1 Minute
3 Minute
5 Minute
15 Minute
Box Funnel By Timeframe
Same relLeg filter, same far-end rule — does timeframe change how many boxes even make it to valid?
1 Minute
28.0%
valid rate · 1,623,972 boxes plotted
3 Minute
26.1%
valid rate · 672,094 boxes plotted
5 Minute
25.2%
valid rate · 428,110 boxes plotted
15 Minute
23.5%
valid rate · 164,616 boxes plotted
Valid Rate Alone Is Misleading
1m has the highest raw valid-rate (28.0%) and 15m the lowest
(23.5%) — but as the sections below show, that ranking flips completely once you look at outcomes. More boxes
form on faster timeframes simply because there's more noise to form legs from — quantity and quality move
in opposite directions here.
Win Rate By Stop & Target
Same base case as the main backtest page: stop = points beyond Level 2, target = points from entry
Expectancy Grid, Side By Side
Avg points per trade — win = +target, loss = −stop. The number that actually matters.
| Stop | Target | 1m | 3m | 5m | 15m |
|---|
| 5pt | 30pt | +4.65 | +6.24 | +7.05 | +9.04 |
| 5pt | 50pt | +4.91 | +6.93 | +8.06 | +11.08 |
| 5pt | 70pt | +4.79 | +7.17 | +8.34 | +12.14 |
| 5pt | 100pt | +4.49 | +7.43 | +9.03 | +13.33 |
| 10pt | 30pt | +4.39 | +5.90 | +6.70 | +8.60 |
| 10pt | 50pt | +4.63 | +6.64 | +7.79 | +10.72 |
| 10pt | 70pt | +4.33 | +6.82 | +8.11 | +11.78 |
| 10pt | 100pt | +3.82 | +7.02 | +8.89 | +13.22 |
15m Wins Every Single Cell
Across every stop/target combo tested, the ranking never changes: 15m > 5m > 3m > 1m. At the best
cell (5pt stop, 100pt target), 15m delivers +13.33 pts/trade — almost 3x 1m's +4.49, and still
well clear of 5m's +9.03. The pattern is simple: more time between candles means less noise contaminating
each swing, and that edge keeps compounding all the way out to 15m.
Leg Size Filter, By Timeframe
Stop=5, target=100 — does the leg-size edge hold up the same way on every timeframe?
| Leg Size | 1m | 3m | 5m | 15m |
|---|
| 10–20pt | +0.17 | +1.65 | +2.16 | +3.10 |
| 20–40pt | +2.72 | +3.83 | +4.18 | +4.74 |
| 40–80pt | +5.92 | +6.16 | +6.39 | +5.33 |
| 80–160pt | +10.47 | +11.53 | +11.80 | +12.16 |
| 160pt+ | +18.19 | +22.57 | +23.62 | +28.55 |
15m Pulls Ahead At Every Leg Size
At small leg sizes (10–20pt) 1m earns barely anything (+0.17) while 15m already clears a real edge (+3.10).
At the biggest legs (160pt+), 15m's advantage actually widens rather than narrows — +28.55 vs 1m's +18.19.
One exception: at 40–80pt legs, 15m briefly dips just under 5m (+5.33 vs +6.39) before pulling back ahead
at the next bucket — a small sample-size wobble, not a real reversal of the pattern.
Best Stacked Combo, By Timeframe
Way1 · leg ≥ 300pt · session hours 7AM–3PM NY · 5pt stop · 100pt target
1 Minute
30.0%
+26.45 pts/trade · n=2,217
3 Minute
35.2%
+31.91 pts/trade · n=3,172
5 Minute
37.4%
+34.29 pts/trade · n=3,164
15 Minute
41.3%
+38.40 pts/trade · n=3,334
Bottom Line
If you can only trade one timeframe, trade 15m. It wins every comparison on this page — funnel
quality (once you look past raw valid-rate), expectancy at every stop/target cell, leg-size response, and the
best stacked combo — while still firing a comparable trade count to 3m and 5m (3,334 vs 3,172 / 3,164
trades over 18 years in the best-combo test). 1m fires more often in raw terms but the extra volume doesn't
make up for the weaker per-trade edge; the pattern is consistent across the whole page — more time
between candles means less noise contaminating each swing, and that edge keeps paying out all the way to 15m.