Skip to main content

MACD for Swings

Updated Jun 24, 2026 at 2:35pm

Research Draft High 1,215 words

MACD (Moving Average Convergence/Divergence), built by Gerald Appel in the late 1970s, is a momentum-trend hybrid: it subtracts a slow EMA from a fast EMA to measure whether trend momentum is accelerating or decaying. For the swing trader holding days to a few weeks, MACD's appeal is exactly its timeframe fit — it is slow enough to filter intraday noise but fast enough to flag the inflection where a multi-day move is gaining or losing fuel. Its core tension is lag versus reliability: every signal is built from moving averages, so by construction MACD confirms what price has already begun to do. Used as a trigger it whipsaws; used as a confirmation and divergence tool it earns its keep.

How it's calculated / formed

MACD has three plotted components, all derived from the daily close (StockCharts ChartSchool; Wikipedia):

  • MACD line = 12-period EMA − 26-period EMA. Positive when short-term price is above the medium-term mean (momentum up), negative when below.
  • Signal line = 9-period EMA of the MACD line. A smoothed lag of the MACD line itself.
  • Histogram = MACD line − Signal line. Added by Thomas Aspray in 1986 to anticipate signal-line crossovers. Bars grow when momentum accelerates, shrink as it stalls, flip sign at the crossover.

The conventional (12, 26, 9) defaults are Appel's originals. They are not magic — the MDPI Nikkei 225 study found standard parameters produced negative returns there while optimized parameters earned positive ones, so the defaults are a starting convention, not an edge. Note MACD is unbounded (unlike RSI's 0–100): its absolute level is meaningless across stocks of different price/volatility; only its behavior relative to its own zero line and signal line carries information.

How it's used in practice (the swing setups)

Swing traders use MACD in three escalating modes, weakest-to-strongest as a standalone trigger:

1. Zero-line (centerline) crossover — regime filter. MACD crossing above zero means the 12-EMA has overtaken the 26-EMA: a confirmed shift to a bullish medium-term regime (StockCharts). Swing use: a filter, not an entry — only take long pullback setups while MACD > 0. It lags the most but whipsaws the least.

2. Signal-line crossover — the classic entry confirmation. MACD crossing up through its signal line is the most frequent MACD signal. The high-conviction swing version stacks conditions: signal-line crossover while MACD is below but rising toward zero (early momentum in an oversold pullback) or just above zero in an established uptrend (continuation). Crossovers far above/below zero are late and prone to immediate reversal.

3. Divergence — the swing reversal tell. Price makes a lower low but the MACD line (or histogram) makes a higher low (bullish divergence), warning the down-leg is losing momentum. This is the setup most prized by swing reversal traders because it can precede the turn rather than trail it. The histogram, being the fastest component, gives the earliest divergence read.

Practical confirmation stack a master swing trader keys on, rather than acting on MACD alone:

  • Confluence with structure: take the signal-line crossover only when it coincides with a bounce off a rising 20/50-day MA, a prior support shelf, or a trendline.
  • Histogram as a tighten/exit cue: a long-held swing whose histogram peaks and starts contracting (bars shrinking) is bleeding momentum — a signal to tighten the stop or scale out before the signal-line crossover even fires.
  • Avoid the chop: suppress MACD signals when price is range-bound (e.g. flat 50-MA, low ADX) — this is where false crossovers cluster.

Common failure modes: in a strong trend MACD divergence can persist for many bars without a reversal (momentum slows but price grinds higher); in a tight range, signal-line crossovers fire repeatedly with no follow-through.

Adoption, debate & evidence

MACD is among the most widely used indicators in retail and professional charting; it ships as a default on virtually every platform. But popularity is not efficacy. The measured evidence is sobering for anyone treating raw crossovers as an edge:

  • A comparative study of MACD strategies on the Dow, Nasdaq and S&P 500 (2015–2021, arXiv:2206.12282) found the win rate of MACD alone is below 50%, improving only when combined with momentum tools like RSI or the Money Flow Index. The authors state performance "is not good without using other momentum indicators."
  • The MDPI Nikkei 225 futures study found standard-parameter MACD lost money; profitability required per-market parameter optimization — a finding consistent with curve-fitting risk.
  • Practitioner backtest aggregators (QuantifiedStrategies) report MACD performs best in volatile, trending markets and worst in flat ones, and emphasize wide variability. One often-cited "81% success rate" figure circulates online but conflicts directly with the peer-reviewed sub-50% findings and should be treated as unverified marketing, not evidence.

Bottom line on folklore vs measured: the concept MACD encodes — momentum acceleration/deceleration — is real and useful. The raw crossover signal as a standalone system has no demonstrated edge in the academic literature. Divergence is widely taught but, like all divergence signals, is hard to test mechanically and is best treated as a heuristic, not a backtested rule.

Strengths & limitations

Works best: as a confirmation/filter layer in a trending swing market, and as an early momentum-loss warning via histogram contraction. The histogram in particular is genuinely useful for managing an existing position.

Fails: in ranges (whipsaw crossovers), at trend extremes (late, reversal-prone signals far from zero), and whenever the absolute MACD value is compared across instruments. The #1 misuse is trading every signal-line crossover mechanically with no trend filter or structural confluence — the configuration the evidence shows loses. Secondary misuse: acting on a single divergence in a strong trend, where momentum can diverge for weeks before (or without) a turn.

Sources

Disputed/flagged: the ~81% "success rate" figure cited by some marketing sources contradicts the peer-reviewed sub-50% win rate — not used as fact. Default (12,26,9) parameters are convention, not a tested edge; divergence is a heuristic, not a mechanically backtested rule.