RSI (Relative Strength Index) Explained: Formula, Uses, and Limits

Origins of the RSI

The Relative Strength Index (RSI) was developed by J. Welles Wilder Jr. and introduced in his 1978 book New Concepts in Technical Trading Systems. Wilder was a mechanical engineer turned technical analyst, and the same book also introduced several other indicators that remain widely used today, including the Average True Range (ATR), the Average Directional Index (ADX), and the Parabolic SAR.

The RSI was designed to measure the speed and magnitude of recent price changes in order to evaluate whether a security is overbought or oversold. Despite being nearly five decades old, it remains one of the most popular technical indicators in use. Every major charting platform includes it by default, and it appears in countless trading strategies, screeners, and algorithmic systems.

Understanding the RSI properly requires going beyond the superficial "buy below 30, sell above 70" interpretation that most traders learn first. The indicator has genuine utility, but also real limitations that are important to understand.

The RSI Formula

The RSI is calculated in two steps. First, you compute the Relative Strength (RS), then you normalize it into an oscillator that ranges from 0 to 100.

Wilder's original recommendation was to use N = 14 periods. This remains the default on virtually all charting platforms.

The "average gain" and "average loss" require some clarification. For each period, you look at the price change from the previous close. If the change is positive, it counts as a gain; if negative, the absolute value counts as a loss. Periods with no change are ignored.

Initial Calculation

For the first RSI value, you compute a simple average of gains and losses over the first 14 periods:

• First Average Gain = Sum of gains over the first 14 periods / 14
• First Average Loss = Sum of losses over the first 14 periods / 14

Subsequent Calculations (Smoothing)

After the initial value, Wilder used a smoothing technique (equivalent to an exponential moving average) for all subsequent values:

This smoothing method means that the RSI incorporates information from all prior periods, not just the most recent 14. The older data has diminishing influence, but it never fully disappears. This is an important detail -- it means the RSI value you see depends on the entire price history, not just a fixed lookback window.

Step-by-Step Calculation Example

Let's walk through a simplified example using 14 periods of daily closing prices. Suppose a stock has the following daily price changes over 14 consecutive trading days:

Day 1: +0.50, Day 2: -0.30, Day 3: +0.70, Day 4: +0.40, Day 5: -0.20, Day 6: +0.60, Day 7: -0.50, Day 8: +0.30, Day 9: +0.80, Day 10: -0.10, Day 11: +0.20, Day 12: -0.40, Day 13: +0.50, Day 14: +0.30

Step 1: Separate the gains and losses. Gains: 0.50, 0.70, 0.40, 0.60, 0.30, 0.80, 0.20, 0.50, 0.30. That is 9 gaining periods summing to 4.30. Losses (absolute values): 0.30, 0.20, 0.50, 0.10, 0.40. That is 5 losing periods summing to 1.50.

Step 2: Compute the first averages. Average Gain = 4.30 / 14 = 0.3071. Average Loss = 1.50 / 14 = 0.1071. Note that you divide by 14 (the full period), not by the count of gaining or losing days.

Step 3: Compute RS = 0.3071 / 0.1071 = 2.867.

Step 4: Compute RSI = 100 - (100 / (1 + 2.867)) = 100 - (100 / 3.867) = 100 - 25.86 = 74.14.

This stock would be considered slightly overbought under traditional interpretation. For subsequent days, you would use the smoothing formula rather than recalculating from scratch.

Traditional RSI Interpretation

The standard interpretation of the RSI uses two threshold levels:

• RSI above 70: The security is considered overbought. This suggests that recent gains have been unusually large relative to recent losses, and the price may be due for a pullback or consolidation.
• RSI below 30: The security is considered oversold. Recent losses have been unusually large, and the price may be due for a bounce.
• RSI around 50: Neutral. Neither buying nor selling pressure dominates.

These thresholds -- 30 and 70 -- were Wilder's original recommendations. Some traders use 20/80 for more extreme signals, or 40/60 for more sensitive triggers. The choice is ultimately arbitrary; there is no mathematical reason why 70 is better than 72 or 68.

