foxalgoThe sweep funnel is a distance curve
Probabilità misurate. Nessuna indicazione operativa. cosa significa →
A funnel widely shared in the futures community starts from the price range of a specific hour of the day, watches for a sweep of one edge of that range, and reports what happens next once price closes back inside: a high rate of reaching the range's midpoint before the 16:00 ET session close. A commonly cited table reports figures like 96% for a 7am window, 87% for 8am, and 94% for the 9:30 window.
What the funnel does
The mechanics: take the high-low range of one hour (for example 07:00-07:59 ET), wait for price to sweep past the high or the low, require a close back inside the range (the "re-entry"), then track whether price reaches the range's midpoint by the 16:00 ET close. The published claim is that this sweep-and-reentry sequence is a strong precondition for reaching the midpoint.
Replication
We rebuilt the rule on NQ, matching the published sample size of roughly 270 days as closely as our data allowed. The low side matches the published figures closely; the high side differs, which traces to a difference in how "re-entry" is defined rather than to a different underlying rate.
| Side | Published | Replicated | Note |
|---|---|---|---|
| high | 97.2% | 91.1% | differs by re-entry definition |
| low | 97.6% | 97.5% | matches to a tenth |
Test against a touch curve
We compare the funnel's realized hit-mid rate against our own touch curve -- the same probability-of-touch surface behind this site's targets tool, fit across the full 2010-2026 history. The touch curve takes two inputs only: the remaining distance to the target, in ATR units, and the remaining time until the session close. It has no knowledge of a sweep having happened, and no knowledge of a re-entry.
On the two windows the funnel actually publishes numbers for (07:00 and 08:00 ET), the touch curve reproduces the realized rate closely: the delta between the realized rate and the curve's unconditional estimate is small in both directions across every market/side cell measured, roughly -0.8 to +0.6 percentage points. A curve with no knowledge of the sweep or the re-entry lands on the same number the funnel reports as its headline result.
| Window (ET) | Market | Side | n (days) | Actual | Curve | Delta (pp) |
|---|---|---|---|---|---|---|
| 07:00-07:59 | NQ | high | 3,256 | 95.5% | 96.0% | -0.44 |
| 07:00-07:59 | NQ | low | 3,229 | 95.9% | 96.1% | -0.25 |
| 07:00-07:59 | ES | high | 3,126 | 94.7% | 95.4% | -0.76 |
| 07:00-07:59 | ES | low | 3,198 | 95.8% | 96.1% | -0.34 |
| 08:00-08:59 | NQ | high | 3,007 | 92.8% | 92.8% | +0.07 |
| 08:00-08:59 | NQ | low | 2,993 | 94.2% | 93.6% | +0.58 |
| 08:00-08:59 | ES | high | 2,897 | 91.6% | 92.1% | -0.50 |
| 08:00-08:59 | ES | low | 2,920 | 93.4% | 93.0% | +0.38 |
| 09:30-09:59 | NQ | high | 2,442 | 74.9% | 80.9% | -5.97 |
| 09:30-10:29 | NQ | high | 2,099 | 55.9% | 68.9% | -13.02 |
| 09:30-10:29 | ES | high | 2,165 | 61.9% | 72.3% | -10.34 |
Later windows tell a different story
Extending the same comparison past the opening hours changes the picture. For the 09:30-10:29 ET window, the realized hit-mid rate runs well below the touch curve's unconditional estimate -- on the order of 10 to 13 percentage points lower, depending on market. After the opening range has already been established, conditioning on a sweep-and-reentry sequence performs worse than simply not conditioning on anything: the curve's plain distance-and-time estimate beats the funnel's own precondition.
The precondition adds little over the unconditioned base rate
Measuring the unconditioned hit-mid rate -- no sweep, no re-entry requirement, just "does price reach the midpoint by the close" -- gives a base rate of roughly 92% for both the 07:00 and 08:00 windows. Requiring the sweep-and-reentry sequence lifts that to roughly 94%, a gain of about two percentage points over the base rate, and roughly zero once compared against the touch curve, which already accounts for the specific distance and time involved. Most of what the funnel appears to add is already present in an unconditioned estimate of distance and time; the sweep-and-reentry condition contributes little beyond that.
Methodology
This measures touch odds, not profit odds: whether price reached the midpoint level at some point in the window says nothing about whether an order placed there would have filled at a usable price, what a trade's risk looked like by the time of the touch, or the cost of holding through the sweep itself. A high touch rate is not a claim that this sequence is profitable to trade.
Price any level yourself
The touch curve behind this comparison is not specific to one funnel, one hour, or one target -- it answers the same question for any distance, any time horizon, and any market this site tracks. Our targets tool lets you price the touch odds for any level you choose, the same measured approach as the table above. For how every number on this site is built and reported, see methodology.
Research and education, not financial advice. Independent, not affiliated with any third party.
FAQ
What is the sweep funnel?
A setup, widely shared in the futures community, built from the high-low range of a specific hour of the day: price sweeps past one edge of that range, closes back inside ("re-entry"), and the claim tracks how often price then reaches the range's midpoint before the session close.
Does the touch curve know about the sweep or the re-entry?
No. It takes only two inputs: the remaining distance to the target in ATR units, and the remaining time until the session close. It reproduces the funnel's published headline numbers without using either condition.
Does this mean the sweep-and-reentry sequence is meaningless?
On the specific windows published (7am, 8am), the sequence adds a small amount over an unconditioned base rate and roughly nothing over the touch curve. On later windows measured (9:30-10:29), it performs worse than not conditioning at all. Distance and remaining time explain most of what the funnel reports.
Is a high touch rate the same as a profitable trade?
No. This measures whether price crossed a level at some point in the window -- not whether an order placed there would have filled at a usable price, or how much risk had already accumulated by the time of the touch.