Stochastic Oscillator: What the Settings Actually Change

Two numbers decide what a stochastic oscillator shows you, and neither of them is the 14 that every settings guide leads with. One controls how much the fast line is smoothed before you ever see it. The other decides whether the range being measured comes from the highs and lows of each bar or from the closing prices alone.

MetaTrader documents both. The explanations that rank for this indicator document neither. That gap is the reason two traders can run what they believe is the same setting and read different charts.

What follows is the settings layer: the four values MetaTrader exposes, what each one changes, and where the standard reading of the result stops holding.

The Four Settings MetaTrader Actually Exposes

The indicator draws two lines. The main line is %K, and the second line, %D, is a moving average of %K. The calculation behind those two lines is set out on our page covering how K and D are calculated, which starts from the same MetaTrader definition, so this article stays on what the settings do rather than re-deriving the formula.

The MetaTrader help lists four variables for the calculation. Three of them are periods and one is a method. The %K period is the number of bars the range is taken over. The %K slowing value controls the internal smoothing applied to %K before it is plotted. The %D period is the number of bars used to average %K into the signal line. The %D method is which kind of moving average performs that averaging.

Most published settings advice names two of the four. It gives a lookback and a signal period, and treats everything else as fixed. It is not fixed. Two of the four are live choices that the platform will let you change and that will move the line.

Fast and Slow Are the Same Indicator at Two Slowing Values

The published treatments present fast, slow and full as three variants, each with its own description. The MetaTrader documentation does not describe three indicators. It describes one indicator carrying a slowing value, and it ties both labels to that single number: set slowing to 1 and what you have is the fast form, set it to 3 and it is the slow form.

That single sentence reorganises the whole subject. The difference between the fast version and the slow version is not a different formula or a different pair of lines. It is one number, and the platform exposes it directly. A reader who has spent an afternoon deciding between fast and slow has been deciding how much smoothing to apply to %K.

The practical consequence shows up when a setting is copied. A guide written around a named variant carries an implied slowing value that the guide never states, because in its vocabulary the variant name carries it. Entered into MetaTrader, where there is no variant name to select, that implied value has to be supplied by hand, and the obvious thing to do is leave whatever was already there.

This is also why the same indicator can look materially different on two charts described identically. Smoothing removes the shallow reversals first, so a higher slowing value produces fewer crossings and later ones. That is the same trade-off that governs momentum indicators generally, and it is a choice about responsiveness rather than about accuracy.

The Price Field Setting That Changes What the Range Means

The MQL4 reference for the stochastic function lists a price field parameter with two permitted values. One measures the range from the low and the high of each bar. The other measures it from closing prices only.

This is the setting with the largest effect on the reading and the smallest presence in the literature. Across the four comparable pages fetched for this article, the terms describing it returned zero occurrences on every one of them. The parameter exists, it is documented, it is exposed in the function signature, and the pages a reader will find do not mention that a choice is being made at all.

The difference is not cosmetic. A bar range taken from highs and lows includes every wick, so a single spike widens the denominator and pushes the reading toward the middle even when closes have barely moved. A range taken from closes ignores those wicks entirely. On an instrument that prints long wicks around news, the two settings can disagree about whether the market is near the top of its recent range.

Neither setting is the correct one. They answer different questions: one asks where the close sits within everything that traded, the other asks where it sits within where the market settled. What matters is knowing which question your chart is answering, because a level compared across two charts using different price fields is not a like for like comparison.

What 14-3-3 and 5-3-3 Set, Position by Position

These two shorthand settings appear constantly in discussion and are almost never expanded. Neither string appears anywhere across the four comparable pages read for this article, and readers ask about both.

Taken against the MetaTrader parameter list, the three numbers are a %K period, a %D period and a slowing value.

So 5-3-3 sets a five bar range, a three bar average for the signal line, and a slowing of three. The worked example in the MQL4 reference uses exactly those three values in that order, with the price field set to the bar extremes.

The trap is the order. Other platforms and other authors write the same three numbers in a different sequence, most often putting the slowing value in the middle rather than last. Read that way, 14-3-3 is a fourteen bar range with a slowing of three and a three bar signal, which happens to produce the same chart only because the last two numbers are equal.

Where they are not equal, the two readings diverge. A setting written as 14-1-3 means one thing if the middle number is the signal period and something visibly different if it is the slowing value, and nothing in the string itself resolves which. The only reliable move is to read the labels in the settings dialog rather than the positions in the shorthand.

The %D Averaging Method Is a Choice, and Guides Assume One

The fourth documented variable is the method used to average %K into %D. MetaTrader offers four: simple, exponential, smoothed and weighted. The default is a simple moving average, and almost every published description of the signal line assumes that default without saying so.

An exponential or weighted average places more weight on recent bars, so the signal line turns earlier and the crossing between %K and %D happens sooner. A trader following a crossover rule has therefore changed the rule by changing a dropdown, without changing either period.

The reason this matters more here than on some other indicators is that the stochastic is frequently traded on the crossing rather than on the level, so the signal line is not decoration. It is half the trigger. This is the same structural point that applies when the oscillator is paired with RSI, where two momentum tools reading the same price can appear to confirm each other while differing only in how they smooth.

Where the 80 and 20 Reading Stops Working

The oscillator answers one question: where did the close finish inside the recent range. A reading near 100 means the close landed near the top of that range. It says nothing about whether the range itself is high, low, rising or falling.

That is why the standard reading fails in a trend. In a sustained advance each bar closes near the top of a range that is itself moving up, so the oscillator sits pinned in the upper zone for as long as the move lasts. Read as a reversal signal, it produces a sell instruction on every bar of an advance, and repeats it until the advance ends for reasons the indicator never measured.

The MetaTrader help describes the level readings as a sequence rather than a state: the interest is in the oscillator falling below a level and then rising back above it, not in it sitting there. That ordering is what makes overbought and oversold levels usable at all, and it is routinely compressed into a claim that a number alone is a signal.

Checking Your Own Chart Before Copying a Setting

Four checks settle what a chart is actually running, and all four are read from the settings dialog rather than inferred from a screenshot.

1. Read the three period boxes by their labels and note which one is slowing, rather than assuming the middle number.
2. Check the price field, since it decides whether wicks are inside the range or excluded from it.
3. Check the averaging method if the plan uses crossings, because it moves the crossing point.
4. Compare against a second setting on the same instrument before treating either as settled, which is the approach behind a strategy built on two stochastics.

Questions Readers Ask About the Stochastic Oscillator