RSI, Explained: What the Relative Strength Index Actually Measures
8 min read·Reviewed by the StockTools.ai Research Team
- ▸RSI compresses the last 14 periods of gains versus losses into a single 0-100 momentum score — above 50 means recent gains dominated, below 50 means losses did.
- ▸Readings above 70 or below 30 mean momentum is stretched, not that a reversal is due.
- ▸Strong trends can pin RSI in extreme territory for weeks, which is why overbought is not a sell signal on its own.
- ▸Divergence — price makes a new high while RSI makes a lower high — hints that momentum is fading, but it needs price confirmation before it means much.
- ▸Disciplined traders treat RSI as context; position sizing and a planned exit do the real risk work.
What RSI actually measures
The Relative Strength Index is a momentum gauge that answers one narrow question: over the last stretch of trading — 14 periods by default, usually 14 days — how big and how frequent were the up moves compared with the down moves? It compresses that answer into a single number between 0 and 100. When recent gains have dominated, RSI sits high. When recent losses have dominated, it sits low. When the two are roughly balanced, it hovers near 50.
The mechanics, sketched in words: take the last 14 periods, average the size of the gains on the up days, and separately average the size of the losses on the down days (losses counted as positive numbers, and days that moved the other way counted as zero in each average). Divide the average gain by the average loss to get a ratio called relative strength. Then map that ratio onto a 0-to-100 scale by taking 100 and subtracting 100 divided by one plus the ratio. The standard version also smooths each new average into the previous one so the line doesn’t jump around, but the intuition is unchanged: it’s a tug-of-war score between recent buying and recent selling pressure.
A quick worked example. Suppose that over the last 14 days a stock’s up days averaged out to a 1.2 percent daily gain across the window, while its down days averaged out to 0.6 percent. The ratio is 1.2 divided by 0.6, which is 2. RSI is then 100 minus 100 divided by 3, or roughly 67. Flip the numbers — losses twice the size of gains — and RSI lands near 33. Perfectly balanced gains and losses put it at exactly 50. That’s all the number is: a normalized summary of who has been winning lately.
The 70/30 convention — and what it really says
The classic reading, dating back to J. Welles Wilder’s 1978 book that introduced the indicator, is that RSI above 70 marks an overbought market and RSI below 30 marks an oversold one. Those levels caught on and are now the default lines drawn on nearly every charting platform.
It’s worth being precise about what those words claim. Overbought does not mean the stock is too expensive or that a fall is coming. It means recent trading has been unusually one-sided to the upside — the average up move has been several times the size of the average down move for a couple of weeks. Oversold is the mirror image. Both are descriptions of the recent past, not predictions about the future.
The convention earns its keep mostly in sideways, range-bound markets. When a stock is oscillating between well-worn levels, a push to RSI 75 often coincides with the top of the range, and a slide to 25 with the bottom, so stretched readings tend to relax back toward the middle. That’s the environment the rule of thumb was born in — and the environment where it quietly stops working is the subject of the next section.
The biggest failure mode: strong trends pin RSI
Picture a stock in a genuine, persistent uptrend: up around 1 percent on ten of the last fourteen days, with the four down days losing only about 0.3 percent each. Run the arithmetic and the average gain across the window is about 0.71 while the average loss is about 0.09 — a ratio near 8, which translates to an RSI near 89. As long as that daily pattern continues, the RSI stays parked in the high 80s. Every single day the indicator screams overbought, and every single day the stock closes higher.
This is not a rare glitch; it’s what RSI is mathematically guaranteed to do in a strong trend. The best-performing stocks in any given year routinely spend weeks — sometimes months — above 70, precisely because sustained above-average gains are what a strong trend is. A trader who sold every touch of 70 would have been repeatedly ejected from the strongest moves on the board, usually early. Symmetrically, a stock in a persistent decline can sit below 30 for a long time; oversold is not a floor, and falling knives frequently look cheap on RSI the whole way down.
The practical lesson is that an extreme RSI reading means very different things in different regimes. In a range, RSI 78 often marks exhaustion. In a fresh breakout backed by real buying, RSI 78 may simply confirm that the move has unusual force. The indicator can’t tell you which regime you’re in — and mechanically fading every extreme reading is one of the most common and expensive habits among newer traders.
Divergence, explained with numbers
Divergence is the observation that price and RSI have started disagreeing. The bearish version: price makes a new high, but RSI makes a lower high than it did on the previous rally. The bullish version is the mirror: price makes a new low, but RSI makes a higher low. In both cases the message is the same — the newest leg of the move had less punch than the one before it.
A mini-scenario makes it concrete. A stock rallies from 42 to 50 dollars in eight brisk sessions, and RSI peaks at 76. It pulls back to 46, then rallies again to 53 — a clearly higher price high. But this second leg takes thirteen sessions, the daily gains are smaller, and three down days are mixed in, so RSI only reaches 64. Price printed a new high; momentum did not. The averages inside the RSI calculation are recording that buyers needed more time and more effort to gain less ground.
