This article explains what the Relative Strength Index measures and how to read it. It is educational and general, not personalised investment advice, and not a claim that any indicator predicts price. Nothing here is a recommendation to buy or sell any security.
The Relative Strength Index, or RSI, is one of the most widely cited indicators in markets and one of the most widely misread. Two confusions do most of the damage. The first is the name: RSI sounds like it measures a stock’s strength against the market, and it does not, it measures the internal speed and magnitude of a name’s own recent moves. The second is the famous 70/30 bands, which get treated as a buy-and-sell switch when they are nothing of the sort. Here is what the number is built from, what “overbought” and “oversold” honestly mean, and why RSI earns its keep only as one input among many.
What RSI actually measures
RSI was introduced by J. Welles Wilder in 1978, alongside the Average True Range, and it answers a narrow question: over a recent window, how much of a name’s movement has been up versus down? The construction is simple in spirit. Take a lookback window, Wilder used roughly 14 periods, and separate the moves into gains and losses. Average the gains, average the losses, take their ratio, and compress the result onto a bounded 0-100 scale. When recent moves have been almost all up, RSI pushes toward the high end; when they have been almost all down, it falls toward the low end; when up and down moves roughly balance, it sits near the middle.
The key word is internal. RSI is computed entirely from a single name’s own price history, it knows nothing about the index, the sector, or any other stock. It is a measure of a name’s recent momentum relative to itself: how fast and how one-sided its own moves have lately been. That is a useful thing to know. It is also a very different thing from what the name suggests.
The bands: “overbought” and “oversold”
Because RSI is bounded, it invites threshold lines, and two became conventional: 70 and 30. A reading above 70 is labelled “overbought,” a reading below 30 “oversold.” The labels are evocative and that is the problem, they sound like verdicts when they are only descriptions. “Overbought” does not mean “too high, sell.” It means, precisely, that recent up-moves have dominated recent down-moves by a wide margin. That is a statement about momentum, not about value, and emphatically not a trade trigger.
The thresholds are conventions, not laws. Some practitioners shift them, tighter bands in quiet conditions, wider ones in trends , precisely because a fixed 70/30 line behaves very differently depending on the regime the name is in.
The disambiguation that matters most
Here is the single most important point in this article, because it is the one that costs people the most: RSI is not relative strength. Despite the name, the Relative Strength Index does not compare a stock to the market or to other stocks at all. It is a self-referential momentum oscillator, a name measured against its own recent history.
What most people actually mean by “relative strength” , how a stock is performing versus the broad universe, the thing that sorts leaders from laggards, is a completely separate measure: the relative strength rating, which ranks a name’s return against every other name over a trailing window. A stock can be screaming higher on RSI (strong internal momentum) while still lagging the market on its RS rating, or sitting at a calm RSI while quietly outperforming everything around it. They answer different questions. Confusing the two, reaching for the oscillator when you wanted the cross-sectional ranking, is one of the most common errors in applied technical analysis, and it is worth getting straight before anything else.
The expensive trap: shorting strength
The most costly mistake RSI invites follows directly from misreading the bands. A name in a powerful uptrend will print “overbought” readings, above 70, and stay there. The naive response is to short it, on the logic that overbought must mean due for a fall. But strong trends sit overbought for a long time, by construction: a name that keeps making one-sided up-moves keeps producing high RSI readings, week after week. Selling into that because the oscillator is “high” is selling the strongest names precisely while they are strongest. The same applies in reverse, buying an “oversold” name in a relentless downtrend catches falling knives. A persistently extreme RSI is more often evidence of a durable trend than a signal it is about to break.
Divergence as a context cue
One genuinely useful pattern is divergence: when price makes a fresh high but RSI makes a lower high than it did at the previous peak, the new high was achieved on less internal momentum than the last one. That can hint that a move is tiring, the buying pressure behind each new high is fading even as the price grinds up. The honest framing is that divergence is a context cue, not a signal. It describes a weakening of momentum; it does not time a reversal, and trends frequently diverge for a long while before they actually turn. Treated as a piece of context to weigh, it is informative. Treated as an entry trigger, it is a reliable way to fight a trend too early.
Why RSI is one input, never a standalone trigger
Every limitation above points to the same conclusion: RSI is a useful descriptor of a name’s internal momentum and a poor standalone decision rule. It says nothing about the broader market regime, the name’s strength versus its peers, the quality of its chart structure, its liquidity, or whether the move is worth the risk. Each of those is a separate question with its own measure. The disciplined use of an oscillator like RSI is as one feature among many, folded into a broader picture rather than fired on alone, the same logic behind composite scoring, where no single indicator gets to decide.
How Shishin uses it
Shishin treats RSI the way the indicator deserves to be treated: qualitatively, as one contextual input, never as a trigger on its own. Its engines, the four guardians Genbu, Suzaku, Byakko, and Seiryū, each tuned to a different market regime, lean far more heavily on a name’s strength versus the universe and on the structure of its setup than on where an oscillator happens to sit. Crucially, Shishin never reads “overbought” as a reason to avoid a strong name: a breakout-oriented engine expects leadership names to carry elevated RSI, because that is what sustained strength looks like from the inside. RSI may add colour, a hint of fading momentum here, confirmation of a one-sided move there, but the cross-sectional ranking and the regime backdrop do the heavy lifting. How those inputs combine into a single ranked, published output is the subject of how a trading signal is made, and the broader role of momentum across the system is covered in momentum investing explained.
So: what is RSI?
RSI is Wilder’s bounded 0-100 oscillator that compresses a name’s recent up-moves against its recent down-moves into a single number, a measure of internal momentum, not of strength versus the market. The 70/30 bands describe how one-sided that recent movement has been; they are not a buy-and-sell switch, and the strongest trends live above the upper one for long stretches. Read as a description of momentum and weighed alongside everything it ignores, RSI is a genuinely useful lens. Mistaken for relative strength, or fired as a standalone trigger against a trend, it is an expensive one.
Sources & further reading
- Wilder, J. W. (1978). New Concepts in Technical Trading Systems. Trend Research., the origin of RSI (and ATR).