Apple and Amazon published their quarterly results on Thursday 30 July, just after the US stock market closed. We trade both as perpetual futures, and we record the state of the order book every second. This page replays what happened to the price and to how easily you could trade — one row per minute, for 1,532 minutes — from Thursday midday through Friday's opening bell. Coinbase is included as a comparison, because it did not report that day and behaves quite differently.
Everything here is built from three measurements. If you read only one section, read this one — the rest of the page assumes these three ideas.
The midpoint between the highest price anyone is currently willing to buy at and the lowest price anyone is willing to sell at. It is the best single answer to "what is this worth right now".
buy 302.60 / sell 302.70 → mid 302.65The gap between those two prices — roughly what it costs you to buy something and sell it straight back. We measure it in basis points, where 100 basis points is 1%. A small spread means trading is cheap. A wide spread means the market has turned expensive or nervous.
10 bps = 0.1% — a normal quiet minute for AppleHow much money's worth of orders is queued up close to the mid price. "Depth within 10 bps" counts every order priced within 0.1% of the mid. More depth means you can trade a bigger size without pushing the price against yourself.
$6M within 50 bps — a healthy book that absorbs sizeLooking at this same window one hour at a time smoothed all of the following away. At one-minute resolution the sequence is unambiguous — and the most interesting part happens before the news was public.
All four charts share one timeline, so the same horizontal position means the same moment in every chart. Shaded areas are the hours when the US stock market was open. The dashed red line is 20:00 UTC — the closing bell, and the moment the results landed. Hover anywhere to read the exact values for that minute.
The same four measurements, zoomed to 19:45–20:15 UTC. Here the collapse and the one-minute spread spike are individual minutes rather than a blip on a long line.
Coinbase's spread sits around 150–180 basis points all week. If the gap between the buy and sell price is 150 bps, the nearest order on each side is roughly 75 bps away from the mid — so when we ask how much is resting within 10 bps, or within 50 bps, the honest answer is nothing. The book is real and healthy at its own scale: widen the question to 100 bps and about $0.9–1.1M appears. That is why the depth charts plot Coinbase at 100 bps and label it so.
The same arithmetic catches Apple and Amazon at Friday's open, where spreads of 26–36 bps push every resting order outside the 10 bps band, so that column legitimately reads zero while the 50 bps column still shows $1.5–2M. A zero in the narrow band is a statement about the spread, not about whether anyone is quoting.
Price and spread come from a once-per-second recording, so every minute averages 60 observations. Depth is different: we photograph the whole order book only about 1.3 times per minute, so a minute's depth figure often rests on a single snapshot, and roughly 14% of minutes have no depth reading at all and are drawn as gaps. Trust the depth lines as a trend, not as individual points.
The raw order-book records are kept for one day and then deleted, so depth can only be rebuilt from 13:17 UTC on Thursday onward. Price and spread cover the full window. This affects the first 77 minutes of the charts and nothing around the earnings event.
The spread is measured between the best buy and sell orders whatever their size, so a single tiny order can make the market look tighter than you could really trade. Apple's quiet-period spreads of a fraction of a basis point are one cent on a $330 instrument — the smallest price step the exchange allows, rather than a measurement error.
Outside normal trading hours the exchange enforces bands that limit where orders are allowed to sit. We do not record those bands, so we cannot tell you whether a wide spread reflected genuine caution or an order simply not being permitted any closer to the mid. That matters most for exactly the overnight stretch this page covers.
Worth stating plainly, because it rules out the dull explanation: across all 1,532 minutes and all three instruments our recorder logged zero seconds of stale data, zero seconds of a crossed book and zero seconds of an empty book. What follows is the market, not our plumbing.
Depth is in thousands of dollars. A dash means no order-book snapshot landed in that minute. The closing-bell minute is highlighted.
| Minute (UTC) | Instrument | Mid | Spread bps | Depth ≤10bps | Depth ≤50bps | Depth ≤100bps | Snaps |
|---|