RSI Explained: Overbought, Oversold & Divergence Trap

Think of RSI like a speedometer, not a GPS. A speedometer tells you how fast the car is going; it does not tell you where the road turns next. RSI is the same — it tells you the recent move is fast or tired, but the price chart still tells you the direction.

Here is the trap, in one short story. A beginner sees RSI touch 70 and feels the chart has finally given a clean answer: sell before everyone else notices. Then the stock keeps rising for another week. He was not wrong that the move was stretched — he was wrong to treat stretched as finished.

RSI sits below almost every price chart on Zerodha Kite, TradingView and most other charting apps, often switched on by default. The line wiggles between 0 and 100 and tells you something the price chart alone does not: how stretched the recent move is, and whether the buyers or the sellers are quietly losing steam.

The trouble starts when beginners treat that wiggling line like a buy-and-sell button. Sell when it hits 70, buy when it hits 30. That rule, as written, can lose money badly in strong trends — in Nifty (India's benchmark 50-stock index), Bank Nifty, or large stocks like Reliance and HDFC Bank. The real value of RSI is more interesting, and a lot more honest.

What RSI actually measures

RSI was created by J. Welles Wilder, an American engineer turned trader, who introduced it in his 1978 book New Concepts in Technical Trading Systems. The idea behind it is simple. Over a recent window — usually the last 14 days — RSI separates the up-moves from the down-moves, averages the size of each, and compares those two averages.

So it is not just counting up-days against down-days; the size of each move matters. If the up-moves dominate, the indicator climbs toward 100. If the down-moves dominate, it falls toward 0. If they are roughly even, it settles near 50.

That is the whole idea. RSI is a score, between 0 and 100, of how lopsided the last 14 days have been.

The default window is 14 periods. On a daily chart that means 14 trading days; on an hourly chart, 14 hours; on a 5-minute chart, 14 candles of five minutes each. A candle is simply one block of time on a chart, so one daily candle is one full trading day.

RSI does not care which timeframe it sits on. It simply looks at the most recent 14 candles.

The famous lines are 70 and 30. Above 70 the recent move has been so one-sided that buyers have done most of the work. Below 30 the sellers have. The middle line, 50, is the neutral mark and is more useful than most beginners realise.

How RSI is actually built

You will never calculate RSI by hand and you should not need to. Every charting platform does it for you. But knowing what is inside the box makes the readings far less mysterious.

RSI takes the last 14 candles and splits them into winners and losers. A winner is a candle that closed above the previous close — the close being the final traded price for that candle or day. A loser is a candle that closed below it. The size of each win or loss is measured in rupees.

Then it does two averages: the average win, and the average loss. The ratio of those two is called the relative strength, or RS. The indicator then squashes RS into a 0-to-100 range using a small formula, so the output is always bounded.

The fact that RSI is bounded between 0 and 100 is the key. Stocks can climb forever, so a price chart never tells you when a move has gone too far. RSI puts a ceiling and a floor on the conversation, which is why it can flag exhaustion that price alone hides.

When all 14 recent candles are winners, the average loss is essentially zero, and RSI prints close to 100. When all 14 are losers, the average win is zero, and RSI prints close to 0. Neither extreme is common. Real markets rarely deliver 14 one-sided candles in a row.

The default setting of 14 has stayed standard for nearly fifty years. Shorter settings like 7 or 9 react faster but throw more false signals. Longer settings like 21 or 25 lag more but filter the noise. For NSE charts (the National Stock Exchange, India's largest), 14 on the daily and 14 on the hourly is simply the widely used default on most charting platforms — a sensible starting point, not a magic number.

Overbought, oversold and the 50-line

The 70 and 30 lines are the famous ones, but the most useful line on the RSI panel is the one most beginners ignore. The 50 line marks the difference between an uptrend and a downtrend in momentum terms — what traders call the regime. Regime simply means the current environment: buyers in control, sellers in control, or sideways.

In a healthy uptrend, RSI tends to live between 40 and 80. It tags 70 or higher on every fresh push, pulls back to around 40 to 50 on dips, and then heads back up. The 40-to-50 zone becomes support on the indicator, mirroring what price is doing on the chart.

In a healthy downtrend, the picture flips. RSI lives between 20 and 60. It hits 30 or lower on every fresh drop, rallies back to around 50 to 60 on bounces, and then turns down. The 50-to-60 zone becomes resistance on the indicator.

This is the part of RSI that beginners most often miss. Overbought is not the same as expensive, and oversold is not the same as cheap. They are statements about the speed of the recent move, not about value or direction.

A stock can sit at RSI 80 for three weeks and grind higher the whole time. A stock can sit at RSI 20 for three weeks and grind lower the whole time. The first 70 reading in a new uptrend is a signal of strength, not weakness. The first 30 reading in a new downtrend is a signal of weakness, not opportunity.

