返回博客

指标

RSI 指标:完整交易指南

RSI 衡量的是动量,而不是方向。 最适用于结构的发散和汇合。

Visual guideRSI is context, not an automatic reversal
RSI is context, not an automatic reversalPrice making a higher high while RSI makes a lower high between the 30 and 70 reference levels.
How to read itDivergence is a warning that momentum changed, not an entry by itself. Confirm it with price structure, location, invalidation, and a complete risk plan.

The Relative Strength Index, or RSI, is a momentum oscillator that compares recent gains and losses on a scale from 0 to 100. It can help describe whether momentum has been strong or weak. It cannot tell you that price must reverse, so it should never be used as an entry signal by itself.

What RSI measures

The common setting uses 14 periods. RSI rises when recent gains outweigh recent losses and falls when losses dominate. Shorter settings react faster but can become noisier; longer settings move more slowly. The setting matters less than using one consistently while you test it.

How is RSI calculated?

RSI uses changes between consecutive closing prices. For each period, separate an upward change into a gain and a downward change into a loss, expressed as a positive amount. An unchanged close contributes zero to both.

For the first 14-period reading, average the 14 gains and the 14 losses separately. Then calculate:

  • Relative strength (RS) = average gain ÷ average loss.
  • RSI = 100 − 100 ÷ (1 + RS).

Suppose the average gain is $1.50 and the average loss is $0.50. RS is 3, so RSI is 100 − 100 ÷ 4 = 75. That describes the strength of recent closing-price gains. It does not mean there is a 75% chance of a rise or that a fall is due.

After the initial calculation, Wilder's method smooths each average: multiply the previous average by 13, add the latest gain or loss, then divide by 14. It does not simply discard the oldest observation and recalculate a rolling mean. Different smoothing methods or available price history can therefore produce slightly different readings across platforms.

With gains but no average loss, RSI reaches 100; with losses but no average gain, it reaches 0. If both averages are zero, the formula is undefined and charting software must choose how to display the flat period. Check the platform's handling before comparing results.

What does a 14-period RSI setting mean?

Fourteen periods means 14 bars on the selected chart, not always 14 days. On a daily chart, it uses daily closing-price changes; on an hourly chart, it uses hourly changes. Keep the timeframe, price source and smoothing method fixed when comparing examples. A daily RSI and a five-minute RSI answer different questions.

Overbought and oversold are context labels

Many charts mark 70 as overbought and 30 as oversold. These levels mean momentum has been unusually strong or weak relative to the chosen lookback. In a strong uptrend, RSI can remain above 70 while price continues higher. In a strong downtrend, it can stay below 30.

Use market structure to decide whether you are looking at a trend, range, or transition before interpreting those levels.

In a range, RSI extremes near established support or resistance may be worth observing for a reversal setup. In a trend, traders may instead watch whether RSI holds a higher range in an uptrend or lower range in a downtrend. These are hypotheses to test, not rules that override price.

Understanding divergence

Bullish divergence describes price making a lower low while RSI makes a higher low. Bearish divergence is the reverse. Divergence can warn that momentum is changing, but it can persist for a long time and can fail completely. Require price confirmation, a defined invalidation, and sensible size.

A cautious RSI workflow

  1. Identify higher-timeframe structure and key levels.
  2. Note RSI condition without treating it as a command.
  3. Wait for price action that matches your written setup.
  4. Define invalidation from price structure, not merely an RSI cross.
  5. Record the context and result in a journal.

The support and resistance guide helps provide the price-level context that RSI lacks.

RSI vs Stochastic: what is the difference?

RSI compares the size of recent closing-price gains with losses. The Stochastic Oscillator compares the latest close with the recent high-low range. Both usually run from 0 to 100, but their numbers describe different things. An RSI of 75 and a Stochastic reading of 75 are not interchangeable.

Adding both may repeat similar momentum information. If you compare them, test the same market, timeframe and written setup, then record whether the second indicator changes a useful decision. Neither replaces a price level or a position-sizing calculation.

Is RSI above 70 a sell signal?

No. An RSI above 70 describes strong recent momentum under a common threshold convention. A trending market may keep rising while that reading remains high. Before treating it as part of a reversal setup, define the price confirmation and invalidation you will require. The same caution applies to buying simply because RSI is below 30.

Avoid indicator overload

Adding several oscillators that measure the same thing does not create independent confirmation. Keep the chart simple enough to see price, levels, and risk. Test one RSI rule with paper trading over different conditions before changing settings or adding filters.

When reviewing, distinguish an RSI observation from a completed setup. “RSI was low” is not enough information to learn from. Record the level, trend, entry confirmation, stop location, and whether the trade followed your rules. This shows whether RSI added useful context or only encouraged more activity.

Settings are a test decision

Changing RSI from 14 periods to a shorter setting may create earlier readings, but it also changes the rule. Do not select settings only because they make a past chart look perfect. Fix them before collecting examples, include losing periods, and compare results after fees and realistic execution assumptions.

Avoid the reversal trap

The most common RSI error is selling simply because it is high or buying simply because it is low. Strong momentum can persist. Ask whether price has actually changed structure or confirmed a level before considering a trade. If not, the correct action may be to keep observing.

For a hypothetical range, RSI near 30 while price tests established support may prompt a trader to watch for a price rejection and a higher low. If support fails, the oversold reading is not a reason to hold longer. In a clear uptrend, the same RSI reading could be a routine pullback rather than a reversal. Write which context you are testing so different situations are not mixed together in the results.

Avoid using divergence as an excuse to fight a strong trend. It can be early, absent, or wrong; price invalidation remains the risk decision.