Somebody Placed the Order
Somebody Placed the Order
- QUESTION
- What does one large order do to a book, and what does it cost the person who sends it?
- METHOD
- We did not place an order. The largest print in our frozen second was placed by somebody else, and we reconstructed it from our own capture: the book before, the print, the book after.
- DATE
- 16 Sep 2026 · 07:24:03.530 to 07:24:04.537 UTC · Binance BTC/USDT
- SAMPLE
- One aggregated print, two depth-20 snapshots, 207 Binance prints in the window.
- LIMITS
- We cannot see who sent it, why, or what fee tier they pay. Binance's aggregated trade stream combines fills of one taker order at one price, so one print may be several fills.
- RESULT
- It paid the screen price. The book around it did not survive the second.
Binance, 470 ms before the second. Best offer $75,807.20 for 6.83704 BTC, worth $518,297. Twenty visible levels on the sell side, $585,236 in all. First eight:
| Price | BTC offered | Worth |
|---|---|---|
| 75,807.20 | 6.83704 | $518,297 |
| 75,807.21 | 0.00098 | $74 |
| 75,807.22 | 0.00014 | $11 |
| 75,808.10 | 0.00200 | $152 |
| 75,808.41 | 0.00014 | $11 |
| 75,808.42 | 0.01764 | $1,337 |
| 75,808.61 | 0.00007 | $5 |
| 75,808.80 | 0.00741 | $562 |
An order to buy. We infer the side: it crossed at the best offer, not the best bid. We cannot see whether it was a market order or a limit order priced to take, or whether it was one slice of something larger.
At +67 ms into the second, one aggregated print: 6.51994 BTC at $75,807.20, worth $494,258. The largest single print on any of the five venues in the second.
It fitted inside the best offer shown in the previous snapshot, 6.83704 BTC at that price, so every unit paid the screen price. The level was not exhausted by this print alone: 0.31710 BTC would have remained.
Slippage against the price on the screen: 0.00 bp.
Fees: unknown. Binance publishes a spot taker fee schedule that starts at 0.10% and falls with volume and holdings. At the base rate this order would have paid about $494; a large firm would pay less. We do not know this firm's tier.
Market impact: see 05.
Between the snapshot before and the snapshot at +537 ms, the best offer rose $36.81, about 4.9 bp. The dollars visible on the sell side fell from $585,236 to $30,094, a drop of 95%.
In that window Binance printed 207 trades for 16.33 BTC, $1,237,816. So our order was about 40% of the bitcoin bought there in half a second, and the rest came after it, while Bybit, OKX and Gate were printing higher too.
| Book at | Best bid | Best ask | Visible sell side | 20th ask level |
|---|---|---|---|---|
| −470 ms | 75,807.19 | 75,807.20 | $585,236 | 75,811.26 |
| +537 ms | 75,844.00 | 75,844.01 | $30,094 | 75,848.89 |
| +1,597 ms | 75,840.33 | 75,840.34 | $638,838 | 75,846.05 |
How much of the thinning was this order, how much was the buying that followed it, and how much was market makers pulling their offers because they saw the buying. A book snapshot shows what is there, not why it left.
One second later the sell side had been rebuilt, deeper than before, about $33 higher.
Who the buyer was. Who the sellers were: several makers may have been resting at that price. Whether the order was the first slice of a larger one. What the buyer paid in fees. Whether anything was hidden behind the visible levels.
The screen price was honest for this order and useless as a description of the market a few hundred milliseconds later. A buyer of the same size arriving at +537 ms would have found less than a tenth of the dollars visible, and could not have filled from the visible book at all.
An order we sent would have been small enough to learn nothing, or large enough to become a story about us. The market had already run the experiment, with a real buyer and real money, inside the second we froze before writing a word. We report it exactly as captured.