MACD was created by Gerald Appel in the late 1970s for stock charts on weekly and daily time-frames. The default parameters — 12-period and 26-period EMAs with a 9-period signal line — reflected the rhythm of those markets: equities that closed once a day, traded for ~250 days a year, and rarely moved more than 2–3% in a session.
Crypto is a different animal. Markets trade 24/7. Hourly candles produce ~720 candles per month. A 12/26-period EMA on hourly crypto data is reacting to roughly the last 1–2 days of price action, while crypto trends frequently complete in hours. By the time the standard MACD signals a cross, the move it was meant to catch is often half over.
This article explains how MACD is built, why the standard settings under-perform on crypto data, what the (8,21,5) "fast" variant looks like, and what the often-misunderstood histogram actually represents.
How MACD is constructed
MACD has three components, calculated in order:
Signal line = EMA(MACD line, signal_period)
Histogram = MACD line − Signal line
The MACD line measures the gap between a faster EMA and a slower EMA of price. When the faster EMA is above the slower EMA, MACD is positive — recent prices have been outpacing older prices, i.e. the trend is up. When the faster EMA is below the slower EMA, MACD is negative — recent prices have been falling faster than the slow average can react, i.e. the trend is down.
The signal line is just an EMA of the MACD line itself — a smoothed version of the MACD. The histogram is the gap between the two. When the MACD line crosses above its signal line, the histogram flips from negative to positive. This crossover is the canonical "MACD buy signal" most traders are taught.
Standard (12,26,9) vs Fast (8,21,5)
The arithmetic difference is straightforward. Faster EMAs react more quickly to recent prices, which means MACD itself reacts more quickly, which means crossover signals fire earlier — at the cost of more false positives.
| Setting | Lag | Best for | Trade-off |
|---|---|---|---|
| (12, 26, 9) — standard | ~2 days on hourly | Slow markets (equities, large-cap crypto on daily) | Robust but late on fast moves |
| (8, 21, 5) — fast | ~1 day on hourly | Crypto, faster intraday markets | ~30% less lag, more whipsaws in chop |
Stryqe's crypto scanner uses the fast variant, while the stocks scanner uses the standard variant. The split is not arbitrary — it reflects the underlying volatility of each market. Crypto's annualised volatility is roughly 60–100% (vs 15–20% for major equity indices), so price covers ground in a fraction of the time. A faster MACD keeps pace with that.
The crossover is not enough
The most common mistake with MACD is treating the crossover as a complete trading signal. It isn't. In any given week of price action, MACD on hourly crypto data will cross its signal line dozens of times — most of those crosses happen in chop, produce no follow-through, and lose money to friction.
Stryqe's engine deliberately treats the cross alone as not a buy. The signal generator only emits MACD = BUY on a fresh upward cross of the histogram from negative to positive, not on a positive histogram in general. The distinction matters: a histogram that has been positive for 20 candles is not a fresh signal; a histogram that just turned positive on this candle is. A cross is an event, not a state.
The engine compares the current histogram to the previous candle's histogram. Fresh BUY requires hist > 0 AND prev_hist ≤ 0. Fresh SELL requires hist < 0 AND prev_hist ≥ 0. A persistently positive histogram is neither — it's already fully priced in.
What the histogram actually shows
The histogram is the most information-dense part of MACD, and the most under-used. It's not a "buy/sell strength meter" the way it's often described. It's the second derivative of trend.
Think of it this way: the MACD line tells you the direction of trend. The signal line tells you the direction of trend, smoothed. The histogram tells you whether the trend is accelerating or decelerating. A growing positive histogram means the MACD line is pulling away from its signal line — the up-trend is gaining speed. A shrinking positive histogram means the MACD line is converging back toward its signal line — the up-trend is losing speed, even if the histogram is still above zero.
This is why divergence at the histogram level is more useful than divergence at the price level. If price is making higher highs but the histogram is making lower highs, the trend's acceleration is fading even while the trend's direction persists. Direction breaks last; acceleration breaks first.
MACD divergence — three flavours
Like RSI, MACD can diverge from price. Unlike RSI, you have to specify which line you're looking at:
- Price ↔ MACD line divergence: price makes a new extreme but the MACD line does not. The slower of the two divergence variants — captures longer-term momentum exhaustion.
- Price ↔ Histogram divergence: price makes a new extreme but the histogram does not. The faster of the two — fires earlier but more false positives.
- Hidden divergence: price makes a higher low while MACD makes a lower low (in an uptrend), or price makes a lower high while MACD makes a higher high (in a downtrend). This is a continuation signal, not a reversal — it suggests the existing trend is intact even though one of the two has briefly disagreed.
Stryqe's confluence engine does not currently emit a separate MACD divergence flag — that work is done by the dedicated RSI divergence layer, which is more reliable on hourly crypto data. MACD's job in the engine is the cross detection plus its contribution to the trend-confluence signal. The combined indicator load means a missed MACD divergence rarely changes the final ALIGNED tag, because at least one of the other 11 indicators usually catches the same underlying condition.
The OBV combo bonus
Inside the CII scoring layer, MACD has a known-good combo with OBV (volume-direction proxy):
THEN comboBonus += 12
The reasoning is causal: a fresh MACD upward cross says momentum has just turned positive; OBV BUY says recent volume is concentrated on up-moves; the combination says the momentum cross is being confirmed by actual buying pressure rather than thin-air price drift. Empirically, this combo has shown a higher hit rate than either signal alone — which is why it carries a +12 bonus to the net score, on top of the indicators' base weights.
Common MACD mistakes
1. Trading every cross
On a 1-hour chart, MACD will cross its signal line many times per week. Most of those crosses are noise. A cross taken without volume confirmation, trend context, or other indicator agreement is, on average, a coin-flip — and friction makes coin-flips a losing strategy. The cross is a starting point for analysis, not a signal in itself.
2. Using MACD as a level indicator
"MACD is at +5, that's high." There is no "high" or "low" for MACD in absolute terms. The MACD line's value depends on the price scale of the asset — Bitcoin's MACD might be +500, Solana's +5, a low-priced alt's +0.005, all describing the same trend strength relative to the asset. Always interpret MACD relative to its own recent range, not against fixed levels.
3. Ignoring the higher time-frame
An MACD buy on a 1-hour chart while the 4-hour MACD is firmly in bear territory is fighting the higher-time-frame trend. The hourly cross might produce a 2% bounce, but the broader move continues against you. Whenever possible, take MACD signals only when the higher time-frame agrees — Stryqe's engine builds this in via the daily-trend signal, which can veto entry signals from lower-time-frame indicators.
MACD, like every momentum-based indicator, assumes price changes follow continuous patterns. News events, exchange outages, regulatory announcements, and large liquidation cascades create discontinuities that MACD can't see coming. The indicator is a useful summary of recent price behaviour, not a predictor of macro events.
The bottom line
MACD is one of the most general-purpose indicators in technical analysis — it captures trend direction, trend strength, and trend acceleration in three numbers. The standard (12,26,9) settings are calibrated for slow markets; on hourly crypto, faster (8,21,5) settings catch moves earlier. The cross is an event, not a state — and even then, it's a signal worth pairing with volume and trend confirmation, not trading in isolation.
On its own, MACD is a 50–55% indicator. As one input in a 12-indicator confluence with combo bonuses for strong pairings, it earns a default weight of 10 out of 100 in Stryqe's CII — modest, but persistent value-add when the rest of the confluence agrees.