What RSI actually measures

The Relative Strength Index (RSI) is a momentum oscillator developed by J. Welles Wilder. It does not measure valuation or “how cheap” a market is. Instead, it compares recent average gains to recent average losses over a lookback window, usually 14 periods.

A common formula is:

[ RSI = 100 - \frac{100}{1 + RS} ]

where:

[ RS = \frac{\text{Average Gain}}{\text{Average Loss}} ]

If average gains have dominated losses, RSI rises toward 100. If losses dominate, RSI falls toward 0. On a daily chart, RSI(14) is the standard default, but many traders also test 7, 9, 21, or 2 depending on the market and timeframe.

What “overbought” and “oversold” really mean

The classic interpretation is:

  • RSI above 70 = overbought
  • RSI below 30 = oversold

But these labels are often misunderstood. “Overbought” does not mean the price must fall immediately. It simply means recent upside momentum has been strong enough that the market may be stretched.

Likewise, “oversold” does not mean the market is guaranteed to bounce. A market can stay oversold for a long time in a downtrend.

A better mental model:

  • RSI is a momentum pressure gauge
  • Extreme readings show recent imbalance, not a reversal signal by themselves

Why the naive RSI strategy usually fails

The most common beginner strategy is:

  • Buy when RSI < 30
  • Sell when RSI > 70

This seems intuitive, but it often underperforms for several reasons.

1) Strong trends break mean-reversion assumptions

RSI threshold strategies implicitly assume prices revert to a mean quickly. That works better in range-bound markets than in trending ones.

In a strong uptrend, RSI can remain above 70 for many bars while price continues rising. If you short every overbought reading, you may repeatedly fight trend momentum.

In a strong downtrend, RSI can stay below 30 while price keeps falling. Buying every oversold reading can become a “catching a falling knife” problem.

2) Fixed thresholds ignore market regime

A 70/30 rule may be too strict for some assets and too loose for others. For example:

  • High-volatility crypto often spends more time in extreme RSI zones
  • Large-cap stocks may show cleaner mean reversion on daily bars
  • Forex pairs can behave differently across sessions and macro conditions

A threshold that works in one regime may fail in another. That is why robust testing across multiple periods and market conditions matters.

3) RSI is lagging, not predictive

RSI uses historical closes. It reacts after momentum has already changed. By the time RSI reaches an extreme, a lot of the move may already be done.

That means the signal often arrives late for reversal trading and can be especially weak if used alone.

4) Transaction costs and slippage matter

Naive RSI strategies often trade frequently. If your average edge per trade is small, commissions, bid-ask spread, and slippage can erase it.

This is especially important for lower timeframes. A strategy that looks profitable on paper can become negative after realistic execution assumptions.

Better ways to use RSI

RSI is usually more useful as a filter than as a standalone entry rule.

1) Use RSI with trend confirmation

Instead of blindly buying oversold conditions, require the broader trend to support the trade.

Examples:

  • Only buy RSI < 30 when price is above the 200-day moving average
  • Only short RSI > 70 when price is below the 200-day moving average
  • Use a trend filter such as higher highs/higher lows or a moving-average slope

This helps avoid fighting persistent trends.

2) Look for divergence, not just extremes

A bullish divergence occurs when price makes a lower low but RSI makes a higher low. This can suggest downside momentum is weakening.

A bearish divergence is the opposite: price makes a higher high while RSI makes a lower high.

Divergence is not magic, but it can be a more nuanced signal than a raw threshold. Still, it should be tested carefully because divergence can appear many times before price turns.

3) Adjust thresholds by regime

Some traders use more flexible bands such as:

  • 80/20 in strong trends
  • 60/40 in range-bound markets
  • Dynamic bands based on RSI’s own distribution over the last 100–252 bars

This is often more realistic than assuming 70/30 is universally optimal.

4) Combine RSI with price structure

A stronger setup might require:

  • RSI below 30
  • Price at a prior support level
  • A bullish reversal candle or breakout above a short-term high
  • A stop loss below the recent swing low

This turns RSI from a blunt trigger into one part of a complete trade plan.

How to test RSI properly

When backtesting RSI ideas, avoid common mistakes:

  • Use out-of-sample testing: don’t optimize on the full dataset
  • Test multiple assets and timeframes: one good chart is not enough
  • Include costs: commissions, spreads, and slippage
  • Check drawdowns: a strategy can have a good average return but unacceptable losses
  • Avoid overfitting: if RSI(13) with 31/69 thresholds looks perfect, it may just be curve-fit noise

A practical workflow is to test a simple RSI rule first, then add one filter at a time:

  1. Baseline RSI threshold strategy
  2. Add trend filter
  3. Add volatility filter
  4. Add exit logic
  5. Re-test out of sample

Tools like a visual backtesting platform can help you assemble these rules without coding and compare variants side by side.

A simple RSI idea that is more realistic

One example of a more robust setup is:

  • Timeframe: daily
  • RSI length: 14
  • Entry: RSI crosses above 30 after price is above the 200-day moving average
  • Exit: RSI crosses above 50, or price hits a trailing stop
  • Risk: 1% of account per trade, stop below recent swing low

This is still not guaranteed to work, but it is more defensible than “buy every RSI below 30.”

Key takeaway

RSI is best understood as a momentum indicator, not a standalone buy/sell signal. Overbought and oversold conditions can persist far longer than beginners expect, especially in trending markets. The naive RSI strategy usually fails because it ignores regime, lag, and trading costs.

If you want to use RSI systematically, test it with trend filters, realistic execution assumptions, and proper out-of-sample validation.

Educational content only, not financial advice.