Exponential momentum
Percent gap between a fast and a slow average plus the last bar's move; above zero leans up.
The indicator keeps two exponential averages of the close: a slow one over period bars (20 by default) and a fast one over period divided by multiplier (10 by default). The score is the fast average's percent distance from the slow one plus the latest bar's percent change. The description promises accelerating momentum, but the code measures no acceleration and predicts no move: it is a short-versus-long trend gap with the last bar's return added at equal weight.
How to read it
- The line sits in its own pane below the price, in percent, with zero as the balance point.
- Above zero the short-term average is over the longer one and/or the latest bar rose; below zero is the reverse.
- Readings beyond about ±0.04 occurred on only 5% of September 2026 SPY one-minute bars on each side. Treat those as notable and confirm the direction on price.
What it outputs
- Momentum score
- Percent distance of a fast average from a slow one, plus the last bar's percent change. Above zero leans up, below zero leans down. Sample range -0.76 to +0.49.
Settings
- period
- Length of the slow average in bars; the fast one is this divided by multiplier. Shorter periods make the reading react bar by bar and look noisier. Longer periods smooth it and let the gap build over slower trends.
- multiplier
- Fast length is period divided by this value. Above 1 the fast average reacts faster and the gap widens. At exactly 1 both averages match and only the last bar's move is left. Below 1 it lags. Negatives diverge; zero errors out.
Pitfalls
- The last bar's percent move enters at full weight beside the average gap, so one noisy bar shifts the reading directly.
- Values are small: 90% of September 2026 SPY one-minute readings sat between -0.04 and +0.04. Judge a reading against its own history, not a fixed level.
- Negative multipliers make the output diverge: at -2 it reached about 1e32 within 390 bars. A multiplier of 0, or minus the period, raises a divide-by-zero error.