One number that tells you how the Bitcoin order book is stacked — on your lock screen, at a glance, all day.
“It’s a widget that lives on top of the lock screen. So every time I tap my phone, I see it… It’s like a clock on the wall.”
— the designer of BitClock
Everything on the BitClock screen exists to serve a single reading: the ratio.
A ratio of 1 means the two sides are equal. Above 1, the sell side is heavier. Below 1, the buy side is heavier. That’s the whole formula — no smoothing, no indicators, just the live book, recalculated every second.
Three real readings from the record, so you can see the range the dial actually covers:
313 BTC bid, 383 BTC asked. An ordinary balanced afternoon — this is the frame on the screenshot opposite. Most of the time, the clock looks like this.
330 BTC bid, 1.9 BTC asked. November 21, 2025 — the record flash: for nineteen seconds the sell side had almost completely vanished. (Deepest sustained floor: 0.038, November 14.)
64 BTC bid, 711 BTC asked. October 28, 2025 — the highest sustained reading on record. Eleven times more supply than demand sitting on the book.
The width is inherited from the depth-of-market chart the designer spent months watching by hand before BitClock existed — it is the exact bandwidth of the view the method was learned on.
Seven things, top to bottom. This is a real capture of the app.
R-HOD — the ratio high of the day (green), with the BTC price in brackets at the moment it printed.
The ratio, live. The big number. Recalculated every second from the resting order book.
R-LOD — the ratio low of the day (red), also stamped with the price at the moment it printed.
Bitcoin’s current price.
The raw depths. BTC currently bid (green) and asked (red) inside the band. The same ratio means different things at 900/100 than at 90/10 — so both sides stay on screen.
24H VOL — traded volume.
The 24H price High and Low — updated the instant a new extreme prints.
The extreme readings flash — they last seconds and don’t hold. In four months of watching the book by hand, the designer never once caught them live. The day’s high and low marks are how the clock remembers them for you — and since August 22, 2026 they are each second’s true high and low, so they can run beyond anything the big number happened to show — in the first hour measured, they did in 92% of seconds.
Nearly eleven months of second-by-second data — 26.0 million readings, October 12, 2025 through August 31, 2026 — show how the dial really spends its time.
The floor. The sell side has emptied out. Historically these prints clustered on capitulation days — big candles, stop runs, panic. Rare, brief, and always worth noticing.
The light side. Bids outweigh asks. Common; on its own it says little — context decides.
The quiet middle. Where the clock spends most of its life. Nothing to do here.
The heavy side. Supply stacking overhead. Historically kept company with weak stretches of tape.
The ceiling. 67 episodes in nearly eleven months, counting a new episode after any gap of more than 30 minutes. When this zone lights up, something unusual is happening.
Asked what he does when the ratio crosses a level, the designer rejected the question: “It doesn’t really work like that. It’s not like there’s a specific action when the ratio crosses above two.” The zones tell you where the needle is on its dial — what to do about it depends on everything else you know.
Each card is rebuilt field-for-field from the second it happened, straight from the archive. This is what the clock showed.
The heaviest sustained book ever logged. 711 BTC of supply against 64 of demand — for 27 minutes. Over the next 25 days the market fell about 27% into its November low.
The emptiest sell side ever logged — 1.9 BTC asked against 330 bid, in the middle of a panic night. The market printed its major swing low the next day and stood about 7% higher a week later.
A one-second flash to 7.2 during a rally — there and gone. The rally printed its top the very next day, and the market went on to lose roughly a third of its value over the following three weeks.
Every field above is the stored value from that second’s archive row. The 24H VOL / High / Low bar shows the app’s 24-hour statistics, which at the time could lag fast moves — on Jan 13 the “high” still trailed the live price. (Today the High and Low update the instant a new extreme prints.) Past behaviour of the order book is not a prediction of future prices.
A toggle in the app’s menu switches between two drawings of the same number — pick whichever your eye prefers.
The ratio exactly as the formula produces it. 1 is the balance point — the line the whole system was designed around, and the face its designer reads: “that’s how it was developed, that’s how it was learned.”
