RSI Indicator: What Relative Strength Index Actually Measures
Ask ten retail traders what the RSI indicator does and most will recite the same folk rule: below 30 buy, above 70 sell. That rule is the reason this chapter exists, because it is not what the relative strength index measures. RSI is a thermometer for recent momentum. It tells you how one-sided the last fourteen sessions have been — and nothing about whether the move was deserved.
This is the RSI chapter of Gale’s technical analysis guide. It explains what the RSI indicator in the stock market actually computes, walks through the RSI indicator formula in plain words, and then spends most of its length on the part that trips people up: how to read 30 and 70 as descriptions rather than instructions.
What is the RSI indicator in the stock market?
The relative strength index is a momentum oscillator — a single line, plotted between 0 and 100 in a panel below the price chart. It is built from closing prices alone. Each day the stock either closed higher than yesterday or lower; RSI compares the average size of the recent up-days with the average size of the recent down-days and compresses that comparison into one bounded number.
The name misleads. Despite the words “relative strength”, RSI does not compare a stock with the Nifty or with its sector peers — that comparison is a different tool with a confusingly similar name, covered in the FAQ below. RSI is entirely self-referential: HDFC Bank’s RSI knows only HDFC Bank’s own recent closes.
It follows that RSI also knows nothing about earnings, order books, pledged shares or valuation. A stock repricing downward after a bad quarter and a sound stock caught in an index-wide fall can print identical RSI values. The number describes the selling; it cannot explain it. If terms like oscillator, close or session are unfamiliar, the stock market terminology glossary defines them, and the opening chapter on technical analysis of stocks sets out what price-based tools can and cannot do.
The RSI indicator formula in plain words
J. Welles Wilder Jr., a mechanical engineer turned trader, published RSI in 1978 with a 14-period lookback that has stayed the default ever since. The construction takes four steps, none of which needs algebra:
- Take the last 14 closes and write down each day’s change from the previous close.
- Sort the changes into two columns — gains on up-days, losses on down-days — and average each column.
- Divide the average gain by the average loss. Wilder called this ratio RS, for relative strength.
- Squash RS onto a 0–100 scale: RSI = 100 − 100 ÷ (1 + RS).
The last step is the only non-obvious one, and the intuition is simple. A raw ratio is awkward to chart — it sits at 1 when buyers and sellers are balanced but can run toward infinity when losses shrink. The squashing formula maps that unbounded ratio onto a fixed scale where balance lands exactly at 50, an all-gain fortnight approaches 100, and an all-loss fortnight approaches 0.
As an illustration only, not a reading from any live chart: if the average up-day was ₹3 and the average down-day ₹1, RS is 3 and RSI works out to 100 − 100 ÷ 4 = 75. Three-to-one dominance by buyers prints as 75. Equal averages would print as 50.
One refinement matters in practice. After the first 14 periods, Wilder did not recompute the averages from scratch each day. Each new session keeps roughly thirteen-fourteenths of the old average and blends in one-fourteenth of the new change. The indicator therefore has a fading memory: one wild session nudges the line rather than flipping it, and two platforms can disagree by a point or two if one uses simple averages instead. “14 periods” also means 14 of whatever bar the chart uses — fourteen days on a daily chart, fourteen weeks on a weekly one — so the same stock legitimately carries different RSI values on different timeframes.
The 30/70 bands: oversold describes, it does not instruct
Wilder drew two reference lines. Below 30, recent losses have dominated recent gains heavily enough to be unusual, and the condition is labelled oversold. Above 70, gains have dominated to the same degree: overbought. Both words are descriptions of the recent past. Neither is a forecast, and neither is an instruction.
The trap is treating oversold as a synonym for cheap. A reading of 24 says sellers have controlled the last fourteen sessions; it does not say they are finished. A stock can stay oversold for weeks. During the 2018–19 NBFC credit squeeze, daily RSI on several financiers sat below 30 week after week while prices kept halving — the indicator was accurately describing relentless selling, not flagging a bargain. Anyone who bought the first touch of 30 learned the difference between a description and a signal at some expense.
The persistent-oversold sequence, drawn above as five full-bodied red candles stepping steadily lower. What it looks like: consecutive bearish candles with substantial bodies, each close below the last, while RSI holds under 30 the whole way. Where it appears: genuine repricings — downgrades, credit stress, governance shocks — and prolonged bear phases. What it suggests: selling pressure that is sustained, not exhausted; the low reading is confirming the trend, not contradicting it. Failure mode for the bargain-hunter’s reading: assuming the elastic must snap back; in a repricing there is no rule that says it must, and averaging down into one has ruined more portfolios than any indicator has saved.
The bands earn their keep as a filter, not a trigger. A reading below 30 is a reason to open the annual report and the latest exchange filings, and to ask why the selling has been this one-sided. Gale publishes today’s RSI ≤ 30 list, computed from exchange data for exactly that use — a research starting point in which some names will be temporarily out of favour and some will be falling for excellent reasons. The screen does not distinguish between them; that is the reader’s job. Gale computes these indicators daily from exchange-published data — the live screens are this chapter applied to the whole market, refreshed each trading day.
| RSI zone | Conventional label | What it actually describes |
|---|---|---|
| Above 70 | Overbought | Recent gains have dominated losses to an unusual degree |
| 50–70 | Bullish side | Up-days are outweighing down-days |
| Near 50 | Balance | Average gains and losses roughly cancel |
| 30–50 | Bearish side | Down-days are outweighing up-days |
| Below 30 | Oversold | Recent losses have dominated gains to an unusual degree |
Everything said about 30 applies in mirror image to 70. Strong stocks visit overbought territory repeatedly during sustained advances; a first touch of 70 in a fresh uptrend has historically marked strength at least as often as excess.
