A chart marked “overbought” can look like an instruction. It is not. Before deciding what an RSI reading means, identify the price series, the time interval and the calculation behind the number. Otherwise, a precise-looking value can answer a different question from the one you intended to ask.
Interest in RSI appeared in the Swiss Google Trends feed checked on 25 October 2026, alongside a commentary about oil-market charts. This guide addresses the financial Relative Strength Index. It does not report a current oil price, endorse that commentary's forecast or recommend a trade. The examples use invented numbers so you can follow the arithmetic without confusing a teaching exercise with market data.
What RSI measures
RSI is a momentum indicator associated with J. Welles Wilder. It expresses the relationship between upward and downward price changes on a scale from zero to 100. The familiar reference levels are 70 and 30, commonly described as overbought and oversold. Fidelity also notes that extreme readings can persist during strong trends. [Fidelity's RSI guide](https://www.fidelity.com/learning-center/trading-investing/technical-analysis/technical-indicator-guide/RSI).
The number is not a probability. An RSI of 75 does not mean a 75% chance of a rise, a 75% chance of a fall or a company being 75% overvalued. Those statements would require information the calculation does not contain. A momentum measure describes a feature of its input history; it is not a valuation of the business behind a security.
That distinction is useful even if you never trade. Market commentary often moves quickly from description to prediction. When a headline includes an indicator, ask whether the author is reporting its value, interpreting a pattern or proposing a future outcome. Those are three separate steps, and each deserves its own evidence.
The calculation in a compact form
For a conventional 14-period Wilder calculation, separate each closing-price change into a gain or a loss magnitude. Seed each average by dividing its first 14 values by 14. Thereafter, update each using its previous average multiplied by 13, add the latest value, then divide by 14. Divide average gain by average loss to obtain RS; RSI is 100 minus 100 divided by one plus RS. [StockCharts calculation documentation](https://chartschool.stockcharts.com/table-of-contents/technical-indicators-and-overlays/technical-indicators/relative-strength-index-rsi).
The following worked examples are original teaching calculations based on that formula. They are not copied price histories, backtests or examples of returns available to an investor. Keep a calculator beside you and check the divisions yourself; understanding the inputs is more useful than memorising a coloured zone on a chart.
A first reading using invented changes
Imagine 14 successive changes between closing prices. Seven are gains of two units each, and seven are losses of one unit each. The total gain is 14 units; the total loss magnitude is seven units. The first average gain is therefore one, and the first average loss is one half.
RS is one divided by one half, or two. Substituting into the final formula gives 100 minus 100 divided by three: approximately 66.67. Nothing in that result says what the fifteenth change will be. It summarises the relationship between the gains and losses supplied to the calculation.
Notice what would happen if every change in this example were doubled. Total gains would become 28 and total loss magnitude 14. Both averages would double, but their ratio would remain two, leaving RSI unchanged. This follows directly from the arithmetic. The indicator does not tell you the absolute size of a potential cash loss merely because it has a standardised scale.
There is another small but important counting detail. Fourteen price changes require 15 consecutive closing observations. The first change needs a starting close and the next close. If you enter only 14 prices into a fresh worksheet, you have only 13 differences. Count the changes, not just the visible rows of prices.
Updating the example one period later
Continue from the average gain of one and average loss of 0.5. Suppose the next invented change is a gain of three units. Its loss component is zero. The new average gain is 16 divided by 14, or approximately 1.142857. The new average loss is 6.5 divided by 14, or approximately 0.464286.
The updated ratio is 16 divided by 6.5, approximately 2.461538. The resulting RSI is about 71.11. Crossing the familiar 70 line in this exercise is a mathematical consequence of the added observation. We have not supplied any information about future demand, company earnings, execution costs or the next closing price.
Now imagine a different next change from the same starting averages: a loss of three units instead. The updated average gain would be 13 divided by 14, and average loss 9.5 divided by 14. That produces RSI of about 57.78. These two branches demonstrate sensitivity to the next observation; they are not alternative forecasts for a real instrument.
The practical lesson is to preserve a calculation's starting state. If two people update different starting averages with the same new price change, their answers need not match. Writing down the inputs makes a disagreement easier to diagnose than arguing about which screen looks more authoritative.
Why two charts can show different numbers
A setting labelled 14 is incomplete without its interval. Fourteen daily bars and 14 hourly bars describe different sequences. A daily close is not a collection of hourly closes inserted unchanged into the same calculation. Changing the interval changes the observations, so there is no requirement for the resulting readings to agree.
Check the instrument as carefully as the interval. Two similarly named charts might represent different contracts, listings or constructed series. Even before considering an indicator, confirm what the price line means. For a Swiss reader, the fact that a topic is trending locally does not make every chart quoted in the discussion a Swiss asset or a Swiss-franc series.
For a reproducible comparison, record the provider, symbol, interval, period setting and timestamp. Note whether the latest bar is complete. A value calculated while a bar is still changing may differ from the value after that bar closes. Comparing screenshots taken at different moments can create a false impression of an error.
Historical starting data also matter. StockCharts explains that smoothing and the available calculation history can affect the result. [StockCharts explanation of smoothing](https://chartschool.stockcharts.com/table-of-contents/technical-indicators-and-overlays/technical-indicators/relative-strength-index-rsi). If all visible settings match but values differ, consult the providers' documented methods rather than adjusting the numbers by hand until they look alike.
