OptiNod Academy

T3 — Six-Stage Smoothed Moving Average

A smooth curve with less lag than an EMA, used to read trend curvature.

T3 is a moving average that applies multiple smoothing steps while reducing lag. It is less noisy than an EMA and does not lag as much as an SMA, which makes it useful when you want to read changes in the slope of the average line.

If you use it as an entry trigger, you lose that advantage. What matters most is whether the T3 is steepening, flattening, or starting to roll over. That curvature change is what acts as a signal that momentum is shifting.

**T3 applies six smoothing steps while correcting for lag**GD correction partially adds back the gap between the source and the smoothed line at each smoothing stage, creating a T3 curve that stays smooth without drifting too far away from price.

The problem is that T3 loses its edge if you use it the same way you used an EMA. If you buy whenever price is above the T3, or enter whenever two T3 lines cross, you inherit the same traps as EMA systems. False signals pile up in ranges, and when a trend starts, the response is still late. T3 is designed to read how quickly slope is changing, so it becomes less effective when treated as a pure entry signal.

Curvature is momentum

Inside a trend, it is common for the T3 to slope upward. The important information is whether that slope is getting steeper, flattening, or beginning to turn. Even within the same uptrend, a steepening T3 marks a strengthening phase, while a flattening T3 marks fading momentum.

If you try to read this with an EMA, the line often appears to flatten and steepen repeatedly. With an SMA, you notice the flattening too late. T3, by contrast, has enough smoothing to absorb noise and leave only the meaningful curvature shifts. That is why the moment the T3 starts to flatten can function as a solid signal that an EMA often fails to capture.

In the two weeks before QQQ made a high near $500 in July 2024, the slope of the daily T3(14, 0.7) first steepened and then flattened. At the same time, the slope change in the EMA(14) was subtle enough to get buried in noise. T3 showed the curvature change more clearly first, and QQQ then dropped to around $460 in the first week of August.

In one uptrend, a steepening T3 means strengthening momentum, a flattening T3 means it is fading

The paradox of six-stage smoothing: fast, but slow at turns

Saying T3 is generally faster than an EMA is only true on average while a trend is already in progress. At inflection points, the opposite happens.

When smoothing is repeated six times, the line needs a large accumulated change before it turns. The correction term reduces lag, but it still takes time for a change to pass through all six smoothing stages and show up across the whole line. Even if price suddenly starts moving the other way, the T3 slope changes slowly.

When you measure this paradox, a clean asymmetry appears. During an ongoing trend, T3 tracks price pullbacks more smoothly than an EMA, but at trend reversals it turns clearly later than an EMA. That makes T3 highly accurate for assessing momentum state, but one of the slowest tools for detecting a directional turn.

If you ignore this asymmetry and use T3 as an entry and exit tool, you will keep entering late and exiting late. T3 belongs in between. Its job is to monitor how momentum is changing while the trend remains alive.

**T3 is fast during a trend and slow at inflection points**When price keeps moving in the same direction, T3 tracks it more smoothly than an EMA. When direction changes, the smoothing structure makes T3 turn late.

Trend phases fall into four stages

By reading T3 slope and curvature together, you can divide a trend into four stages.

Start stage: The T3 slope turns decisively from near zero in one direction. Because T3 identifies this stage late, use another tool, such as a break in price structure, to catch the start. Use T3 only for confirmation.
Acceleration stage: The T3 slope is getting steeper. The trend is strengthening, so pullback buys tend to work best, and pullbacks into the T3 become solid buy zones.
Maturity stage: The T3 slope stays steady. The trend is still alive, but new entries should be treated more conservatively, while existing positions are usually worth holding.
Exhaustion stage: T3 curvature begins to flatten. Increase profit-taking, stop adding new entries, and assume the trend itself may soon end.

> ETH is in an uptrend on the daily chart, trading above the 200 EMA with ADX at 23 or higher.

> T3(14, 0.7) is in the acceleration stage, with its slope steepening relative to the average of the previous 10 bars.

> Price pulls back into the T3, then a candle closes back above the T3.

> Enter long at that candle's close, and place the stop 0.5 ATR below the T3 line.

> If T3 curvature shifts into the flattening stage, or price closes below the T3, the setup is considered invalid.

When ETH moved from the $2,500 area to the $3,200 area in November 2024, nearly every pullback buy taken during the T3 acceleration stage worked. In early December, after the same setup shifted into the exhaustion stage, it stopped out twice in a row.

**T3 slope divides a trend into four stages**Classifying start, acceleration, maturity, and exhaustion by changes in the curve's slope separates where to add new entries and where to scale them back.

Volume factor: what the 0-to-1 range means

T3's second parameter, v, also called vfactor or b, is a value between 0 and 1. Intuitively, it controls the strength of the correction.

When v is set to 0, the correction disappears and T3 becomes an EMA smoothed six times. It is very smooth, but it responds very late. When v is set to 1, the correction is at maximum strength, so T3 follows price with almost no lag, but that speed creates overshoot. Even after price stops, T3 may briefly continue higher before turning back.

Tillson used 0.7 as the default because testing showed it produced the most natural balance between smoothness and responsiveness. He chose it empirically, from experiment.

There is one adjustment rule: find the v value that makes curvature changes visible without creating false signals. If v is too low, curvature changes appear too late. If v is too high, overshoot makes the curvature wobble too often, making it impossible to separate the stages. Backtest the ratio of real trend endings to false flattening signals for each asset and timeframe to find a suitable value. BTC daily charts usually work well around 0.6-0.7, while single stocks such as NVDA more often respond better to 0.7-0.8.

**Volume factor balances smoothness and overshoot**A lower v is slower. A higher v is faster, but it can jump at inflection points. 0.7 is a reasonable starting balance.

The warm-up trap: 6 x N is only the minimum

Because T3 applies six smoothing stages, it takes time for the values to stabilize. Roughly 6 times N, about 84 bars for T3(14) and about 120 bars for T3(20), should be treated only as the minimum. The line needs much more data than that minimum to pass through all six smoothing stages and fully settle. Values right after 6 x N may still contain early instability.

This is often ignored in two places. First, if you use T3(20) on a short backtest window, such as 100 bars, nearly the entire dataset is warm-up, so the signals themselves are not meaningful. Second, using early data from a newly listed asset to judge a T3 trend fails for the same reason. Use 6 x N as the minimum standard, but in practice it is safer to start with at least about one year of data for the asset, roughly 365 bars, and only accept signals from bars where the T3 line has had enough time to stabilize.

Three places where T3 signals break down

Simple crossover entries: Entering on golden crosses or dead crosses between two T3 lines creates the same trap as EMA crossovers. Whipsaws build up in ranges, and even in trending markets, by the time the crossover appears, part of the trend has already played out.
Relying on inflection-point exits: If you wait until T3 turns, your exit will be even later than with an EMA. At inflection points, price structure, such as a break below the prior swing low, or an RSI close below 50 gives a much faster signal.
Using the default v unchanged: 0.7 is just an average value, and the right value differs by asset. If the stages are not separating cleanly on your asset, adjust v in 0.1 increments.