Use the 50 line to decide the regime, and use 70 and 30 to spot exhaustion inside that regime. That is the framework most professional traders run on, and it is far more useful than the textbook rule of selling at 70 and buying at 30.

What RSI divergence really is

Divergence is where RSI earns its keep. It is also where most beginners blow themselves up. Understanding both halves of that sentence is the point of this section.

The idea is simple. Price and RSI usually move together: new highs in price come with new highs in RSI, and new lows in price come with new lows in RSI. Divergence is when they stop agreeing.

Bearish divergence is when price makes a higher high but RSI makes a lower high. The chart looks healthy and bullish on the surface, but momentum under the bonnet is fading. Each new high in price needed less buying power to print than the last one.

Bullish divergence is when price makes a lower low but RSI makes a higher low. The chart looks heavy and bearish on the surface, but the sellers are running thin. Each new low in price needed less selling pressure than the last one.

Two often-studied examples from the Nifty 50 show both sides. In the run-up to the January 2008 top, price kept tagging fresh highs into the first week of January, while daily RSI is widely noted to have rolled over from above 80 toward the 60s.

The chart looked unstoppable; the indicator was quietly less convinced. Nifty peaked near 6,357 in January 2008 and fell to around 2,250 by October 2008 — a drop of nearly two-thirds in roughly ten months.

The opposite is often pointed to at the bottom of the March 2020 Covid panic. Nifty made fresh lows into late March — closing at 7,610 on 23 March 2020, its low for that crash — even as daily RSI is commonly described as turning up and making higher lows. That is the textbook shape of bullish divergence into the lowest fear of the cycle.

Large, liquid stocks often show RSI divergences on daily charts, and Bank Nifty produces them on hourly charts too — frequently around the monthly expiry, the day index and stock derivatives are settled. The pattern is not rare. It is the question of how to act on it that catches people out — so scan for these setups, but verify each one before trusting it.

The divergence trap

If divergence is so useful, why call it a trap? Because in strong trends, divergence can keep appearing and keep failing for weeks before any real reversal happens.

A stock in a powerful uptrend can show bearish divergence at RSI 70, then at 75, then at 80, and still grind higher the whole time. Every fresh push prints a lower high on the indicator because the move is so stretched that no single push can match the earlier ones in raw momentum. The price chart, meanwhile, keeps trending up.

Beginners spot the first divergence and short. The market does not reverse. They add to the short on the next divergence; the market still does not reverse.

By the third divergence, the position is large and the stop is far. Then the trend does eventually turn, but by then the account is already wounded.

This is the divergence trap. Divergence tells you momentum is fading. It does not tell you the trend has ended. Confusing those two ideas is the most expensive RSI mistake a retail trader can make.

The fix is to treat divergence as a warning, never as an entry. The entry needs a confirming signal from price itself.

For a bearish divergence, wait for price to break a clear swing low — a recent dip, a visible valley on the chart. For a bullish one, wait for price to break a swing high — a recent peak.

A close back through a moving average — a smoothed line that averages the last several closing prices — counts too. So does a break of the trendline: a straight line joining a run of rising lows or falling highs that marks the shape of the move.

Until price confirms, divergence is just a yellow light. It says be careful, take some profit off, tighten the stop. It does not say slam the brakes and go the other way.

One more practical filter helps. Divergence works far better at the second or third extreme reading than at the first. A market that has hit RSI 80 once and is showing divergence is far less likely to actually top than one that has tagged 80 three or four times across a few weeks and is now showing fading momentum on each touch.

Strong trends rarely turn on the first warning. They turn after the third or fourth.

How to actually trade with RSI

Once the framework is clear, the rules become straightforward. RSI is not a stand-alone system. It is a filter that improves trades you would consider anyway from price action.

Use it for three jobs, in this order. Decide the regime, time the entries inside that regime, and listen for early warnings of exhaustion.

For the regime, look at where RSI lives on the chart. If most readings over the last few weeks are above 50, the bias is up. If most are below 50, the bias is down. Trade in that direction; do not fight it.

For the entries, in an uptrend wait for RSI to pull back to the 40-to-50 band and then turn up. Combine that with a price-based confirmation, such as a bounce off support, a hammer candle (a candle with a long lower wick, a sign buyers stepped back in), or a moving-average touch, and the entry has both momentum and price agreeing.

For the exhaustion warnings, watch for divergence after the third or fourth tag of an extreme reading. When you see it, do not flip the trade.

Take partial profit, trail the stop closer, and wait for price itself to break before reversing.

One small practical note. The 14-period default is the right starting point. Resist the urge to tune it heavily. Most beginners shorten RSI to chase faster signals and end up with a noisier indicator that fires on every wiggle.

The other temptation is to stack RSI with three or four other oscillators on the same chart. MACD, Stochastic, Williams %R, all measuring something similar to RSI. The more indicators you stack, the less each one means. One momentum indicator, used well, is enough.