Are crypto order books fake liquidity?
Updated
Are crypto order books fake liquidity?
A lot of resting size is pulled before it trades, so a single depth snapshot overstates what you can actually hit. But that is not a reason to ignore the book; it is a reason to measure its behaviour. How much liquidity is placed and then removed, how long a level survives, and how often levels flicker are measurable properties, and they are exactly what BlackForge records.
Concede the premise
The objection is right about the thing it points at. Much of the size resting in a crypto order book is never traded; it is placed and cancelled. Even the founder of a major historical-data vendor has said the common view is that the order book is mostly noise. If you treat one depth snapshot as liquidity you can take, you will be wrong, often.
Now measure it
The mistake is treating the book as a single snapshot. Spoofing and flicker are behaviours over time, and behaviours over time are measurable. BlackForge closes a five-minute window and records how the book actually behaved during it, not just how it looked at one instant. Three families of columns capture the exact thing the objection describes.
Liquidity added versus removed
The gross size placed into each side of the book during the window, and the gross size that left it. When added and removed are both large and roughly equal while little trades, most of the posted depth was placed only to be pulled. For BTC-USDT on Binance in the latest window, the bid side added 305,669,784 and removed 283,946,974, and the ask side added 367,928,874 and removed 346,011,496. Read them per metric on bid liquidity added and bid liquidity removed.
Price-level lifetime
The median time a price level stands in the book before it clears. Short lifetimes mean size is being posted and pulled quickly rather than resting. BTC-USDT on Binance showed a median level lifetime of 1,800 milliseconds in the latest window. See median level lifetime.
Level flicker
How many price levels were rebuilt after clearing, a direct count of the place-cancel-replace churn that spoofing produces. BTC-USDT on Binance flickered 7,970 levels against 100,608 total level changes in the window. See level flicker count.
Where the skeptics are right
- A single depth snapshot, taken alone, does overstate available liquidity.
- You cannot prove a specific order was a spoof from public data; only the aggregate footprint is measurable.
- Top-of-book quotes can be posted and pulled in milliseconds, faster than any five-minute window resolves.
The point is not that the book is honest; it is that its dishonesty is measurable. Read added-and-removed, level lifetime and flicker together, and the pulled liquidity the objection worries about becomes a number you can see rather than a reason to look away.
Aren’t crypto order books fake liquidity?
A large share of resting order-book size is pulled before it ever trades, so treating a single depth snapshot as liquidity you can hit is a mistake. But that behaviour is not noise, it is a measurable property of the book: how much size is placed and then removed within a window, how long a price level survives before it clears, and how often levels are rebuilt after clearing. BlackForge measures exactly that.
Can spoofing be measured?
Yes, in aggregate. You cannot label a single order as spoofed from public data, but you can measure the footprint spoofing leaves: liquidity added versus liquidity removed, the median lifetime of a resting level, and level flicker. A book full of size that is constantly placed and pulled shows high added-and-removed volume and short level lifetimes, and those are columns you can read.
What is the difference between resting and pulled liquidity?
Resting liquidity is size sitting in the book. Pulled liquidity is size that was placed and then cancelled rather than traded. Over a five-minute window BlackForge records the gross liquidity added and the gross liquidity removed on each side, so you can see how much of the posted book was real depth and how much was placed only to be withdrawn.
Sources
- Tardis.dev founder, on the common view that the order book is mostly noise (Hacker News): news.ycombinator.com/item?id=20663766
Liquidity added and removed, defined · Resting order lifetime, defined · How BlackForge measures the market · Pricing