Stochastic describes a Close relative to a recent price range, or a smoothed version of that calculation. A reading of 85 means neither an 85% chance of a fall nor an instruction to sell. Check the line, smoothing convention, data and closed-bar status before interpreting it.

Raw position and smoothed position are different

For a window with low L, high H and H > L, define raw R = 100(Close − L)/(H − L). Raw 0 means Close at L; raw 100 means Close at H. A zero-width window requires an explicit implementation rule. A smoothed line need not equal the latest raw position.

The names %K and %D do not fully specify their construction. TradingView’s documented %K smooths the raw ratio with an SMA; %D then averages that %K. MetaTrader exposes slowing and an averaging-method setting. Compare named fields and the actual algorithm, not just the label. The mechanics of %K and %D examines those differences.

The same Stochastic 80 can describe different distances

One range has a low of 100, a high of 110 and Close 108. Another has a low of 100, a high of 150 and Close 140. Both raw readings equal 80, but Close is 8 or 40 price units above the low.

Scroll sideways to see the full diagram

Equal readings in narrow and wide rangesOn a shared distance scale, the first range is 10 units wide and the second 50. Both closes lie 80% along their ranges, despite the fivefold width difference.80 points, two different rangesA: low 100 · high 110C=108B: low 100 · high 150C=140A: (108−100)/10 = 80%B: (140−100)/50 = 80%Distance above the low: 8 or 40
Both segments use the same price-unit scale. The second is five times wider.

Stochastic normalises the close’s position within the selected range. Comparing absolute range width requires another number. Equal readings of 80 do not imply the same stop distance or monetary risk.

Assumptions: Fictional ranges and prices using the same unit. We calculate the raw result without K or D smoothing and without assigning it to a real instrument. Formula or data verification: MetaTrader 5: Stochastic formula.

Threshold, crossing and divergence: specify the event

Traditional 80/20 zones describe a selected indicator line. StockCharts notes that extreme readings can persist in a trend; they do not establish fundamental value or an inevitable turn. Above 80 excludes exactly 80; at least 80 includes it. An unrounded value of 79.96 can display as 80.0 while still failing an above-80 rule.

%K > %D is a current relation, not necessarily a new crossing. An upward crossing could require previous K ≤ D and current K > D. State the equality rule and compare like-for-like observations. A transient open-bar crossing is not the same event as a final closed-bar crossing.

Divergence compares specified price and indicator extremes. Define how points are paired and when they can be recognised; it is not automatically the strongest signal or a probability of reversal. An ADX threshold or the label consolidation also does not certify a successful trade.

Why smoothed 85 need not mean Close in the top 15%

Assume three consecutive closed EUR/USD observations. Each corresponding 14-bar window has low 1.1000 and high 1.1200. These are assigned window parameters, not a supplied full OHLC market history. Smooth the three raw readings with an SMA and do not calculate an additional %D here.

Fixed window bounds, three assigned closes
ObservationCloseRaw RSMA3 of shown R
t−21.119095Not calculated here
t−11.119095Not calculated here
t1.11306585

The latest raw result is 100×0.0130/0.0200 = 65, while (95 + 95 + 65)/3 = 85. Close fell 0.0060 USD per EUR, or 60 pips with pip 0.0001. It is 65% up the range, outside the top 20%, although the smoothed line remains above 80. Neither result predicts the next candle or a profitable short.

RSI instead compares smoothed gain/loss magnitudes. Sharing a 0–100 scale does not establish equal meanings or that RSI universally makes fewer errors and Stochastic always signals sooner. Define the configuration and error criterion before comparing them.

A rising Close can lower raw Stochastic

In a separate hypothetical window, keep L = 1.1000. Initially H = 1.1200 and Close = 1.1180 give R = 90. A new candle raises Close to 1.1190 but sets High = 1.1400; the former low is assumed to remain in the window. The new raw result is 100×0.0190/0.0400 = 47.5.

Price change and reference-range change are separate
QuantityBeforeAfter
Close1.11801.1190
Window Low / High1.1000 / 1.12001.1000 / 1.1400
Raw R9047.5

Close rises 10 pips and R falls 42.5 points. For fixed L and positive C − L, this happens when ΔH > ΔC(H − L)/(C − L). Here the High increase must exceed 11.11… pips; it is 200. At exactly that boundary, R would stay unchanged. Smoothing adds another dependence on earlier raw values.

Write settings in words and retain information time

TradingView documents K window 14, D period 3 and additional K smoothing 3 as defaults; a smoothing period of 1 disables that extra averaging. A tuple 14,3,3 without field names, algorithm and input source can be incomplete. Periods count bars, not necessarily days.

Record provider, symbol, price series, interval, time zone, averaging methods, missing-value policy and bar status. Compare identical observations before changing parameters. Selecting settings until past failures disappear does not evaluate unused data; a default is not an optimum.

A trading rule needs entries, exits, execution, costs and no-trade conditions. A plotted value, a line crossing and a decision are separate stages. Calling the setting a range or trend does not guarantee an outcome, and three selected divergences do not validate a strategy.

Reproduce the model before testing a trade

Change only the last Close in the fixed-bound table to 1.1160. Its raw R becomes 80; SMA3 of 95,95,80 is 90. The raw result is exactly at 80, not above; the smoothed line is above it. No number alone is a sell instruction.

The diagram separately compares [100,110] with Close 108 and [100,150] with Close 140. Both raw results are 80, yet distances from the low are 8 and 40 units. The quiz uses [100,120] and Close 116: raw 80 and distance 16. Equal normalised values do not imply equal price distances or monetary risk.

Use the technical analysis section on ForexMechanics for related reading. Keep a record of the rule, data and calculation first, then evaluate any trading hypothesis separately.