Why RSI 70 and 30 Are Not Automatic Sell and Buy Signals
Understand how RSI compares gains and losses and defines overbought and oversold zones. Examples explain why extreme readings can persist in strong trends and how averaging methods differ.
RSI compares upward and downward price changes over a specified period and expresses the result on a scale from 0 to 100. Readings above 70 and below 30 are conventionally called overbought and oversold, respectively, but these thresholds describe the state of price changes. Overbought does not mean overvalued, and oversold does not mean undervalued.
Compare the size of changes, not just the number of up days
The core RSI ratio is the smoothed average gain divided by the average loss, transformed using 100−100÷(1+ratio). Knowing that there were many up days is not enough to determine the reading precisely: one large down day can offset several small up days. Standard implementations separate gains and losses and smooth each, treating losses as positive magnitudes. Indicators with the same period can produce different readings if smoothing methods or initial values differ, so do not assume the result always equals a ratio of simple totals.
Persistent extremes can also reflect a trend
When prices rise repeatedly, the average gain can dominate and RSI can remain high for a long time. Assuming an imminent reversal merely because RSI first crosses 70 can miss the continuation of a strong trend. The same applies to low readings: a sharp decline can be followed by further losses. In practice, record recent RSI highs and lows, price highs and lows, and time spent in extreme zones alongside threshold crossings. A decline from a high RSI reading and an actual reversal in the price trend are different events.
Changing the period changes the question
A short period is sensitive to nearby changes, while a longer period combines changes over a broader window. Continually adjusting the period and thresholds to match a stock’s past lows does not establish a rule that will work in the future. First choose the observation period and whether signals require a final close, then compare bull and bear markets using the same rule. RSI does not use revenue, cash flow, or credit risk as inputs. If a low reading coincides with deteriorating business performance, keep chart-based oversold conditions separate from judgments about business value. Understanding the formula’s inputs takes priority over the label attached to the reading.
Worked example
Hypothetical example: If the average gain and loss calculated with the same smoothing method are 3 won and 1 won, the ratio is 3 and RSI is 100−100÷4=75. If the average gain later becomes 2 won while the average loss is 1 won, RSI is about 66.67. These two readings alone cannot tell you how far price fell or the probability of a rebound the next day. Keep the averages used and the closing-price series together to identify what caused the indicator to change.
What to check
- Check the period and smoothing method.
- Do not interpret 70 and 30 as valuation thresholds.
- Consider time spent in extreme zones alongside the price trend.
- Distinguish RSI on an unfinished bar from a confirmed signal.
Sources and primary materials
This is educational material about research methods. For an actual company decision, verify the filings and primary sources for the relevant period.