Hook — a speed meter is not a map

A beginner sees RSI at 80 and concludes that the next price must fall. Another sees RSI at 25 and thinks the next candle must rise. Both conclusions go beyond what the indicator measures. The relative strength index, or RSI, is a momentum oscillator derived from recent upward and downward price changes. It can describe how the balance of those changes has behaved over a chosen observation window, but it does not see tomorrow's buyers or sellers. Its usual display runs from 0 to 100. Those numbers are index points, not percentage profit, a probability of success or a count of transactions. The indicator can remain high during a sustained rise or low during a sustained fall. Learning RSI means understanding the arithmetic, the period and smoothing convention, and the distinction between a descriptive reading and an unsupported forecast. We will use fictional price changes and avoid any instruction to trade.

Concept in depth — gains, losses and a bounded scale

RSI compares the magnitude of average positive changes with the magnitude of average negative changes over a selected number of bars. A gain is an upward close-to-close change; a loss is the absolute magnitude of a downward close-to-close change, not an investor's realized loss. A common calculation first forms average gain and average loss, then relative strength RS = average gain / average loss, and RSI = 100 − 100/(1+RS). Many platforms use Wilder's smoothing after initialization; a simplified arithmetic average is suitable for explaining the formula but is not always identical to a platform's production RSI. The commonly displayed 14-period setting is a convention, not a universal rule. An RSI reading of 80 does not mean an asset has an 80% chance of declining. It also does not tell us that the price is 80% above a starting level. The scale compresses the balance of recent average gains and losses into one bounded measure. Its sensitivity changes with the lookback and the data feed. Educational background is available from CME Group and Charles Schwab; exact platform conventions still need independent checking.

How it works — separate direction from size

Start with a single instrument, quotation unit, completed candle series and a declared lookback length. For each adjacent pair of closes, subtract the earlier close from the later close. A positive difference contributes to gains and zero to losses; a negative difference contributes its positive magnitude to losses and zero to gains. Do not treat a flat close as both a gain and a loss. Form the average gain and average loss according to the chosen initialization and smoothing rule. For our fictional illustration, we are given average gain ₹4 and average loss ₹1. RS = 4/1 = 4. Then RSI = 100 − 100/(1+4) = 80. The result is 80 index points, not ₹80 or 80% profit. If average loss is zero, ordinary division by zero is undefined; software needs an explicit edge-case convention, often showing RSI near or at 100 when gains are positive. If both gains and losses are zero, platform treatment may differ. Before comparing screenshots, check their lookback, closing-price source, bar timestamps and smoothing. A current unfinished candle can change the latest computed RSI.

Desi example — a mandi clerk counts rising and falling days

Desi example: Imagine a fictional onion-market clerk who records the daily closing quote of a crate. Some days the quote rises, other days it falls. The clerk writes each positive change in one column and each negative change's magnitude in another. A ₹4 upward move belongs in the gain column, while a ₹1 downward move belongs in the loss column as a magnitude of one. The clerk then compares typical changes, not the number of profitable traders. If the average gain is four units and the average loss is one unit, the illustrative ratio is four to one. That resembles a high RSI reading, but it does not say whether the clerk bought onions, paid transport costs or earned any money. Nor does it tell us what tomorrow's wholesale price will be. A harvest disruption could change the next observation. The analogy clarifies the difference between measuring historical price movement and calculating a merchant's business profit.

ILLUSTRATIVE ONLY — compute 80 and compare 50

Given fictional average gain ₹4 and average loss ₹1, RS = 4 and RSI = 100 − 100/5 = 80. For a separate imaginary series with average gain ₹2 and average loss ₹2, RS = 1 and RSI = 100 − 100/2 = 50. The difference between these readings is 30 RSI points. It is not a ₹30 price difference, 30% guaranteed return or a comparison of two exchange rates. A further invented case has average gain ₹1 and average loss ₹4. Then RS = 0.25, so RSI = 100 − 100/1.25 = 20. These examples isolate the formula; they do not reproduce the full time-series initialization of a real chart. The average gain and loss values are supplied for the classroom, not claimed as observed market data. If a fictional investor bought one unit at ₹100 and later sold at ₹99, the gross price difference would be −₹1 before fees regardless of an RSI reading. The oscillator cannot replace transaction records.

Common Mistakes — 70 and 30 are not orders

A widely used convention marks RSI 70 as an upper reference and 30 as a lower reference. These are observation thresholds, not laws forcing price to reverse or instructions to sell and buy. Another mistake is treating RSI 80 as 80% certainty or an 80% price increase. RSI is not a probability model. Some students confuse price gain with investor gain; a candle may rise without a particular person owning the asset. Others compare RSI readings across different timeframes without noting that the underlying changes differ. A 14-period indicator on five-minute bars is not the same as one on daily bars. The smoothing convention also matters: a simple classroom average and Wilder-smoothed RSI can give different results. Some screenshots hide the price source, and an unfinished bar can move the latest value. The index cannot reveal spreads, financing costs, fills or tax obligations. When settings are missing, write NEEDS-REVIEW rather than assert that a high or low reading guarantees anything.

Real-world-style scenario — the confident screenshot

An imaginary social post shows an RSI gauge at 78 with the caption 'sell now, crash guaranteed'. The screenshot crops out the instrument, timeframe, calculation settings and candle timestamps. Even if the 78 reading is authentic, it describes a particular calculation on historical observations. It cannot establish the next return or a person's ability to transact at a displayed price. A careful student asks whether the indicator uses completed bars, which price series it measures and whether the platform uses standard Wilder smoothing. The student can compare the reading with a clearly labelled independent calculation, noting any differences rather than inventing an explanation. The screenshot's author might have confused a conventional reference threshold with a prediction. This is a fictional example of misleading reasoning, not an allegation about any actual person or service. A responsible explanation reports what the RSI measures and explicitly states what remains unknown.

Summary and recap — RSI is a measure of past momentum

RSI is a bounded momentum oscillator based on relative magnitudes of recent positive and negative close-to-close changes. Its standard scale is 0–100. In a simplified invented example, average gain ₹4 and average loss ₹1 give RS 4 and RSI 80; equal averages produce RSI 50. These numbers are index points, not personal returns or probabilities. The selected period, smoothing convention, quote source and completion of the current bar determine how a reading is calculated. Reference levels such as 70 and 30 are conventions, not guarantees of reversal. An RSI calculation cannot establish actual trade execution or future market direction. The practical educational skill is to check the inputs, reproduce the arithmetic where possible and separate the measurement from any claim about what must happen next. All examples are fictional.