If price takes this out: history says continuation, not reversal — after 32 sweeps of significant highs, price was higher 3 days later 69% of the time (avg +2.6%). Do: nothing — don't short it, don't chase it. The trend model is already positioned; only a Sunday close moves us.
If price takes this out: no reliable bounce — swept lows were a coin flip (win 50%, avg -0.0%/3d over 32 sweeps). Do: nothing intra-week — not a buy-the-dip signal, not a panic-sell signal. Our exits fire only on the Sunday-close ladder on the signal page.
How to use this page: glance at the two levels above; the flags light up live when a sweep is in progress. Both outcomes are pre-briefed — so whatever happens mid-week, the answer is already written and it's never an intra-week trade.
Where the stops are resting
Unswept highs above (buy-stops pool)
| $126,200 D+W | +63.2% |
| $116,400 D | +50.5% |
| $97,924 D+W | +26.6% |
| $82,850 D+W | +7.1% |
live · $77,324
Unswept lows below (sell-stops pool)
| $62,535 D | -19.1% |
| $62,275 D | -19.5% |
| $57,800 D+W | -25.2% |
| $52,550 D+W | -32.0% |
| $49,000 D+W | -36.6% |
These levels are real: stops do cluster beyond significant highs and lows, and price does sweep them. What the data refuses to support is the profitable part of the story — see the evidence below before trading any of this.
Recent daily-level sweeps · what happened next (7d)
| 2026-06-25 | low swept | $59,794 | +3.0% |
| 2026-06-24 | low swept | $61,078 | -1.7% |
| 2026-06-05 | low swept | $61,056 | +4.1% |
| 2026-06-04 | low swept | $63,886 | -0.4% |
| 2026-04-14 | high swept | $74,132 | +3.0% |
| 2026-02-03 | low swept | $75,770 | -9.1% |
Fading a swept high lost money. Shorting after a significant daily high was swept-and-rejected returned -2.6% over 3 days (win rate 31%, t=-2.7, n=32). Sweeps of major highs were continuation, not reversal — the market that just paid up through everyone's stops usually keeps going.
Buying swept lows: -0.0% over 3 days (win 50%, t=-0.0, n=32) — indistinguishable from noise, below fees. 4-hour and weekly variants: also noise. Intraday (5m/1h) variants: tested separately, no edge.
What this concept claims, and why charts make it look real
The claim. Stops cluster beyond significant highs/lows. When price sweeps them, one-sided emotion peaks and price sharply reverses — so fade the sweep with a stop beyond the wick.
Why it looks real. Every major reversal, by definition, happens at a price extreme — which in hindsight is always "a swept level." Chart threads show you the sweeps that preceded reversals and never count the ones that didn't. That's survivorship bias applied to patterns, and it makes any level-based story look prophetic.
What we measured instead. Counting every sweep since 2017: highs that got swept kept rising (fading them cost -2.6%/3d), and swept lows bounced no better than any random day. This coheres with everything else we've measured on BTC: strength continues. It's why our actual system (the trend signal) buys breakouts rather than fading them.
If you still want to watch it, that's what this page is for: the levels update with each weekly publish, and the sweep log shows what actually followed — keep score honestly.