The belief

RSI below 30 means the stock is oversold. Oversold means it has fallen too far. When something has fallen too far, it bounces. Therefore: RSI below 30 is a buy signal.

This logic is everywhere. It is in Zerodha Varsity. It is in every YouTube tutorial that covers technical analysis. It is probably the first thing most retail traders learn after candlestick patterns. The idea is intuitive, easy to understand, and very easy to code into a screener. You can pull up a list of stocks with RSI under 30 in about twenty seconds on any platform in India.

Which is exactly why we decided to test it properly.

The problem with this belief

The belief treats RSI as a standalone reversal signal. It doesn't account for trend context, timeframe, or the type of market the stock is operating in. These omissions are not minor details - they are the entire difference between a useful signal and a coin flip with worse odds.

A stock can have an RSI of 28 and continue falling to RSI 12. Then RSI 9. It happens regularly. During a strong downtrend, RSI spends extended periods below 30. Each new reading below 30 looks like "more oversold" - but what's actually happening is that sellers are in control and momentum is persistently negative. The RSI reading is accurate. It is correctly describing the situation. The problem is that "oversold" and "about to reverse" are not synonyms.

RSI is a momentum oscillator. It measures the speed and magnitude of recent price changes relative to a period average. It does not measure value. It does not know whether the stock is fundamentally cheap or expensive. It does not know whether the broader market is in a risk-on or risk-off state. Using it as a reversal predictor without asking those questions first is where retail traders consistently lose money.

Important distinction RSI < 30 tells you that selling momentum has been dominant over the recent period. That is a factual measurement. What it does not tell you is whether that momentum is about to reverse. Whether it reverses depends almost entirely on the market regime - and RSI alone cannot tell you what regime you are in.

What we tested

We wanted to strip away all interpretation and test the raw signal in isolation. No confluence, no filters, no additional conditions. Just the RSI-below-30 rule as most people apply it: see it, buy it.

Methodology

  • Universe: Nifty 50 constituents (index composition as of each date, survivorship-adjusted)
  • Timeframe: Daily bars, 5-year lookback period
  • Indicator: RSI with standard 14-period setting
  • Signal: Every instance where RSI crossed below 30 (first touch, not continuation)
  • Measurement: Closing price return at 5, 10, and 20 trading days after signal
  • No additional filters - trend, volume, sector, or otherwise
  • Total signals identified: 1,847 across the universe and period

The methodology is intentionally simple. We are not testing a sophisticated setup - we are testing the belief as it is most commonly held and applied by retail traders. If it works even in this crude form, that's useful to know. If it doesn't, that's equally useful.

The result

The raw signal, applied without any filter, produces the following outcomes:

52.3% Win rate at 5 days.
Barely above a coin flip.
50.1% Win rate at 10 days.
Essentially random.
51.8% Win rate at 20 days.
No meaningful edge.

Win rate alone, however, is not the full picture. A strategy with 52% win rate could still be profitable if the wins are significantly larger than the losses. The opposite is true here.

Key findings

Average loss magnitude vs. average gain magnitude −1.4x
Expectancy per trade (raw RSI signal, no filter) Negative
Win rate in confirmed downtrends (price below 200-day MA) 33%
RSI < 30 followed by further decline, in strong downtrends 67%
Win rate in range-bound conditions (ADX < 20) 61%
Expectancy in range-bound conditions Positive

The pattern is clear once you split the data by market regime. In a ranging market, RSI extremes are genuinely useful - prices oscillate between support and resistance, and an RSI reading below 30 does correlate with a near-term mean-reversion move. In a trending market, particularly a downtrend, RSI below 30 is not a reversal signal. It is a momentum confirmation signal. It is telling you the sellers are dominating. Acting against it without a strong structural reason is, on average, a losing trade.

The number that matters most: in strong downtrends, RSI below 30 preceded further decline 67% of the time over the following 5 trading days. The signal is predictive - just not in the direction most traders assume.

What this actually means

RSI is a momentum oscillator. That is a precise description, not a casual label. It measures momentum - the rate and direction of recent price change. When RSI is below 30, it is telling you that downward momentum has been significant over the past 14 periods. That is a factually correct and useful piece of information.

The mistake is treating this momentum reading as a reversal prediction. Using RSI below 30 to decide "this is a buy" is like using a speedometer to decide whether to turn left. The instrument is measuring something real. It is just not answering the question you are asking.

For RSI to function as a mean-reversion signal, the underlying market structure needs to support mean-reversion. That structure exists in ranges, where price is bounded between a floor and a ceiling and has no directional intent. It does not exist in downtrends, where the prevailing structure actively works against mean-reversion. The indicator does not contain the information to distinguish between these two conditions. You have to bring that context yourself.

The right way to use it

We are not going to prescribe an exact setup here - that depends on your strategy, timeframe, and risk framework. But the directional guidance is straightforward:

  • Identify market structure first. Before reading any indicator, establish whether the instrument is in a trend or a range. Price structure, moving averages, and directional indicators like ADX give you this. RSI alone does not.
  • In ranges, RSI extremes are legitimate signals. When price is oscillating between defined levels and the trend is absent, an RSI reading below 30 carries genuine mean-reversion probability. This is the context where the original intuition holds.
  • In downtrends, treat RSI below 30 as momentum confirmation. Not a reversal trigger. If you are looking for a counter-trend entry in a downtrend, you need structural reasons: a major support level, an exhaustion pattern, a divergence, a capitulation candle. RSI below 30 alone is not a structural reason.
  • RSI is never a standalone trade. It is one piece of evidence in a confluence-based argument. If RSI below 30 is the only reason you are buying, the position has no structural foundation.

What this means for where you are in the learning curve

If you are currently using RSI below 30 as a buy signal without accounting for trend context, you are not doing anything unusual - this is how the majority of retail traders use it, because this is how it is taught. The gap between what is taught and what the data shows is precisely what this series is about.

Understanding RSI properly - as a momentum tool rather than a reversal oracle - requires first building a framework for reading market structure. You cannot use RSI in the right context if you do not yet have the tools to identify context. This is a sequencing problem, not an intelligence problem.

In the Bharath Shiksha curriculum, Stage 2 covers momentum indicators including RSI in full technical depth: how it is constructed, what it actually measures, how it behaves differently across market regimes, and where it genuinely adds information versus where it is noise. Stage 3 is where you learn to build and backtest structured setups - so you are not relying on received wisdom, but on your own tested data. Stage 4 is where you run tests like the one in this note and build your own edge documentation.

The progression matters. Stage 1 to Stage 4 is a systematic build - each stage is designed to give you the prerequisite understanding before the next layer adds complexity.