Short answer: the rsi indicator, short for relative strength index, is a momentum oscillator that compares the average size of recent gains with the average size of recent losses and scales the result between zero and one hundred. Readings above seventy are conventionally called overbought and below thirty oversold. It measures the speed of recent price movements, not their direction, and that distinction decides how it should be used.
What the calculation actually measures
J. Welles Wilder introduced the indicator in 1978 alongside several other tools that are still standard on trading platforms today. The default is a 14-period RSI, and the arithmetic runs in three steps.
First, separate each period's price change into a gain or a loss. Second, average the gains and average the losses over the lookback, using a smoothing method that lets older data fade rather than drop out abruptly. Third, express the ratio between them on a zero to one hundred scale.
The consequence worth understanding is what a reading of eighty actually says. It does not say the instrument is expensive. It says that over the last fourteen periods, the average up move has been substantially larger than the average down move. In a strong trend that condition can persist for weeks, which is exactly why the overbought label misleads so many traders.
| Reading | Conventional label | What it actually indicates |
|---|---|---|
| Above 70 | Overbought | Recent gains have dominated recent losses |
| 50 to 70 | Bullish territory | Upside momentum with room left |
| Around 50 | Neutral | Gains and losses roughly balanced |
| 30 to 50 | Bearish territory | Downside momentum with room left |
| Below 30 | Oversold | Recent losses have dominated recent gains |
Reading the four signals
Extremes. The classic reading, and the most abused. In a range-bound market extremes mark the edges reliably. In a trending market the indicator pins near one extreme and stays there, and selling every overbought print is a steady way to lose money.
The fifty line. Underused and more robust. RSI holding above fifty on pullbacks describes an intact uptrend, and rejection from below fifty describes an intact downtrend. This reading works in exactly the conditions where the extremes reading fails.
Failure swings. A move to an extreme, a pullback, then a second attempt that fails to reach the previous extreme before breaking the intervening low. This is a structural signal within the oscillator itself and does not require price confirmation to identify, though it still requires it to trade.
Divergence. Price makes a new extreme and the indicator does not. Our guide to RSI divergence covers the four types and the confirmation sequence in full, because divergence is where most of the technique's difficulty lives.
Adjusting settings to the instrument
Fourteen periods is a convention from an era of hand-drawn charts, not an optimum. Shortening it to seven or nine produces more signals with more noise, which suits shorter holding periods. Lengthening it to twenty one produces fewer, steadier readings.
The more useful adjustment is usually the thresholds rather than the lookback. In a persistently trending instrument, shifting the bands to eighty and twenty cuts a large share of premature countertrend signals. In a range-bound instrument, seventy and thirty may still be too wide and sixty five and thirty five works better.
The mistake is adjusting after a losing trade, which guarantees you always run the configuration that just failed. Choose once based on how the instrument behaves, then leave it long enough to gather a meaningful sample.
A framework for using it in a trading strategy
- Classify the environment first. Trend or range, decided on the timeframe above the one you trade. Every subsequent reading depends on this answer.
- In a range, fade the extremes. Enter as price turns back from an extreme, with the stop beyond the range boundary rather than at the indicator level.
- In a trend, buy the shallow dip. An oversold print during a pullback within an uptrend is a continuation signal, not a reversal one. This inversion is the single most valuable habit with this tool.
- Use the fifty line as the trend filter. Take long setups only while RSI holds above fifty, and short setups only below it.
- Require price confirmation. The indicator never tells you where a trade is wrong. Stops come from swing points or a volatility measure such as the ATR indicator.
- Cap attempts. Extremes cluster. Decide before the session how many countertrend entries you will take.
Combining it with other tools
Momentum plus direction is the useful pairing. RSI describes force, and something else has to describe direction and location. A moving average or market structure supplies direction. A support and resistance level supplies location. A volatility measure supplies stop distance.
What does not help is stacking a second oscillator. RSI, stochastic and the MACD histogram are all derived from recent price and largely agree, so running all three produces confidence rather than information. Our guide to the best indicators for day trading covers how to combine families that actually measure different things.