Overbought does not mean the move is finished
Schwab's educational discussion warns that markets can remain at extreme technical readings and that those labels alone are not buy or sell instructions. It distinguishes such readings from fundamental assessments and emphasises market context. [Schwab on overbought and oversold markets](https://www.schwab.com/learn/story/how-to-tell-if-market-is-overbought-or-oversold).
Think about the wording of a claim. “The indicator is above its reference line” is an observation. “The price must fall tomorrow” is a prediction. A colourful chart can make the transition between them seem effortless, but the second sentence does not follow logically from the first.
The same caution applies below 30. A low reading is not a certificate that a security is cheap or safe. RSI does not contain a company's debt obligations, a fund's mandate or an individual's capacity to absorb losses. Those subjects need their own analysis. Treating an indicator as a shortcut around them changes the question without answering it.
This article offers no threshold-based trading strategy. Its purpose is to make financial commentary easier to read and numerical claims easier to audit. Understanding a tool includes knowing which questions it cannot settle.
Relative strength is not always RSI
Infront's documentation distinguishes Wilder's RSI from comparing one security with another security or index. [Infront's indicator documentation](https://infront-portfolio-manager.helpcenter.infront.co/en/pm/latest/relative-starke-index-nach-welles-wilder). The similar wording can conceal a major difference in inputs.
StockCharts describes relative-strength ratio charts built from two datasets, such as a security divided by a benchmark. [StockCharts on ratio and difference symbols](https://help.stockcharts.com/data-and-ticker-symbols/ticker-symbols/ratio-and-difference-symbols). A ratio comparison and an RSI calculation are not interchangeable simply because both appear beneath a price chart.
Here is an invented ratio example. Suppose asset A moves from 100 to 105 while benchmark B moves from 100 to 110. A has risen in absolute terms, but A divided by B has fallen from one to approximately 0.9545. It has lagged that benchmark over the example period. That observation does not provide A's RSI; we have not supplied the sequence needed to calculate it.
When someone says a security shows “strong relative strength,” ask which definition they mean. Requesting the chart label is often enough to clarify the conversation. Otherwise, two people can discuss different calculations while believing they are disagreeing about the same evidence.
Be especially careful with a spread headline
The October 23 commentary linked from the trend feed discusses a Brent–WTI spread and monthly RSI. It is attributed market analysis, not a verified instruction about future prices. [The original FOREX.com commentary](https://www.forex.com/ie/news-and-analysis/brent-wti-spread-hits-11-year-wall-and-rsi-hits-2011-highs/).
A difference between two series can change because either component changes. For a simple invented illustration, A at 90 and B at 80 produces a difference of ten. If A remains at 90 while B falls to 75, the difference widens to 15 despite A not rising at all. Reading only the difference would not reveal the full movement of both components.
Before interpreting an indicator on a constructed series, identify the construction. Is it a subtraction, a ratio or something else? Which component comes first? Reversing a subtraction reverses its sign; replacing subtraction with division changes the calculation entirely. The words “spread” and “relative” are not substitutes for a clear definition.
This guide does not independently validate the oil commentary's historical-high claim. It uses the headline as a reason to explain the indicator, while keeping price forecasts and trade recommendations outside its scope.
A worksheet for checking a claim
Use a short record that another person could reproduce. Start with the source link and the date you read it. Then copy the symbol and settings exactly, keeping your observations separate from the writer's interpretation.
- Identify the underlying series, including whether it is a single instrument or a calculation involving two.
- Record the chart interval and RSI look-back setting.
- Note the timestamp and whether the latest observation is complete.
- Write the stated reading without converting it into a probability.
- Separate the observed value from any forecast attached to it.
- Record missing information instead of silently assuming a default.
A useful final line is “What would allow someone else to check this?” If the answer requires an unavailable chart, hidden settings or a missing timestamp, the claim is not yet reproducible from the material you have. That does not prove it false; it identifies a gap that should remain visible.
Common calculation mistakes
First, do not average only the positive days when seeding average gain, or only negative days when seeding average loss. In the 14-change example, both totals were divided by 14. Dividing by seven instead happens to preserve the initial ratio there, but produces different starting averages and can change subsequent smoothed updates.
Second, keep losses as positive magnitudes in the loss column. A downward change of minus three contributes a loss magnitude of three. Inserting a negative average loss into the ratio would no longer be the intended calculation.
Third, avoid rounding every intermediate value to a single decimal place. Preserve precision while calculating and round the final displayed result. In the worked update, the fractions 16/14 and 6.5/14 are more reliable than prematurely shortening them to 1.1 and 0.5.
Finally, document zero-value handling. A completely flat example can leave both averages at zero, making their ratio undefined mathematically. Check the chart provider's implementation instead of assuming that every system will display an identical value for that edge case.
Conclusion: read the inputs before the signal
RSI can organise a price history into a convenient number, but convenience is not certainty. Its meaning depends on the series, interval, smoothing and observations supplied. The worked examples show how the arithmetic changes; they cannot tell you what a market will do next.
For a reader encountering RSI in a headline, the strongest next step is verification: identify the chart, reproduce the settings and distinguish a description from a forecast. That habit is useful well beyond this one indicator.
Editorial note
Researched on 25 October 2026. Educational information, not personalised investment advice. All numerical examples are invented and calculated for this explanation; no returns, trading performance or live prices are claimed. The original generated illustration is conceptual, not a market-data screenshot. Source links support the definitions and distinctions above; examples and the checking worksheet are original explanatory material.