- Range break +
bookSignal - Depth +
spreadConfirmation - ~20 HzResolution
- Crypto +
moreMarkets
Price breaking a range is the easy part to spot. The hard part is knowing whether to believe it. The order book gives you the confirmation: a breakout that coincides with depth pulling and the spread widening is a different animal from one that fizzles back into the range.
On any market, most range breaks fail. The ones worth acting on tend to leave a signature in the book, liquidity moving out of the way and quotes repricing. With high-frequency snapshots you can see that signature form moment by moment, not just guess at it after the fact.
A two-part signal
The range break
Define the consolidation band from recent mid-price action, then flag when price closes decisively outside it.
- Recent high/low band
- Decisive close beyond
- Direction tagged
The book confirmation
A real break usually comes with depth thinning ahead of price and the spread widening as makers reprice.
- Depth pulls aside
- Spread expands
- Imbalance leans the break’s way
The fields that carry the confirmation
You do not need anything exotic. The range comes from mid-price history; the confirmation comes from the spread and the two depth totals on the same snapshots. Because they share a timestamp, the price break and the book reaction are measured at the same instant, no stitching two feeds together.
mid_priceDefines the range and the break- spreadWidens as makers reprice the move
bid_depth_totalask_depth_totalImbalance leans the break’s way- ~20 Hz (BTC)Fast enough to time the confirmation
Why confirmation cuts false signals
A price poke past the range with the book unchanged, full depth still resting, spread still tight, is exactly the kind of move that snaps back. Requiring the book to confirm filters those out. It will not catch every winner, but it raises the hit rate by ignoring breaks the liquidity does not support.
- 1Define the consolidation band from recent mid-price, a rolling high and low over a quiet window.
- 2Flag a candidate break when mid closes decisively beyond the band, and tag its direction.
- 3Check the book at the break: depth thinning ahead of price and the spread widening, both leaning the break’s way.
- 4Only accept the signal when price and book agree, and log the ones that broke without book support to study reversals.
The book is the second opinion
Treat the range break as the hypothesis and the order book as the test. Depth getting out of the way and the spread widening in the break’s direction is the market agreeing with you. No book reaction is the market disagreeing, and on a thin prediction market, the book is often right.
The honest limits
It’s a filter, not an oracle
Confirmation trims false breaks; it also trims some real ones that move before the book reacts. You are trading recall for precision.
- Fewer false positives
- Some real breaks missed
- A deliberate trade-off
Read of intent, not fills
Depth thinning is resting size being pulled, a read on maker intent, not a ledger of executed trades.
- Resting size, not volume
- Can be cancelled
- A proxy, honestly
Sampling rate sets the edge
On fast crypto books the ~20 Hz capture catches the confirmation in time; on slower categories the read is coarser.
- Crypto near 20 Hz
- Sports slower
- Tune the window per market
A breakout is a hypothesis. The order book is the cheapest test you can run on it, and on a thin market, the test is usually right.
Backtest the idea
Test a book-confirmed breakout rule in the Strategy Builder, or pull the depth snapshots to prototype it yourself.
Frequently asked questions
How does the order book confirm a breakout?
A genuine break usually coincides with resting depth thinning ahead of price and the bid-ask spread widening as makers reprice. If price pokes past a range but the book is unchanged, depth intact, spread tight, the move is more likely to reverse.
What data do I need for book-confirmed breakouts?
Mid-price history to define the range and detect the break, plus the depth and spread fields to confirm it. All of these come from the same historical snapshots, captured at roughly 20 Hz on crypto markets so the confirmation is timely.
Does confirmation guarantee the breakout works?
No, nothing does. Requiring book confirmation filters out breaks the liquidity does not support, which tends to raise the hit rate, but you should still backtest the rule and model fills honestly before trusting it.
Does requiring book confirmation mean I miss some real breakouts?
Yes, that is the trade-off. Demanding the book agree trims false breaks but also drops some genuine ones that run before makers react. You are trading recall for precision. Whether that trade is worth it depends on your strategy, so log the breaks that fired without book support and backtest both versions before committing to the filter.