What divergence does not do is time anything. Divergences can stack up two or three times in a row while a trend grinds on, and each one looks prophetic only in hindsight. Traders who use divergence seriously treat it as a yellow light rather than a trigger: it raises attention, and then they wait for price itself to confirm — for example, a break below the swing low at 46 in the scenario above — before treating the momentum fade as meaningful.
How disciplined traders actually use RSI
The common thread among experienced users is that RSI is context, never a standalone trigger. A typical adaptation is to let the trend reset the levels: in an established uptrend, pullbacks often bottom with RSI in the 40-to-50 zone rather than at 30, so that zone becomes the area of interest for adding to or entering a position idea. In a downtrend, weak rallies often stall with RSI near 50 to 60. The trend defines what normal looks like; RSI just measures the stretch around it.
RSI also gets weighed against independent evidence rather than read in isolation — where price sits relative to obvious support and resistance, what volume is doing, how the broader market is behaving. In that role it can strengthen or weaken a thesis the trader already has. It does not generate the thesis, and a stretched reading that conflicts with everything else on the chart is usually just noise.
Most importantly, the survival math happens elsewhere. Whatever role RSI played in finding an entry, the questions that determine long-run outcomes are how much is at risk if the idea is wrong, where the exit is, and whether the potential reward justifies that risk. Sizing a position so the planned stop is an affordable loss, and checking the reward-to-risk ratio before entering, matter far more than whether RSI read 28 or 32 at the time. An indicator can inform a decision; it can’t manage the risk of one.
Honest limits
RSI is computed entirely from past prices, so it contains no information that isn’t already on the chart. It’s a restatement — a genuinely convenient one — of what recent candles already show. Anyone claiming RSI predicts the future is claiming that the last 14 days of price action predict the future, which is a much harder claim than it sounds.
Its parameters are conventions, not laws. The 14-period window and the 70/30 lines were calibrated on daily commodity charts in the 1970s. Shorten the window and the line gets jumpy and generates far more signals; lengthen it and signals nearly vanish. Because every settings change reshuffles the signals, it’s easy to tune RSI until it looks brilliant on past data — and that curve-fit brilliance rarely survives contact with the next year of trading.
Finally, its usefulness depends on the market regime, and the regime is only obvious in hindsight. Mean-reversion readings work until a range breaks into a trend; trend-following interpretations work until the trend dies. None of this makes RSI useless. As a compact, standardized momentum thermometer it’s a fine tool for describing what just happened. It’s a thermometer, though — not a forecast — and this article is education, not a recommendation to buy or sell anything.
FAQ
Is RSI above 70 a sell signal?
Not by itself. Above 70 means recent gains have been unusually one-sided, and in strong uptrends RSI can stay above 70 for weeks while the stock keeps rising. Traders who use it well combine the reading with the trend, key price levels, and a risk plan rather than acting on the number alone.
Why 14 periods, and can the setting be changed?
Fourteen is simply the default Wilder published in 1978, and platforms kept it. Shorter settings like 7 react faster but whipsaw far more; longer settings like 21 or 28 are smoother but nearly silent. There is no correct number — just a trade-off between sensitivity and noise, which is why tuning it to fit past data is a trap.
Does RSI work on intraday and weekly charts?
The math is identical on any timeframe — 14 five-minute bars, 14 days, or 14 weeks. The interpretation shifts, though: very short timeframes are dominated by noise, so extreme readings appear constantly and mean little, while weekly readings move slowly and describe longer regimes. Whatever timeframe is used, the trend-pinning failure mode applies there too.
Is RSI the same thing as relative strength versus the market?
No, and the naming collision confuses a lot of people. RSI compares a stock’s own recent gains with its own recent losses. Relative strength (as in relative strength ratings or RS lines) compares a stock’s performance against an index or its peers. A stock can have a low RSI while still showing strong relative strength versus the market, and vice versa.
What is RSI divergence and how reliable is it?
Divergence is when price makes a new extreme but RSI makes a less extreme one — for example, a higher price high with a lower RSI high — suggesting the latest leg had less momentum. It’s a warning sign, not a timing tool: divergences can repeat several times before anything changes, so experienced traders wait for price confirmation, such as a break of a recent swing level.
What RSI level is best for buying a stock?
There isn’t one, and any specific number offered as a universal buy level should raise suspicion. The same reading means different things in a range versus a trend, and RSI says nothing about a company’s value or your risk. Position sizing and a defined exit matter far more than any entry threshold — and nothing here is a recommendation to buy or sell.
Put it to work
More to learn
Educational only — not financial advice. Concepts simplified for clarity; markets are messier than definitions.