Centerline Crossover

The RSI crossing above 50 is sometimes interpreted as a bullish signal (average gains are exceeding average losses), while crossing below 50 is bearish. This interpretation is essentially a trend-following use of the RSI, rather than a mean-reversion use.

RSI Divergences

One of the more nuanced RSI applications is divergence analysis. A divergence occurs when the price and the RSI move in opposite directions, potentially signaling a weakening trend.

Bearish Divergence

A bearish divergence occurs when the price makes a new high, but the RSI makes a lower high than its previous peak. This suggests that while the price is still rising, the momentum behind the rally is fading. The gains are becoming smaller relative to losses compared to the previous push higher.

Bullish Divergence

A bullish divergence occurs when the price makes a new low, but the RSI makes a higher low than its previous trough. This suggests that selling pressure is diminishing even as the price continues to fall.

Failure Swings

Wilder himself considered failure swings to be more reliable signals than divergences. A failure swing is a specific pattern in the RSI that does not require comparing RSI to price.

Bullish Failure Swing

The RSI drops below 30 (oversold), bounces above 30, pulls back but stays above 30, then breaks above the prior RSI high. The key feature is that the RSI finds support above 30 on the pullback -- it fails to swing back into oversold territory. This suggests the selling pressure has exhausted itself.

Bearish Failure Swing

The RSI rises above 70 (overbought), drops below 70, rallies but fails to get back above 70, then breaks below the prior RSI low. The RSI fails to swing back into overbought territory, suggesting buying pressure is fading.

Failure swings are considered more reliable because they use only the RSI itself and are more precisely defined than divergences. However, like all technical patterns, they produce false signals and should not be used in isolation.

RSI in Trending vs. Ranging Markets

This is one of the most important practical considerations for using the RSI, and it is frequently overlooked by newer traders.

In a strong uptrend, the RSI can stay in overbought territory (above 70) for extended periods. During the powerful bull markets of 2020-2021, many stocks maintained RSI readings above 70 for weeks or even months. Selling every time the RSI hit 70 would have been catastrophic -- you would have been shorting or exiting one of the strongest rallies in history.

In a strong downtrend, the opposite applies. The RSI can remain below 30 for long stretches. Buying every dip to RSI 30 during a sustained bear market means repeatedly catching falling knives.

Constance Brown, in her book Technical Analysis for the Trading Professional (1999), proposed adjusting RSI thresholds based on the trend. In a bull market, she suggested using 40 as oversold and 80 as overbought. In a bear market, 20 as oversold and 60 as overbought. While these specific numbers are somewhat arbitrary, the underlying principle -- that RSI ranges shift with the trend -- is well-supported by observation.

Common RSI Variations

RSI-2 (Connors RSI)

Larry Connors popularized the use of a 2-period RSI for short-term mean reversion trading. The logic: a 2-period RSI is extremely sensitive to recent price changes, making it useful for identifying short-term oversold conditions in stocks that are in longer-term uptrends.

The typical strategy involves buying when the 2-period RSI drops below 10 (or even 5) on a stock that is above its 200-day moving average, then selling when the 2-period RSI rises above 90. Connors published backtesting results showing this approach generated positive returns, particularly in large-cap and index-level applications.

Weekly RSI

Applying the standard 14-period RSI to weekly charts produces a longer-term momentum indicator that filters out daily noise. Weekly RSI readings above 70 or below 30 are less common and therefore potentially more meaningful than the same readings on daily charts.

RSI with Different Periods

Shorter periods (e.g., RSI-7 or RSI-9) produce a more volatile indicator that crosses overbought/oversold thresholds more frequently. Longer periods (e.g., RSI-21 or RSI-25) produce a smoother indicator that generates fewer signals. The trade-off is always sensitivity versus reliability.

Academic Evidence: What Research Actually Says

This is where honest assessment matters. The academic evidence on the RSI's predictive power is, at best, mixed.

Most rigorous academic studies find that the RSI, used alone as a mechanical trading rule, does not consistently produce risk-adjusted returns that exceed a buy-and-hold strategy after accounting for transaction costs. This is consistent with the broader academic literature on technical analysis, which generally finds that simple trading rules have weak standalone predictive power.

