The relative strength index

The relative strength index compares the size of recent gains against the size of recent losses and expresses the result from 0 to 100. The standard setting, and the one used on the charts here, looks back over fourteen periods.

The calculation averages the gains across those periods and the losses across the same periods, divides one by the other to give relative strength, and converts it to a bounded scale. If the average gain over fourteen sessions is 1.0 and the average loss is 0.5, relative strength is 2, and the index reads 100 minus 100 divided by 3, which is 67.

The bounding is the point. Momentum has no natural ceiling, and dividing it this way produces a number that always sits between 0 and 100 and can therefore be compared against itself over time. If every one of the last fourteen sessions was up, RSI sits near 100; if every one was down, near 0; if gains and losses have been similar in size, near 50.

Readings above 70 are conventionally called overbought and readings below 30 oversold. Neither term refers to value, and both are frequently misread as one.

An RSI of 80 records that the price has risen persistently and that the up sessions have been larger than the down ones. It says nothing about earnings, about what the company is worth, or about whether the rise was justified. A company can be overbought and cheap, or oversold and expensive, at the same time.

RSI is built for range-bound conditions, where a price oscillates around a level and extremes tend to revert. In a sustained trend it saturates and stops discriminating: an indicator pinned near 80 for a month reports the same thing every day, and the reading that was supposed to be exceptional becomes the normal state.

Strong uptrends routinely hold RSI above 70 for weeks. Acting on each such reading means selling into every stage of a rise, which is the opposite of what the indicator is usually reached for.

It is also derived entirely from price. It carries no information about earnings, cash flow, an acquisition or a product failure, so it cannot distinguish a company falling because sentiment turned from one falling because the business is deteriorating. Both produce an RSI of 25.

The more informative reading is divergence rather than level. A price making a new high while RSI does not indicates the latest move was driven by fewer and smaller advances than the one before it, which is a statement about the composition of the move rather than about its direction.

Bearish divergence is a higher price against a lower RSI peak; bullish divergence is a lower price against a higher RSI trough. Both are observations about momentum thinning out, and neither carries timing: divergence can persist for months while the price continues in the same direction.

The fourteen-period default came from Welles Wilder, who introduced the indicator in 1978, and it has stuck as a convention rather than as a result. Shorter settings, such as 7 or 9, react faster and cross the thresholds far more often, producing many more signals of which a larger share are noise. Longer settings, such as 21 or 25, lag and cross rarely.

Because the period determines how often the thresholds are reached, any rule expressed in terms of 70 and 30 is a rule about a setting somebody else chose. Changing the period without changing the thresholds changes the rule.

The period on the charts here is the standard fourteen. RSI is calculated on whatever interval the chart is showing, so fourteen periods on a daily chart is fourteen sessions and fourteen periods on an intraday chart is fourteen of those bars.

Reading overbought as expensive. The two words describe different things, and the indicator has no access to the information that would make it a claim about value.

Acting on a threshold crossing in a trending market, where the indicator saturates and every day looks like a signal.

Comparing RSI across intervals. A reading of 75 on an hourly chart and 75 on a weekly chart describe different lengths of time and are not the same observation.

Treating divergence as a timing tool. It describes momentum thinning, and prices can continue in the same direction for a long time after it appears.

Using RSI alone on a company in the middle of news. Price-derived indicators know nothing about the event driving the price, and this is precisely when they are least informative.