Every cascade starts by moving against you. We tell you one is forming about ten minutes out, how far it runs, and how deep the fake-out goes before it turns.
No card · Free tier runs indefinitely · 1,649 signals public
Dashed lines mark the band four in five signals stay inside. A stop tighter than the lower line gets hit by the fake-out, not by being wrong.
of what? A hit rate quoted against nothing is a statement about Bitcoin, not about us.
We ran 41,225 random moments over the same 121 days and scored them identically. Against a random moment we are 11.4 points ahead, and the twenty minutes that follow carry 1.29× the volatility. Against the harder control — random moments matched on the previous half-hour of price movement — the gap narrows to 5.1 points and 1.10×. That second number is the one a sceptic would compute, so we print it ourselves.
Watch it call one live.
The free tier sends every alert. Nothing is held back except the numbers you'd size a stop with.
What it does, and what it refuses to claim.
What the ten minutes feel like.
The hard part is the first three minutes, when price is going the wrong way and you have to sit still.
The alert fires
Order-book depth thins, flow turns one-sided, liquidations start accelerating. You get the conditions, the expected size, and the fake-out band.
It goes against you
Around 0.16% in normal conditions, 0.28% when volatile. Late positions are being flushed. This is the part that shakes people out.
The cascade develops
Forced selling feeds itself and the real move builds. Four in five signals stay inside the band quoted at minute zero.
Peak, then graded
The move tops out and the signal is scored against what actually happened. Win or lose, it goes on the public record.
Running right now.
It works in storms, not in calm.
Conditions predict signal quality better than anything else we measure. Alerts now go out only in the two where we beat the control.
Start free. Pay when it saves you something.
Enough to see the pattern is real.
- Cascade alerts, long or short
- Market conditions and session
- Fake-out warning on every alert
- Public track record
The numbers that let you place a stop.
- Everything in Free
- Fake-out depth with the 80% band
- Expected move size per signal
- Outcome follow-ups when signals resolve
- Early-exit warnings on weak signals
The order book, and a direct line.
- Everything in Pro
- Order-book depth and spoof detail
- Entry timing guidance
- Private channel
- Developer and API access
Cancel any time · Every result public, failures included · No card for the free tier
Straight answers.
Is this a trading bot?
No. It does not trade, hold funds, or touch your exchange account. It watches Binance order flow and tells you when a cascade is forming. What you do next is yours.
Can you predict which way price will move?
No. Across 1,649 signals the favourable move beat the adverse one 50.4% of the time. A coin flip scores 49.8%. Reversing the direction on every alert we have ever sent would change our detection rate by about two points.
This is a volatility forecaster. It tells you a move is coming, roughly how large, and how much it is likely to hurt first. Anyone selling ten-minute directional calls on Bitcoin is selling a coin flip with a subscription attached.
Why does price go against me right after the alert?
Because that is what a cascade looks like from the inside. Late positions are force-closed before the move completes, so price dips against the call first — around 0.16% in normal conditions, 0.28% when volatile, for roughly three and a half minutes.
Four in five signals stay inside 0.25% in normal conditions and 0.44% when volatile. A stop tighter than that gets hit by the fake-out rather than by being wrong. Being stopped out and then watching the move happen is the most common reason people quit in week one, which is why the chart at the top of this page leads with it.
Can I just trade every signal?
No, and we would rather say so here than take your money for a month. Average favourable movement is 0.21% and average adverse movement 0.19%. After fees, trading every alert mechanically loses money. This is a risk and timing tool — protect a position, avoid a bad entry, size down before volatility hits.
Why do I get fewer alerts than I used to?
Because most of them were not worth sending. Against a volatility-matched control the engine has a real edge in extreme and high volatility — 7.0 and 8.9 points — and essentially none in normal or quiet markets. Alerts now go out only in the two conditions that earn it: roughly four a day instead of thirteen, at 78% detection instead of 68%.
Do you hide the bad calls?
Every signal is graded when its window closes and stays on the public record, including the ones that went nowhere. The live feed above shows everything, with nothing removed.
Do you have an API?
Yes — signed webhooks with HMAC, filtering by market condition and confidence, and a per-key delivery audit trail. It is sold separately from the consumer plans. Get in touch through the links below.
Sit through the first
three minutes.
Free tier, no card, running live right now. Watch a week of alerts before you decide anything.