Rsi Strategy

What it is

Buying oversold dips and selling overbought spikes using RSI(14).

RSI, published by J. Welles Wilder in 1978, measures the ratio of average gains to average losses over a lookback window, normally 14 periods, and scales it from 0 to 100. Mean-reversion traders use it as a stretch gauge: a reading under 30 says recent selling has been unusually one-sided, which in a range-bound market often precedes a bounce. Crucially, RSI says nothing about direction — it describes how price got here, not where it goes next.

Market conditions that matter

The method only makes sense in a market that is oscillating around a value area: a large-cap stock consolidating after a run, a major FX pair inside a monthly range, BTC chopping between well-defined levels. In a strong trend RSI can hold above 70 for weeks, and every 'overbought' short is a loss. Check the 50 and 200 EMA relationship first: if they are widely separated and sloping, this is a trend regime and mean reversion should be skipped.

Best suited to

Range-bound markets, large-cap stocks, and major crypto pairs.

Badly suited to

Strong trends — RSI stays overbought/oversold for weeks and you'll get run over.

The steps

• Add RSI(14) to your chart.
• Wait for RSI < 30 (oversold) or > 70 (overbought).
• Confirm with a candlestick reversal pattern on the same bar.
• Enter, stop just beyond the reversal candle, target the 20 EMA.
• Skip the trade in obvious strong-trend regimes — check the 50/200 EMA first.

Worked example

BTC RSI dips to 26 at $92K with a bullish engulfing — buy, stop $91K, target $94K.

The numbers behind it

In range conditions a 57% win rate at 1:1.2 reward-to-risk gives (0.57 x 1.2) - (0.43 x 1) = +0.25R. Apply the same rules to trending conditions at a 32% win rate and expectancy falls to (0.32 x 1.2) - (0.68 x 1) = -0.30R. The identical setup is profitable in one regime and clearly negative in the other, which is why regime classification — not indicator settings — is where the work belongs.

How it fails

• Taking the signal without confirmation. RSI under 30 alone is not an entry; wait for the reversal candle or a reclaim of a prior level so there is a defined place to be wrong.
• Shorting an overbought reading in an uptrend. This is the single most expensive misuse of the indicator, and it feels most compelling exactly when it is most dangerous.
• Tuning the lookback until history looks profitable. An RSI(9) that backtests beautifully on one asset and one year is usually curve-fitting, not an edge.

Practising it safely

Run this method for at least thirty simulated trades with fixed sizing before judging it, and record every trade in the journal. A handful of winners proves nothing. Educational simulation only — not financial advice.