RSI 30/70 on crypto: does it work? A 10-coin backtest

The answer first: RSI 30/70 is not a broken rule, it is a slow one. On daily candles it made money on 8 of 10 large coins, with two trades out of three closing in profit — and still finished far behind simply buying and holding on 8 of them. On 4-hour candles it made money on only 4 of 10. Against random entry dates with the same holding time it looks no better than chance: only 7 of 20 runs did better than the median random run, and none made the top 10%.

Setup

• Rule, as in the textbook, untuned: RSI with period 14 on the run's timeframe. Buy when RSI crosses back above 30 (it was at or below 30 on the previous candle and is above it now). Sell when RSI crosses above 70. No stop-loss, no take-profit, no time limit — a trade stays open until RSI reaches 70. One position at a time, long only.
• Coins: BTC, ETH, BNB, XRP, ADA, DOGE, LTC, LINK, BCH and TRX — ten large coins that have Binance spot history on the platform from the start of 2020.
• Market and period: Binance spot, 4h and 1d candles, 2020-01-01 → 2026-10-01. 20 coin × timeframe runs.
• Costs and fills: 95% of equity per trade, 0.1% fee and 0.05% slippage on each side, orders filled at the next candle's open after the signal, 10,000 USDT per run.
• Two baselines: buy and hold of the same coin over the same run, as the platform reports it; and a random-entry baseline computed from candles — 2,000 simulated runs per cell with the same number of trades and the same average holding time as the rule, entered at random dates, with the same size and costs.

How such a test is set up in general is in the guide how to backtest a crypto strategy; the metrics on the cards are explained in backtest metrics explained.

Results: profit, but far behind holding

The rule is right more often than not — about two of three trades closed in profit on both timeframes — and that is exactly what the 30/70 levels promise: buy a dip, sell a bounce. The trouble is the trades that go wrong. With no stop, a trade bought on the way down stays open until RSI finally reaches 70, which in a falling market can take months. And in a rising market the rule is mostly out: RSI rarely drops below 30 during a bull run, so on BTC much of the 2020–2021 and 2024 rallies happened without it.

Backtest

RSI 30/70 reversal (RSI 14) — article study

BTCUSDT · 1d · Spot · 2020-01-01 → 2026-10-01

Return

+68.14%

Max drawdown

48.05%

Sharpe

0.40

Win rate

70.0%

Trades

10

Open the report →

On BTC daily the rule traded 10 times in almost seven years. Between its exit on 2020-04-30 and its next entry on 2021-05-21 it held nothing, while BTC's daily close went from 8,620 to above 60,000 USDT and back to about 40,000. Its worst trade was bought on 2022-05-15 and held until 2023-01-12 — through the rest of the 2022 bear market and the FTX collapse — for -40.58%.

Backtest

RSI 30/70 reversal (RSI 14) — article study

BTCUSDT · 4h · Spot · 2020-01-01 → 2026-10-01

Return

+107.33%

Max drawdown

70.52%

Sharpe

0.48

Win rate

72.5%

Trades

69

Open the report →

On 4-hour candles BTC traded 69 times and did better than on daily ones; on ETH it was the reverse — -10.35% daily (run) and -61.35% on 4h (run), while ETH itself went up +1,969.60%.

The best and the worst cell

Backtest

RSI 30/70 reversal (RSI 14) — article study

LINKUSDT · 1d · Spot · 2020-01-01 → 2026-10-01

Return

+291.31%

Max drawdown

75.89%

Sharpe

0.66

Win rate

66.7%

Trades

9

Open the report →

Backtest

RSI 30/70 reversal (RSI 14) — article study

ADAUSDT · 4h · Spot · 2020-01-01 → 2026-10-01

Return

-88.74%

Max drawdown

94.26%

Sharpe

-0.25

Win rate

48.0%

Trades

50

Open the report →

Even the best of the 20 runs stayed behind holding the same coin. The two runs that did beat holding — LTC and BCH on daily candles, +142.23% and +36.82% — did it on the two coins that went almost nowhere over the period (+17.13% and -9.03% for holding).

Was it the signal or just the market?

A long-only rule can make money simply because the coin went up while it was in a trade. The random-entry baseline takes that away: for each run, 2,000 simulated runs with the same number of trades and the same average holding time, but entries at random dates.

If the RSI signal added nothing, about half of the 20 runs would end above their random median and about two would land in each 10% tail. The rule ended above the median in 7 of 20 runs — 5 of 10 on daily candles, 2 of 10 on 4-hour ones. None reached the top 10% of its random runs; three fell into the bottom 10%, all on 4-hour candles: ETH, ADA and DOGE. The median run beat 37.0% of its random runs. The best cases, LTC and LINK on daily candles, beat 84.8% and 79.9% of theirs — on 9 trades each. One simplification: every random trade is held for exactly the rule's average time, while the rule's own trades last from days to many months.

All 20 runs:

Each return links to its run where it was saved; the others are re-runs of the same strategy that were not stored — the saved runs reproduce them to the cent. "Random runs beaten" is the share of the 2,000 random-entry runs that ended below the rule.

Conclusion

• The textbook RSI 30/70 rule made money on most large coins on daily candles — 8 of 10 — and two of three of its trades were winners.
• It beat buy and hold in only 2 of 20 runs, both on coins that went nowhere: it sits out bull runs and, without a stop, holds losers for months.
• Against random entry dates with the same holding time it was no better than chance — above the random median in 7 of 20 runs, and below 90% of its random runs in 3.

The same test for two other classic signals: MACD and Bollinger Bands on crypto. A trend rule on the same kind of coins: the golden cross on the top-20.