Schaff Trend Cycle

MACD run through two rounds of stochastic normalization — faster-turning than MACD, at the cost of more whipsaw.

What it is

The Schaff Trend Cycle (STC) is a momentum oscillator attributed to Doug Schaff that combines two familiar tools: it starts with a MACD line and then normalizes it twice using stochastic-style calculations. The result is a bounded 0-100 line that turns noticeably sooner than the MACD it is built from, usually read against reference levels around 25 and 75. The often-repeated claim is that STC is 'faster than MACD' — and in the narrow sense of reduced lag at turns, the double stochastic normalization does achieve that. What the marketing usually omits is the cost: normalizing an oscillator to its own recent range makes it react to every wiggle in that range, so STC trades lag for whipsaw rather than eliminating either. It also frequently pins at 0 or 100 for extended stretches, for the same reason StochRSI does. STC is a deterministic transformation of past prices — descriptive, not predictive. This is educational material, not financial advice, and all trading involves risk of loss.

How it works

STC is a four-stage pipeline. Stage one: compute a MACD line as the difference between a fast EMA and a slow EMA of price — the STC convention uses 23 and 50 rather than MACD's usual 12 and 26. Stage two: apply a stochastic calculation to that MACD line, locating it within its own highest-to-lowest range over a cycle length (commonly 10 bars) and scaling to 0-100. Stage three: smooth that value with a recursive averaging step (each bar moves the smoothed value a fixed fraction, typically 0.5, toward the new reading). Stage four: repeat stages two and three on the smoothed result — a second stochastic over the same cycle length followed by the same recursive smoothing. The final output is the Schaff Trend Cycle. Each stochastic pass re-expresses its input relative to its recent extremes, which is what makes STC snap quickly from one end of the scale to the other: once the MACD line reaches a new short-term extreme, the normalization saturates at 0 or 100 regardless of how large the underlying move actually is. The smoothing fractions keep the line from being pure noise, but the range-relative logic dominates its behaviour.

How traders read it

• Reference levels: readings above ~75 are conventionally described as an 'up cycle' extreme and below ~25 as a 'down cycle' extreme; crossings of these levels are the events most STC descriptions focus on.
• Saturation is normal: STC spends long stretches flat at or near 0 or 100 during sustained moves — that is the double normalization saturating, not extra conviction.
• Direction of turn: the line curling up from near 0 or down from near 100 describes the underlying MACD line reversing relative to its recent range.
• Compare with MACD: plotting STC and its source MACD together shows exactly what you gained (earlier turns) and paid (more false turns) — the trade-off is visible on any chart.
• In ranging markets, expect frequent full swings between 0 and 100 as the MACD line drifts inside a narrow band; each swing looks decisive but reflects small underlying changes.
• As always, these are descriptions of past momentum, not instructions to act.

Common settings

The standard parameters are fast EMA 23, slow EMA 50, cycle length 10, and a smoothing factor of 0.5 applied after each stochastic pass. These come from the original publication's convention, not from any optimality result. Shortening the cycle length makes saturation and whipsaw more frequent; lengthening it makes STC behave more like a smoothed MACD. Because there are four interacting parameters, small changes can alter signals substantially — a sensitivity worth testing explicitly rather than assuming.

Strengths

• Bounded 0-100 scale with consistent reference levels, unlike the unbounded MACD it is built from.
• Genuinely reduced lag at turns compared with the raw MACD line, thanks to the range-relative normalization.
• Fully decomposable pipeline — EMA difference, stochastic, smoothing, repeated — so every value can be reproduced and verified.
• Clear, discrete level-cross events that are easy to define and test objectively.
• A useful case study in indicator design: it shows concretely how normalization trades lag for sensitivity.

Pitfalls to watch

• Whipsaw is the price of speed: the same normalization that reduces lag amplifies small MACD wiggles into full-scale swings, producing many false turns in ranging markets.
• Pinning at 0/100: double stochastic normalization saturates easily, so extreme readings are common and carry less information than they appear to.
• Four stacked parameters (two EMA lengths, cycle length, smoothing factor) interact, making STC easy to curve-fit and hard to compare across sources.
• Implementation variance: published descriptions differ on smoothing details, so ports on different platforms can produce different values from identical data.
• Still lagging: STC is built entirely from past prices via EMAs; 'faster than MACD' does not mean early, and it does not mean predictive.
• No indicator, fast or slow, removes the risk of loss.

Pine v6 example

//@version=6
indicator("Schaff Trend Cycle Example", overlay = false)

fastLen  = input.int(23, "Fast EMA Length", minval = 1)
slowLen  = input.int(50, "Slow EMA Length", minval = 1)
cycleLen = input.int(10, "Cycle Length", minval = 1)
factor   = input.float(0.5, "Smoothing Factor", minval = 0.1, maxval = 1.0)

macdLine = ta.ema(close, fastLen) - ta.ema(close, slowLen)

// First stochastic pass on the MACD line, then recursive smoothing
loM = ta.lowest(macdLine, cycleLen)
hiM = ta.highest(macdLine, cycleLen)
stoch1 = 0.0
stoch1 := hiM - loM > 0 ? (macdLine - loM) / (hiM - loM) * 100 : nz(stoch1[1])
pf = 0.0
pf := nz(pf[1]) + factor * (stoch1 - nz(pf[1]))

// Second stochastic pass on the smoothed result, then smoothing again
loP = ta.lowest(pf, cycleLen)
hiP = ta.highest(pf, cycleLen)
stoch2 = 0.0
stoch2 := hiP - loP > 0 ? (pf - loP) / (hiP - loP) * 100 : nz(stoch2[1])
stc = 0.0
stc := nz(stc[1]) + factor * (stoch2 - nz(stc[1]))

plot(stc, "STC", color = color.blue)
hline(75, "Upper", color = color.gray)
hline(25, "Lower", color = color.gray)

Pro tip: Put STC in one pane and its source MACD(23,50) in another and step through a ranging period bar by bar. Count how many full 0-to-100 STC swings correspond to trivial MACD movement — that count is the honest measure of what the extra speed costs on that instrument and timeframe. Speed and reliability are a trade-off, not a free upgrade. Educational context only, not advice; all trading carries risk of loss.

Educational only — not financial advice, not a recommendation to trade. No indicator is predictive; trading involves substantial risk of loss.