Free long price histories make settings testing practical. The S&P 500 daily series published by the St. Louis Fed provides decades of data for checking whether a threshold change helps or merely fits history.
How it behaves across different markets
The maths is identical everywhere, but the reading is not, because what counts as an extreme depends on how the instrument moves.
Currency pairs trend in policy-driven legs and range for long stretches in between. The fifty line filter earns its place here more than the extremes do, and the same pair can need different thresholds in a trending quarter than in a quiet one.
Index products smooth away single-name noise, which produces cleaner oscillator readings than individual instruments. Extremes on a major index are less frequent and correspondingly more meaningful.
Metals produce long directional runs punctuated by sharp reversals. RSI can hold above seventy for weeks in a gold advance, which is precisely the environment where the textbook reading destroys accounts.
Cryptocurrencies move far enough that the indicator pins at an extreme routinely. Wider thresholds are not optional here, and a volatility measure alongside is close to mandatory for sizing.
The unifying point is that the number on the scale means something different per instrument. Before trusting a threshold, check how often that instrument actually reaches it over the past year.
Limitations and common pitfalls
- Treating extremes as signals. In trends they mark strength, not exhaustion.
- Comparing readings across instruments. The scale is bounded, but what counts as an extreme differs by how the instrument moves.
- Trading divergence without a structural break. Momentum can weaken for many periods while price continues.
- Optimising the lookback against history. Testing fifty settings and keeping the best fits the past rather than describing behaviour.
- Forgetting it lags. Every input is an average of completed periods.
- Ignoring transaction cost. Shorter settings mean more trades, and spread plus commission charges accumulate faster than most traders expect.
The honest summary is that the indicator is a good description of one narrow thing and a poor substitute for a plan. It has no concept of where a trade is invalidated, which is the decision that determines results.
Where evaluation rules change the reading
Traders working through a prop firm evaluation face a fixed daily loss limit, which changes the calculus on countertrend entries specifically. A run of premature fades in a trending market can end an attempt before a good setup appears. Pivex publishes its daily and total loss parameters before an attempt begins, and the practical response is to run the trend-aligned version of the tool: use the fifty line as a filter and take oversold prints only in the direction the higher timeframe supports.
Mean reversion approaches lean on this tool more heavily than trend approaches do, and our guide to mean reversion trading sets out when that premise holds and when it breaks.
A worked read across one cycle
Following a full cycle makes the tool concrete in a way that rules do not.
Start with a market that has been declining for several weeks. RSI sits below thirty and keeps printing new lows alongside price. Every one of those oversold readings is a continuation signal, not a bottom, because the environment is a trend and the trend is intact. Traders who buy each oversold print here lose steadily.
Price then stabilises. RSI lifts off the low but price makes one more marginal low. The indicator does not follow, which is bullish divergence. Nothing is tradeable yet, because momentum can decelerate for weeks without price turning.
Next, RSI crosses back above fifty. This is the reading that matters most and gets the least attention. Crossing the midline means the average gain over the lookback has caught up with the average loss, which is a genuine change in the balance rather than a change in speed. Combined with a break of the most recent swing high in price, it is a structural signal.
The market then trends higher. RSI spends weeks between fifty and eighty, pulling back toward fifty on dips and never reaching thirty. During this phase the correct use is exactly the opposite of the textbook: buy pullbacks toward fifty, and treat readings above seventy as evidence of strength.
Finally the trend matures. RSI makes a lower high while price makes a higher high, the pullbacks start breaking below fifty, and the environment shifts back toward a range. The same tool, read the same way throughout, would have been wrong in three of these five phases. Reading it against the environment is what makes it work.
Next step: take one instrument, mark every RSI reading above seventy over the past six months, and split them by whether the market was trending or ranging at the time. The difference in what happened next is the whole lesson, and it costs nothing but an hour. Pivex publishes its rule set before an attempt begins, so the sample you build can be sized against limits you already know.
Related guides: mean reversion trading and RSI divergence.
Before participating, review Pivex's current program rules and decide whether a simulated proprietary trading evaluation fits your experience and risk limits. Participation does not guarantee a reward, income, or any particular trading result.