Exponential trend strength
How one-sided recent moves are, scaled by acceleration; it shows strength, not direction.
It averages the true range and the up-move and down-move sizes over the last window bars, then scores how lopsided the up-moves are against the down-moves, from 0 to 100. That score is multiplied by one plus the average of each bar's change in up-move size minus its change in down-move size, divided by the average range. The result is capped at 100 but not floored at zero, and it never says which way price is moving.
How to read it
- The line sits in its own pane below the candles and is capped at 100 at the top.
- At defaults the median is about 36, so over half of bars read above 30; the top 5% sit above 92.
- The score has no direction. Read the price direction from the candles, then use the line to judge how one-sided that move is.
- On an up-move, a drop from a peak while price keeps climbing means the up-moves are slowing.
What it outputs
- Trend strength
- How lopsided recent up-moves are against down-moves, from 0 to 100, scaled by acceleration. Can dip below zero and carries no direction.
Settings
- window
- Bars averaged for the true range, the up-moves, down-moves and their acceleration. Shorter, such as 5, swings faster and more often; longer, such as 30, smooths the line and lags. The first reading appears after window plus one bars.
Pitfalls
- The boost follows up-move growth minus down-move growth, not the trend's own acceleration. A slowing sell-off raises the score and an accelerating one lowers it.
- The name and docstring overstate it: the factor is linear, not exponential; it is not the standard ADX (no second smoothing); and the 30 and 60 cut-offs appear only in the source docstring.
- Values below zero are rare, under 5% of bars, but the deepest reached about -733 in a month of SPY one-minute bars, so it is not a bounded 0 to 100 gauge.