RSI divergence, honestly framed
Divergence is the observation people find most persuasive about RSI, so it deserves the most careful framing. A bullish divergence occurs when price makes a lower low but RSI makes a higher low: the second sell-off carried the stock to a new low, yet the fourteen-session arithmetic behind that low was less one-sided than before. A bearish divergence is the mirror — a higher price high on a lower RSI high.
The price leg of a bullish divergence, drawn above; the RSI line beneath such a chart would print a higher low at the second, deeper price low. What it looks like: a hard sell-off, a bounce, then a push to a lower low on visibly smaller bearish bodies — here with a long lower wick — before a green candle appears. Where it appears: late in extended declines, at retests of earlier lows, and near long-standing floors of the kind the support and resistance chapter maps. What it suggests: the second decline needed less force to reach a lower price — sellers may be tiring. Failure mode: divergence is a condition, not an event. Price can print a third and fourth lower low while RSI keeps diverging, and each new low resets the pattern. In hindsight the “real” divergence is always obvious; in real time, three analysts will draw it from three different swing points.
That last point is the honest core. Divergences are abundant in any long trend, most resolve in the trend’s favour, and the persuasive textbook examples are survivors selected after the outcome was known. The MACD chapter discusses the same trap with that indicator’s divergences, which suffer it equally.
RSI in trends versus ranges
The single biggest improvement in reading RSI comes from asking one prior question: is this stock trending or ranging? The bands behave differently in each regime.
In a sideways range, RSI does what the textbook promises. Price oscillates between a floor and a ceiling, RSI swings between its bands roughly in sympathy, and readings near 30 tend to occur near the floor where buying interest has repeatedly appeared. The oscillator and the chart structure tell one story.
In a trend, RSI shifts its operating range. Stocks in sustained uptrends often spend months oscillating between roughly 40 and 80, rarely touching 30 at all — pullbacks bottom out with RSI in the low 40s. Sustained downtrends mirror this between roughly 20 and 60. A fixed 30/70 reading therefore means different things in different regimes: RSI at 42 can be a routine pullback reading in an uptrend and a feeble bounce in a downtrend.
| Market condition | Typical RSI behaviour | Reading that carries weight |
|---|---|---|
| Sideways range | Full swings between the bands | Band touches near tested floors and ceilings |
| Established uptrend | Holds roughly 40–80; 30 rarely seen | Pullback lows in the low 40s; repeated 70s as persistence |
| Established downtrend | Holds roughly 20–60; 70 rarely seen | Bounce highs stalling in the high 50s |
Regime, in turn, is what the moving averages chapter measures — position relative to the 50- and 200-day averages is the quickest first pass. Participation matters too: a momentum reading built on thin trade says less than the same reading on heavy turnover, which is why the high-volume screen sits alongside the oversold list. And a stock pressing a long-tested ceiling on the near-breakout screen with RSI in the 60s is showing trend pressure, not an overbought extreme waiting to be faded.
For the price-action side of the same chart — what the individual candles are saying while RSI summarises them — the chapters on reading candlestick charts and candlestick patterns complete the picture.
FAQ
Is RSI below 30 a buy signal?
No. It describes fourteen sessions in which selling dominated; it says nothing about the fifteenth. Stocks in genuine trouble can stay below 30 for weeks while falling further. A low reading is a reason to investigate — starting with why the selling happened — not a reason to act.
What is the difference between the relative strength index and relative strength?
RSI compares a stock’s own recent gains with its own recent losses. Relative strength (no “index”) compares one stock’s performance against a benchmark or peer — Infosys against the Nifty IT index, say. A stock can have a high RSI while showing poor relative strength, and vice versa.
Why 14 periods?
It was Wilder’s choice in 1978 — he reasoned from a roughly 28-day trading cycle and took half — and convention has preserved it. Shorter settings swing harder and touch the bands more often; longer settings smooth and lag. No setting is “correct”; each answers a slightly different question about how much recent history to weigh.
Does RSI work on weekly charts and on indices?
The arithmetic works on any consistent series of closes — weekly bars, the Nifty, a commodity. What changes is meaning: a weekly RSI below 30 describes fourteen weeks of dominant selling, a far rarer and heavier condition than the same reading on a daily chart. Comparing readings across timeframes as if they were interchangeable is a common error.
Can RSI be combined with other indicators?
It usually is — but combining RSI with another price-derived oscillator adds less than it appears to, since both are computed from the same closes. Pairings that add genuinely new information come from a different dimension: volume, chart structure, or the trend measures covered in the moving averages chapter.
What to weigh
RSI compresses fourteen sessions of closing prices into one number between 0 and 100. That is both its usefulness and its whole content. It will tell you, faster than scanning the chart, that selling has been unusually one-sided; it will never tell you whether that selling was panic or accuracy. The readings that deserve attention are the ones that arrive with context — a low reading at a floor that has held for years, in a stock whose filings show nothing broken, reads differently from the same number in a leveraged business losing its lenders. Weigh the regime before the reading, the turnover behind the reading, and above all the reason for the move, which lives in disclosures rather than derivations of price. The bands describe; the homework decides.
Gale is not a SEBI-registered investment adviser or research analyst; this chapter is education about how an indicator is built and read, not a recommendation to buy or sell any security.