Larry Connors profile picture

Larry Connors

RSI(2) mean reversion

Verified1DRecommended Channel

A short-term mean-reversion system: while price is above its 200-day average, buy extreme two-day weakness (RSI(2) under 5) and exit on the first close back above the 5-day average.

Win rate

73.3%

Share of trades that closed in profit

Trades

30

Completed trades in the test

Profit factor

2.07

Above 1 means profit beat loss

Return

42.1%

Account change over the window · buy & hold 963.8%

Max drawdown

20.4%

Worst fall from a peak

Most-tested backtest · BTCUSDT 1D2018-09-022026-08-31 · 8.0 years

Step by step, exactly as executed

1

Timeframe

1D

2

Setup / filter

close > SMA(200)

3

Entry

RSI(2) < 5 → market at the next candle open

4

Stop

NONE — no stop-loss rule was provided, so no stop is placed

5

Target / exit

Target: no fixed target · Rule exit: a close above SMA(5) closes the trade at the next open

6

Risk

1% of equity as trade notional (there is no stop, so this is not risk to a stop)

5

Executable rules

2

Published sources

4

Known limitations

The published rules, verbatim

1

Timeframe

daily.

2

Trend filter

the close is above the 200-day simple moving average.

3

Entry

RSI(2) closes below 5.

4

Exit

the close crosses back above the 5-day simple moving average.

5

No stop loss

the published rules exit on the moving-average signal only.

Sources

What this test cannot reproduce

Short trades (the mirror rules below the 200-day average) cannot be executed on Spot, so only the long side is tested.

The published rules use NO stop loss, so R-based statistics are undefined and the position size is a fixed notional, not risk to a stop.

Connors published this for US equities and ETFs; this test runs on the crypto pair you named.

Connors reports RSI(2) thresholds of 10, 5 and 2; the book's core table uses 5, which is what is tested.

Portfolio — every backtest of this strategy

Each row prints the exact window it was measured on. Rows with different windows measure different market periods, so their numbers differ without contradicting each other.

Sign in to run these rules yourself and see your own results here.