Triple Exponential Moving Average (TEMA)

An aggressive lag-reduction moving average — three-level EMA correction. Faster than DEMA, noisier than EMA. For short-term execution where reaction speed matters most.

Updated 2026-05-24
7 min read
intermediate

The fastest of the traditional moving average family — three EMA passes combined into a single line that lags real price by almost nothing. Use when reaction speed dominates every other consideration.

Placeholder · Video / Clip

EMA vs DEMA vs TEMA — three speeds at the same period

Author hint: MES 5m, dark theme, ~50-bar window. EMA(14), DEMA(14), TEMA(14) overlaid as three lines with distinct colors (cyan = TEMA, mid-tone = DEMA, dim = EMA). Show one sharp price reversal where the three lines visibly stack from fastest to slowest. ≤ 8 seconds, no audio. Save to /public/images/guide/indicators/tema/chart.mp4.

What it measures

  • Trend direction with minimal lag. Tracks price more closely than any other classical moving average at the same period.
  • Momentum changes earlier than DEMA or EMA. The slope flips first.
  • Smoothed enough to be a reference line. Faster than raw price but still produces a clean curve.

Formula

How AlgoLift computes it

TEMA = (3 × EMA1) − (3 × EMA2) + EMA3

Where EMA1 is the EMA of price, EMA2 is the EMA of EMA1, and EMA3 is the EMA of EMA2 — all at the same period. The three-term correction cancels nearly all of the lag introduced by the underlying smoothing.

The math extends the same idea behind DEMA: subtract out the smoothing artifacts that come from nesting EMAs. Where DEMA cancels first-order lag, TEMA cancels second-order lag too — at the cost of amplifying high-frequency noise.

Inputs in AlgoLift

SettingDefaultRangeNotes
Period142–500Lookback in bars. TEMA's aggressive smoothing means very short periods (3–7) become viable where they'd be unusable on EMA.

Recommended periods

  • 5–9 (fast): Short-term execution. TEMA(5) reacts almost as fast as raw price changes but with a recognizable line.
  • 14 (default): Intraday day-trading. Balanced speed/smoothness.
  • 20–30: Slower reference. Most users prefer EMA at these periods because TEMA's extra reactivity becomes noise rather than signal.

Outputs in AlgoLift

HandleTypePlottedNotes
TEMANumericAlwaysThe TEMA line.
SlopeNumericOn selectChanges direction very quickly — useful for early momentum reversal detection, dangerous in chop.
Placeholder · Screenshot

TEMA node — default state

Author hint: The TEMA node on a fresh canvas with default Period=14. Both output handles labeled. Dark theme. Tight crop. Save to /public/images/guide/indicators/tema/node.png.

How to read it

  • Price above TEMA: Up-state, confirmed faster than any other moving average.
  • Price below TEMA: Down-state.
  • TEMA slope flip: Often the earliest possible "trend changed" signal you can extract from a moving average. Also the earliest possible false signal.
  • TEMA crossover with EMA or SMA at the same period: A common diagnostic — when TEMA leads EMA by a wide margin, the trend is genuinely fast.
Key Takeaway

TEMA reacts faster than anything else in the moving-average family. Use it when lag is the binding constraint on your strategy. Don't use it when you need stability — that's what SMA is for.

Best in / worst in

Best in very-short-timeframe execution (sub-5m bars) and on instruments with sharp directional moves. Pairs well with breakout strategies where catching the initial impulse matters more than filtering false starts.

Worst in sideways markets, on choppy instruments, and as a primary trend filter. The same reactivity that makes TEMA useful for execution makes it useless for regime classification.

Three setups

1. TEMA slope reversal entry

Skip the value, trade only the slope flip.

  • Long: TEMA(9) slope was negative for 3+ consecutive bars AND just turned positive.
  • Short: Symmetric.

This is essentially a momentum reversal signal. The bar-count requirement filters out the most common false signals.

Placeholder · Screenshot

TEMA slope reversal entry with Tally confirmation

Author hint: Node graph showing TEMA(9) Slope output feeding into a Tally node (count negative-slope bars), then into a comparison with the latest slope, gating an Entry. Save to /public/images/guide/indicators/tema/setup-01.png.

2. TEMA / EMA divergence

When the fast and slow versions disagree significantly, the market is in a strong directional move.

  • Long: TEMA(14) is more than 0.5% above EMA(14) — strong upside momentum.
  • Short: TEMA(14) is more than 0.5% below EMA(14) — strong downside momentum.

Use as a trend-strength filter rather than a primary signal.

3. TEMA fast execution + slow EMA filter

The classic "TEMA where speed matters, EMA where regime matters" pattern.

  • Long entry: Price crosses above TEMA(9) AND price > EMA(50).
  • Short entry: Price crosses below TEMA(9) AND price < EMA(50).

The slow EMA prevents most false TEMA signals; the fast TEMA gives entries earlier than an EMA-only system would.

Advanced patterns in AlgoLift

TEMA-triggered scale-in entries. Use a slow moving average (EMA or SMA) crossover as your initial entry, then use TEMA's faster slope flips to trigger Scaled Entry additions to winning positions. The pattern catches the slow setup and presses the position when fast confirmation arrives.

TEMA in a Cluster. TEMA + slow trend filter + ATR-based stop forms a complete short-term execution kit. Save the combination as a Cluster once and drop it into every short-horizon strategy you build.

Common mistakes

  • Using TEMA as a regime filter. It's far too reactive — a single sharp counter-bar flips the trend signal. Use SMA(200) or EMA(200) for regime, TEMA for execution.
  • Long periods on TEMA. TEMA(200) defeats the purpose. If you want a slow average, use one. TEMA earns its compute cost only at short periods.
  • No cooldown on TEMA crossovers. The faster the average, the more crossovers occur, the worse machine-gun trading becomes. Always wire a Cooldown.
Common Misconception
Myth
TEMA is the best moving average because it has the least lag.
Reality
Least lag isn't a universal good. Less lag means more reactivity, which means more false signals. The right moving average depends on what your strategy is doing — fast execution favors TEMA, regime classification favors SMA, and most things in between favor EMA.