Stochastic oscillator: how to read it, what divergence is, and what happens when you trade by it alone

The Stochastic Oscillator measures where the last close sits within the price range of the recent period (usually 14 candles), on a scale of 0 to 100. A close at the top of the range gives 100, at the bottom 0. Above 80 the price is "overbought", below 20 "oversold". The two lines on the chart are the fast line (%K) and its average (%D), and the classic signal is a crossover between them in the extreme zones. Simple, long-standing (George Lane, the 50s), and that's exactly why it appears in so many ads.

When we ran its classic crossover signals on 184 stocks over the last three years, they gained +19% while the index rose +95%A full explanation of the oscillator, of the divergence, and of why the number came out this way, below.

The formula in one sentence, and the definitions

%K = 100 × (the close minus the low of 14 candles) divided by (the high of 14 candles minus their low). In the "slow" version, the common one, %K is smoothed by an average of 3 to get the fast line; %D is a 3 average of that line. The setting "14,3,3" in charting software is exactly that.

Shorter settings (5,3,3) react faster and produce more signals and more noise. Longer ones are slower. There's no "correct" setting, only a trade-off for each choice.

The three common signals

• Crossover in an extreme zone: The fast line crosses %D upward below 20 (buy), or downward above 80 (sell). This is the signal drawn in every ad: a green circle below, a red circle above.
• Exit from an extreme zone: A return above 20 or below 80, without a crossover. Fewer signals, less early.
• Divergence: The price makes a new high and the oscillator makes a lower high (negative divergence), or a new low in price with a higher low in the oscillator (positive). This is a sign that the move is losing strength, not an instruction to act. Divergence can last for months while the price keeps going.

The measurement: the strategy from the ads, on real data

The rules we ran: buy when the fast line crosses %D upward below 20, sell when it crosses downward above 80 or at a stop of 8%; 10% of capital per trade, up to 10 positions, entries only when the S&P 500 is above its 200 average. The measurement: 184 American stocks and ETFs, daily candles, May 2023 to September 2026 (840 trading days), execution at the open following the signal, cost of 0.05% per side, long only and without leverage. The universe is today's and not 2023's — stocks that collapsed and disappeared are not included, and this biases every result upward. The market rose over most of the period.

• It's right most of the time, and doesn't profit. 57% success sounds great; an average profit of half a percent per trade, after costs, doesn't add up to anything. The success rate is the least important number in a strategy.
• It sells too early. "Overbought" in an uptrend is exactly the moment when most of the money is made, and the strategy exits there. 2024, the year of the chip stocks, ended for it in the red.
• It's not a standalone tool. The oscillator is useful as a filter: don't buy a breakout when the oscillator is already above 90, look for an entry on a pullback when it resets in an uptrend. As a standalone strategy it lost to the index by a huge margin.

How to use it correctly

Like RSIthe Stochastic tells you where the price is within its recent range, and that's all it knows. A logical combination: define the trend using Moving averagesentry only in the direction of the trend, and the oscillator only times the entry within the pullback. Divergence suits a warning signal on an existing trade, not opening a trade against the trend. On the TechAnalyzer Pro chart the Stochastic is available as an indicator, and the report flags RSI divergence when it exists, in exactly this capacity: a warning, not an instruction.