What Is Order Book Imbalance (OBI)?
Order book imbalance is the normalised difference between resting buy and sell volume near the top of the book: (bid volume − ask volume) / (bid volume + ask volume), on a −1 to +1 scale. Positive means more resting bids than asks. It is a short-horizon pressure gauge, not a directional forecast.
Chart indicators are computed from prices that have already happened. Order book imbalance is computed from orders that have not yet executed — it is one of the few retail-accessible signals that describes intent rather than history. That is its appeal, and its limits follow directly from the same property.
- OBI is bounded −1 to +1, which makes it comparable across instruments of any price or size.
- It measures resting limit orders — passive intent, which can be cancelled at any moment.
- Its predictive horizon is seconds to minutes, not hours. Treating it as a trend signal is a category error.
- Depth choice changes the number completely: 5 levels and 20 levels can disagree in sign.
- Spoofing exists, which is why aggregate flow confirmation matters more than the raw reading.
How is order book imbalance calculated?
An order book is the list of resting limit orders — bids below the current price, asks above it. Order book imbalance compares the total volume on each side within a chosen depth:
OBI = (bid_volume − ask_volume) / (bid_volume + ask_volume)
The division by total volume is what makes the measure useful. It normalises to a fixed −1 to +1 range, so a reading on BTC and a reading on a small-cap alt are directly comparable despite wildly different absolute volumes.
| Reading | Meaning |
|---|---|
| +1.0 | All resting volume is on the bid side; no asks in range |
| +0.5 | Bids are three times ask volume — strong buy-side pressure |
| 0.0 | Balanced book |
| −0.5 | Asks are three times bid volume — strong sell-side pressure |
| −1.0 | All resting volume is on the ask side |
Worked example: 150 BTC of bids and 50 BTC of asks within your depth window gives (150 − 50) / 200 = +0.5.
Depth window: the choice that changes the answer
"Within your depth window" is doing heavy lifting there. How many price levels you aggregate is not a detail — it changes the reading, sometimes including its sign.
- Top 5 levels — very responsive, dominated by fast-cancelling orders, noisy.
- Top 10–20 levels — the usual compromise; captures meaningful resting interest without excessive churn.
- Percentage band — everything within, say, 0.5% of mid. More robust across instruments with different tick sizes, and generally the better choice if you are comparing across a universe.
A book can be bid-heavy in the top 5 and ask-heavy across the top 20. Both readings are correct; they answer different questions. Fix your depth definition before you compare anything, and never compare readings taken at different depths.
Why does order book imbalance matter?
Price moves when aggressive orders consume resting liquidity. If resting bid volume greatly exceeds resting ask volume, then all else equal it takes less aggressive buying to move price up than aggressive selling to move it down. The path of least resistance is upward.
That is the entire mechanism, and it is worth stating plainly because it also delimits what OBI can do. It describes the current shape of resting liquidity. It does not know what is about to arrive.
Passive intent, and its consequences
Resting limit orders are passive intent: someone is willing to trade at that price if the market comes to them. Three consequences follow, and they are the whole story of why naive OBI trading fails:
- Cancellation is free. Most resting orders are cancelled rather than filled. A wall can disappear in milliseconds without a single trade.
- Displayed is not total. Iceberg orders show a fraction of their size, so the visible book systematically understates real interest at some levels.
- Spoofing exists. Large orders placed with no intent to fill, specifically to create a reading like the one you are measuring. It is prohibited on regulated venues and demonstrably present in crypto.
This is why practitioners treat OBI as one input rather than a signal: it is a genuine measurement of a genuinely informative quantity, and that quantity can be manufactured.
What is a meaningful OBI reading?
There is no universal threshold, and anyone quoting one without naming the instrument, the depth window and the venue is selling something.
Rough working bands for a liquid major, at a 10–20 level depth:
| Absolute reading | Interpretation |
|---|---|
| 0.0 – 0.15 | Noise. Books are rarely exactly balanced. |
| 0.15 – 0.30 | Mild lean. Weak on its own. |
| 0.30 – 0.50 | Meaningful imbalance. |
| 0.50+ | Strong — and worth asking whether it is real. |
Calibrate against the instrument's own distribution rather than importing numbers. A reading of +0.4 is unremarkable on a thin book that swings wildly and significant on a deep one that normally sits near zero.
What OBI is not
Not a trend indicator. Its informational half-life is measured in seconds to minutes. A reading from an hour ago tells you nothing about now.
Not a standalone entry. Buying every time OBI exceeds +0.4 produces a great many trades and no edge after costs — particularly given that OBI strategies tend to use tight stops, and round-trip cost in R scales inversely with stop distance.
Not comparable across venues. The same asset has different books on different exchanges. An imbalance on one venue may be absent on another, and cross-venue disagreement is itself information.
Not immune to manipulation. See spoofing above.
OBI divergence
A more robust use than the raw level: compare the direction of price with the direction of imbalance. Price grinding to new highs while OBI deteriorates means the advance is being made against thinning bid support — the move is being carried by aggression rather than by underlying demand. That disagreement is often more informative than either series alone, and it is harder to manufacture than a single large resting order.
Where to go next
This page covers what the measure is. Turning it into something tradeable — depth selection, threshold calibration, persistence requirements, confirmation from executed flow, and the cost problem that kills most OBI strategies — is covered in building a strategy on order book imbalance.
TradeFloor computes OBI across Hyperliquid, Binance and Bybit and exposes it as a trigger condition, so a rule can require an imbalance threshold together with up to four other conditions rather than acting on the reading alone.
Frequently asked questions
What does order book imbalance measure?
The relative weight of resting buy versus sell limit orders near the current price, normalised to a −1 to +1 scale. Positive means more resting bid volume than ask volume within your chosen depth. It measures passive intent — orders waiting to be filled — not orders that have executed.
How do you calculate order book imbalance?
Sum bid volume and ask volume within a chosen depth, then compute (bid_volume − ask_volume) / (bid_volume + ask_volume). With 150 BTC of bids and 50 BTC of asks the result is +0.5. The division normalises the measure so readings are comparable across instruments of any size.
What is a good OBI threshold for trading?
There is no universal number — it depends on the instrument, the depth window and the venue. On a liquid major at 10–20 levels of depth, readings below 0.15 are usually noise, 0.30–0.50 is meaningful, and above 0.50 is strong enough to be worth asking whether it is genuine. Calibrate against the instrument's own distribution.
Can order book imbalance be manipulated?
Yes. Spoofing — placing large orders with no intention of filling them — creates exactly this reading, and unlike executed volume it costs nothing to cancel. Iceberg orders create the opposite distortion by hiding real size. This is why executed-flow confirmation matters more than the raw book reading.
How long does an OBI signal remain valid?
Seconds to minutes on liquid instruments. Resting orders are cancelled and replaced constantly, so the informational content decays quickly. Treating OBI as a multi-hour trend signal is a category error — it describes the current shape of the book, not a persistent condition.
Is order book imbalance the same as volume imbalance?
No, and the distinction matters. Order book imbalance measures resting limit orders that have not executed — passive intent that can be cancelled. Volume or delta imbalance measures orders that actually traded — realised aggression. They frequently disagree, and that disagreement is informative.
Order Book Imbalance: Building a Strategy on Order Flow
Depth selection, threshold calibration, persistence, flow confirmation, and the cost problem that kills most OBI strategies.
// questions or corrections · [email protected] · more essays · /blog