks 1d x40

Run id
t1010-cumrsi2-s, run 10 October 2026
Frozen parameters
timeframe 1d; rsiLen=2, sumDays=2, entry=35, trend=200, exitSma=5
Data source
vendor not recorded in the result file; data fingerprint sha256 f8bd56551e6502b2...
Instruments and coverage
40 assets: AAPL, NVDA, AMZN, GOOGL, META, TSLA, JPM, V, UNH, XOM, JNJ, COST and 28 more (full list in the result file). 9 corrupt vendor bar-day(s) across 9 assets were dropped, never repaired
Date window
3 July 2017 to 14 March 2025 (discovery cutoff 15 March 2025)
Sample
2,198 completed trades over 3 July 2017 to 14 March 2025 (7.7 years)
Cost rule
0.05% per side; charged on every change in position; benchmark pays none; no funding/borrow on shorts
Fill rule
not recorded
Benchmark
equal-weight long of the same assets, rebalanced daily across assets trading that day (no costs)
Run record (public)
research/hunt2/ledger.jsonl, line with id t1010-cumrsi2-s
Full result file
not published yet

No. None of the versions we tested passed. None held up once we allowed for how many other rules we tested. That covers the versions, costs and benchmark described here, not every version of the rule.

Tested a rule of your own? Check whether your backtest is just luck. It is free and needs no account.

This is one line in a bigger check. The research desk has run 1,346 strategies, most against an equal-weight benchmark of the same assets rebalanced daily that pays no costs (the strategies pay theirs). After correcting for how many we tried (the luck check, Benjamini‑Hochberg, 1,778 tests), 0 passed. That does not prove no strategy works, and it says nothing about versions we did not test.

Short answers: the verdict and the rule, as questions

Does the Connors RSI(2) strategy work?

No. None of the versions we tested passed. After costs: US stocks −16.6 pts/yr vs the benchmark; crypto pair −43.2 pts/yr vs the benchmark; US stocks −14.9 pts/yr vs the benchmark. None held up once we allowed for how many other rules we tested. That covers the versions, costs and benchmark described here, not every version of the rule.

What exact rule did Tickfloor test?

The 2-period RSI dropping to an extreme oversold level to enter long, exiting on a bounce back above the midline, Connors' published version of the rule, not a looser or tighter threshold.

Is this the same as a classic RSI 30/70 overbought or oversold rule?

No. Connors' RSI(2) uses a much shorter 2-period RSI with more extreme thresholds on a specific pullback-and-exit rule, and that exact version is what we tested. A slower 14-period RSI crossing 30/70 has no matching row in the registry yet.

Is this financial advice?

No. This measures a publicly claimed strategy rule, not a recommendation. General information only, not personal advice.

See the full numbers on the Trading books, Mean reversion, Popular indicator rules family pages, or the full method and every result.

Linked from this page

Other rules we've tested

Terms explained

Next steps

General information only. It doesn't consider your objectives, finances or needs. Tickfloor holds no financial services licence and never places trades.