If five indicators agree, how many opinions do you have?

A chart with RSI, stochastics, Williams %R, a momentum line and a MACD histogram all pointing the same way feels like overwhelming evidence. Traders call it confluence. It is usually one opinion, printed five times.

A proof you can do with a calculator

Take a 14-session window: the highest high was 120, the lowest low 100, and today's close is 116. It is drawn below, with those three numbers marked, so you can check the arithmetic against the bars rather than against my typing.

• Stochastic %K = 100 × (Close − Low₁₄) ÷ (High₁₄ − Low₁₄) = 100 × (116 − 100) ÷ 20 = 80
• Williams %R = −100 × (High₁₄ − Close) ÷ (High₁₄ − Low₁₄) = −100 × 4 ÷ 20 = −20

Now do the algebra rather than the arithmetic. Because (High − Close) = (High − Low) − (Close − Low), Williams %R is exactly %K − 100, for every instrument, on every bar, forever. Two names, two panes, two "confirmations" — and one number with a constant subtracted.

That is the clearest case, but it is not the exception. RSI, rate-of-change and the MACD histogram are all functions of the same closes; CCI and the stochastics reach into the same window's highs and lows as well. Different inputs, overlapping history — and on the same window their daily values typically correlate very tightly. You can measure it yourself, and you should.

Why the illusion is expensive

Correlated readings don't accumulate into evidence; they accumulate into confidence. That gap is where money goes missing. Adding a sixth momentum oscillator changes how you feel about the trade and changes nothing about what you know. Worse, the felt certainty tends to arrive right when it is least earned — in a fast, one-sided move every bounded oscillator pins near its ceiling at once while the unbounded ones stretch alongside them, and the chart looks unanimous.

What does add information

Different families and different inputs:

• A volume indicator reads a series that a price oscillator never sees.
• ATR measures dispersion, which says nothing about direction.
• A longer window answers a different question from a short one.

Even then, "different" is not "independent" — all of it descends from the same tape. An honest count of independent evidence on a single chart is usually one or two items, not five.

The professional habit

Name the question first, then allow exactly one indicator per question. Trend gets one line. Speed gets one. Dispersion gets one. Volume gets one. If a fifth pane appears, it should be because a fifth question appeared. And when several panes agree, describe it as it is: "three views of the same fourteen closes are consistent with each other" — which is a statement about arithmetic, not about tomorrow.

Try it now

1. Read the three marked numbers off the schematic above and compute both %K and Williams %R from them by hand. Confirm the second is the first minus 100, and then prove to yourself from the two formulas that it always is.
2. Now check the identity against numbers you did not choose. Take one settled bar, Apple on 25 September 2026, below. Over the fourteen sessions ending that day the highest high was 345.34 (22 September) and the lowest low 309.90 (9 September); find both on the month of candles with a Level if you want to see them. Compute %K and Williams %R by hand from those two and the close, and confirm the identity. Then compare with Apple's published stochastic, the third table: its %K will not match yours, because the standard published line is the slow %K, the three-session average of the fast one. Compute fast %K for 23 and 24 September too (the fourteen-session high is the same 345.34 on both days and the low the same 309.90; the closes were 337.02 and 335.92), average your three, and you land on the published figure for the 25th.

3. Take any chart of your own, remove two of your momentum panes, and ask honestly whether it now tells you less. Keep whichever one you can defend with a sentence.