However, there are some nuances worth noting:

As a confirmation filter: Several studies have found that the RSI adds value when used as a filter or confirmation tool within a broader strategy, rather than as a standalone signal generator. For example, requiring RSI confirmation before acting on a fundamental signal can improve hit rates.

In combination with other indicators: Research has shown that combining momentum oscillators like the RSI with trend-following indicators (such as moving averages) can produce better results than either approach alone. The idea is that the trend indicator determines direction, while the RSI helps with timing entries within the trend.

Short-term mean reversion: There is somewhat stronger evidence for RSI-based mean reversion strategies at very short time horizons (2-5 days), particularly in liquid, large-cap stocks and indices. This is consistent with the microstructure literature showing short-term return reversals.

Limitations of the RSI

Lagging Indicator

The RSI is calculated from past price data, so it is inherently backward-looking. By the time the RSI reaches 70, the move has already happened. The indicator tells you what has occurred, not what will occur. This is true of all indicators based on historical prices, but it is worth stating explicitly because many traders treat the RSI as if it were predictive.

Does Not Account for Volume

The RSI is calculated purely from closing prices. It does not incorporate volume information at all. A price move on extremely high volume (which may indicate strong conviction) produces the same RSI reading as the same price move on negligible volume. Indicators like the Money Flow Index (MFI), sometimes called the "volume-weighted RSI," attempt to address this limitation by incorporating volume into a similar oscillator framework.

Arbitrary Thresholds

The 30/70 thresholds are conventions, not scientifically derived boundaries. There is no mathematical property of the RSI that makes 70 a meaningful cutoff. Wilder chose these levels based on his experience and judgment. While they have become self-fulfilling to some degree (because so many traders watch the same levels), they are ultimately arbitrary.

Sensitivity to Lookback Period

Changing the lookback period from 14 to 7 or 21 changes the RSI readings substantially. There is no objective way to determine the "correct" lookback period. This introduces a degree-of-freedom problem: if you test many different lookback periods, you will inevitably find one that worked well historically, but this is data-mining rather than a genuine edge.

False Signals in Trending Markets

As discussed above, the RSI generates systematic false signals in trending markets. Overbought readings in uptrends are not sell signals; they are confirmation of strength. This is the single most common mistake traders make with the RSI.

Practical Guidelines for Using the RSI

Given everything above, here are some practical principles for incorporating the RSI into a trading process:

1. Never use the RSI in isolation. It works best as one input among many. Combine it with trend analysis, volume, support/resistance, and ideally some form of fundamental information.
2. Identify the market regime first. Before interpreting the RSI, determine whether the market or stock is trending or range-bound. Apply overbought/oversold interpretation only in ranging markets.
3. Use divergences as alerts, not entry signals. A divergence tells you to pay closer attention, not to immediately trade.
4. Consider the RSI as a measure of momentum, not direction. A declining RSI in an uptrend does not mean the trend is over -- it means the pace of gains is slowing.
5. Be wary of optimization. If you find a specific RSI parameter set that works beautifully in backtesting, be suspicious. Robust strategies should work across a range of parameter values.

RSI in the Context of Multi-Factor Signal Analysis

The RSI is most powerful when integrated into a broader analytical framework. Quantitative trading systems rarely rely on a single indicator. Instead, they combine multiple signals -- each capturing different information -- and use the convergence of signals to generate high-conviction trades.

For example, consider a scenario where a stock shows the following characteristics simultaneously: insider buying detected through SEC Form 4 filings, RSI pulling back to the 40-50 range within an established uptrend, above-average volume on the recent bounce, and the stock holding above its 200-day moving average. Each of these factors alone is a weak signal. Together, they form a much more compelling case.

The RSI's role in this type of multi-factor analysis is to assess the timing of the entry. The insider buying provides the fundamental information edge -- insiders know their business. The RSI helps identify a favorable entry point where the stock has pulled back enough to offer a reasonable risk/reward ratio, but not so much that the trend may be broken.

Alpha Suite incorporates RSI-14 as one of several technical overlays in its signal scoring pipeline, alongside momentum, volume-confirmed breakouts, and relative strength versus the S&P 500. The RSI is used as a confirmation filter, not as a primary signal generator -- exactly in line with what the academic evidence suggests is its most effective application.