How to Read the Stochastic Oscillator: %K, %D and 80/20

What the stochastic measures: position within the range

The stochastic shows, on a scale of 0 to 100, where the current close sits between the highest high and lowest low of the last 14 bars. %K = (close − 14-bar low) ÷ (14-bar high − 14-bar low) × 100, and the high and low are taken from the highs and lows of the 14 bars including the current one. For example, if the 14-bar high is 110, the low is 90 and the close is 105, %K is 75. A reading of 100 means the bar closed at the 14-bar high; 0 means it closed at the 14-bar low. George Lane, who popularized the indicator, built it on the observation that while buyers have the upper hand, closes tend to finish near the top of the range, and while sellers dominate, they tend to finish near the bottom. In effect it summarizes who held the price at the end of the bar as a position within the range. One thing to watch is that the denominator is the 14-bar high and low, so a single long wick that stretches the range makes the readings look suppressed for the next 14 bars. If price sits at one value for all 14 bars so that the high equals the low, the denominator is zero and no value is produced; this occasionally happens in thinly traded markets.

%K and %D, fast and slow

%D is a 3-bar simple average of %K. Using the formula above as it is gives the fast stochastic; first averaging %K over 3 bars to smooth it gives the slow stochastic. In the slow version, the fast %D becomes the new %K, and a further 3-bar average of that is %D. Settings are usually written as three numbers: (%K period, %K smoothing, %D smoothing). TradingView's default is (14, 1, 3), the fast version, while the chart in this article uses the slow version, (14, 3, 3). Because the slow version averages once more, it is less jagged than the fast one but moves about one bar later. If two charts show different values, first check whether these three numbers match. Used as is, the fast version often jumps %K from 20 to 80 from one bar to the next and produces too many crossovers, so in practice many people use the slow version. Conversely, raising the smoothing a lot makes the line smooth but reports a touch of the range's edge several bars late. Whatever the setting, you cannot escape the trade-off of 'less jitter but more lag', and two stochastics with different numbers are different lines even if they share a name.

• Fast %K: the formula as it is (smoothing 1)
• Fast %D = slow %K: a 3-bar average of %K
• Slow %D: a 3-bar average of slow %K

How it differs from RSI

Both run from 0 to 100, but they measure different things. RSI uses only closes to measure the ratio of recent gains to recent losses, and because of Wilder's averaging, older moves linger a little (see the RSI article). The stochastic uses highs, lows and closes to measure position within the 14-bar range, and once 14 bars have passed, that bar's high and low drop out of the calculation completely. So on a bar where a big high leaves the window, the value can jump sharply even if the close has not moved, and it reaches 0 or 100 more often than RSI. On the same chart you can see RSI in the 60s while the stochastic is in the 90s. This difference affects interpretation too. A high RSI means recent gains were bigger than recent losses; a high stochastic only means the close is near the top of the range and says nothing about how strongly price rose. After a sideways stretch, even a small rise can push the stochastic near 100 if the range is narrow. When you show both and one is at an extreme while the other is in the middle, first check whether the range is narrow or whether an old large bar has just dropped out of the window.

The 80/20 lines and crossovers

By convention, above 80 is treated as overbought and below 20 as oversold. Some read a crossover, %K crossing %D up or down, as a sign the move is changing, and many give more weight to crossovers that happen above 80 or below 20. However, this course's measurements (past Binance bars) do not include the stochastic, so we cannot put a number here on how well these readings worked. RSI, which is similar in character, actually saw a higher-than-baseline share of rises after overbought bars (see the RSI article), but since the calculation is different, that result cannot be transferred directly to the stochastic. A commonly used order of checks for crossovers goes like this. First judge from the price chart whether the market is ranging or trending, then see in which zone the crossover happened, and finally check whether it was confirmed on a closed bar. The numbers 80 and 20 are a convention, not a law, and some people use wider levels on very volatile coins. Another approach is to watch not whether a line crossed the level, but whether it came back inside the level afterwards.

How it looks in a range and in a trend

While price oscillates within a range, the stochastic swings regularly between around 20 and around 80, and the crossovers of %K and %D occur at the top and bottom of the range. But once price breaks out of the range and starts rising steadily, closes keep finishing near the 14-bar high, so %K stays pinned around 80 and above. In the rising stretch of the chart below, %K crossed below %D above 80 several times, yet price only paused briefly and then rose again. The same levels and crossovers lead to completely different outcomes depending on the stretch. In a range, sellers near the top and buyers near the bottom keep showing up and price turns back, which suits an indicator that measures position within a range. In a trend, each new high keeps stretching the range upward and the close is always near the top, so staying above 80 is simply what the trend looks like. That is why many people first check whether highs and lows are rising and whether moving averages are sloping before looking at the stochastic. During a trend, some change how they read it, paying attention to crossovers in the other direction that come out of pullbacks rather than to overbought crossovers.

Stochastic RSI: an indicator stacked on another

Stochastic RSI applies the stochastic formula to RSI values instead of price. It shows, on a scale of 0 to 100, where the current RSI sits between the highest and lowest RSI of the last 14 bars, and TradingView's default is RSI 14, stochastic 14, %K smoothing 3 and %D smoothing 3. If RSI has stayed between 45 and 55 over the last 14 bars, RSI moving from 45 to 55 is enough to take the unsmoothed value from 0 to 100. It is that sensitive, so it hits 0 and 100 often and gets labeled overbought or oversold far more frequently. Stacking one indicator on another moves you further from the original price. A stochastic RSI of 100 does not mean price is high, nor that RSI is high; it means RSI is at its highest point of the last 14 bars. Even with RSI sitting in the 40s, the stochastic RSI can print 100. So when you look at this value, showing the raw RSI alongside it reduces mistakes. On short timeframes it flips between 0 and 100 every few bars and becomes close to noise, so note which timeframe you used, or comparisons will not make sense.

What this article does not say

The stochastic is a number that summarizes the close's position within the last 14 bars' range; it does not calculate anything outside that range or the direction ahead. Changing the period and smoothing values greatly changes how often the levels are crossed and when crossovers occur, so this article makes no judgment about which setting is better. This site's RSI Radar and Pro Trading Chart show RSI, not the stochastic. Putting them side by side with another chart that shows the stochastic lets you see how the two indicators summarize the same stretch differently. The pitfalls of testing settings on past data are covered in the backtesting article. A common mistake is treating anything above 80 as a place about to fall, because of the word 'overbought'. The name is just a convention; what the value tells you is one thing, that the close is in the upper part of the recent range. Also, because this indicator produces many crossovers, it is easy to remember only the few scenes that worked out. If you scroll back and count every crossover, including the ones that did not work, you can see for yourself how your impression differs from reality.

How it looks different in crypto and stocks

Because the stochastic uses highs and lows directly, gaps affect it heavily. Stocks often open the next day away from the previous close after the regular session ends, and on a day with a big gap up, that bar's high widens the 14-bar range at a stroke and changes the basis for %K for the next two weeks or so. Examples include Samsung Electronics or SK hynix gapping open the day after an earnings report, or a large US tech stock reporting after the close and opening far away the next day. Crypto trades around the clock, so such gaps are rare, but sudden moves that leave long wicks in a short time are common, so the same problem shows up as wicks. Korean stocks have a daily price limit, and on a day that closes at the upper limit the close equals the high, so %K sticks at 100; that means trading was blocked with buy orders still waiting, not that sellers absorbed buying at the top of the range. The length of 14 bars also differs. Fourteen daily stock bars are about three weeks of trading days, skipping weekends and holidays, while 14 daily crypto bars are two calendar weeks. Volatile altcoins swing between 0 and 100 far more often than large-cap stocks, so you need to build a separate feel for each market.

A practical checklist

On a live chart the stochastic swings a lot because of the forming bar. The close of the current bar is the current price, and if a new high or low appears, the 14-bar range itself changes, so %K can travel widely between 0 and 100 within a single bar. If %K crosses above %D during the bar and then drops back below it just before the close, that crossover never happened. The slow setting averages and so jitters less, but the last bar's value is still not final. Most statements that interpret indicators, including the chart in this article, are based on closed bars. On shorter timeframes, bars close and new ones open in quick succession, so if you do not separate confirmed crossovers from forming ones, signals seem to appear and vanish endlessly. Checking the screen in the following order helps reduce mistakes.

• Check that the three settings and the timeframe are what you intended
• First judge from the price chart whether the market is ranging or trending
• See where %K sits and in which zone (above 80, below 20, middle) a crossover came
• Count only crossovers confirmed on a closed bar
• Check whether a long wick or gap in the last 14 bars is stretching the range
• Show RSI as well and compare whether the two tell the same story
• Confirm that the move agrees with the higher timeframe