Its one quirk, drawn below: the heavy-sell side stretches from 1 up toward 15, while the entire heavy-buy world is squeezed between 1 and 0.
Built to fix exactly that quirk. At 1 and above: the raw ratio, uncapped — the 15.052 record reads +15.05. Below 1: the mirror, −1 ÷ ratio — a 2:1 buy-heavy book reads −2.00, 3:1 reads −3.00, exactly the sell side’s number. No floor, no cap.
A stop-hunt print that reads a cramped 0.05 raw reads −20.00 here — as loud as the event actually is.
| Book state | Raw | Normalized |
|---|---|---|
| The record ceiling | 15.052 | +15.05 |
| Ceiling zone begins | 5.000 | +5.00 |
| Heavy zone begins | 2.250 | +2.25 |
| Sell side 2:1 | 2.000 | +2.00 |
| Balance | 1.000 | +1.00 |
| Quiet middle ends | 0.800 | −1.25 |
| Buy side 2:1 | 0.500 | −2.00 |
| Floor zone begins | 0.200 | −5.00 |
| Stop-hunt territory | 0.100 | −10.00 |
| The record floor (a 19 s flash) | 0.005776 | −173.14 |
Reciprocal-mirror values: raw at 1 and above, −1 ÷ ratio below; nothing prints between −1.00 and +1.00. Normalized values are computed from the unrounded ratio. Zone edges from page 4; deepest sustained floor 0.0377 (−26.54), November 14.
The ten heaviest and ten emptiest days in ten months of second-by-second history — each day’s single most extreme reading, the book that produced it, and what price did next. Confirmed feed errors are excluded.
| Date (UTC) | Raw | Norm. | Bid BTC | Ask BTC | BTC price | +24h | +3d | +7d | +30d | 30d max | R pre |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 2026-08-19‡ | 15.052 | +15.05 | 7.85 | 118.15 | 69,417 | +4.1% | +10.9% | +12.5% | — | +17.3%† | — |
| 2026-08-21§ | 13.512 | +13.51 | 130.09 | 979.81 | 78,170 | −1.4% | +0.9% | −0.7% | — | +4.2%† | — |
| 2026-08-22§ | 12.940 | +12.94 | 64.90 | 499.07 | 77,550 | −1.6% | +3.9% | +0.2% | — | +5.0%† | — |
| 2026-08-27§ | 12.356 | +12.36 | 116.29 | 1,245.14 | 80,490 | −2.1% | — | — | — | +1.2%† | — |
| 2026-08-25§ | 12.032 | +12.03 | 21.06 | 253.34 | 80,704 | −2.2% | −0.7% | — | — | +0.9%† | — |
| 2025-10-28 | 11.131 | +11.13 | 63.87 | 710.91 | 115,295 | −3.3% | −5.1% | −12.5% | −20.7% | −29.8% | 0.23 |
| 2025-12-09 | 10.653 | +10.65 | 26.25 | 279.61 | 90,426 | +1.7% | +2.0% | −4.1% | −0.9% | −6.6% | 0.36 |
| 2025-10-26 | 9.544 | +9.54 | 18.42 | 175.83 | 115,000 | −0.3% | −2.8% | −4.1% | −24.4% | −29.4% | 0.23 |
| 2025-12-18 | 9.285 | +9.29 | 51.41 | 477.35 | 88,944 | −1.2% | −1.5% | −1.6% | +7.2% | +9.8% | 4.67 |
| 2025-10-20 | 8.327 | +8.33 | 12.04 | 100.28 | 111,211 | −2.9% | −1.1% | +4.0% | −17.5% | −19.5% | 0.12 |
“30d max” = biggest excursion within 30 days; “R pre” = the most extreme ratio in the 72 hours leading into that price extreme. ‡/§The top five rows are the live era, from the production feed — the same numbers the dashboard serves — with 30-day windows still open (—/†); “R pre” fills in when a window closes. Aug 19’s price has run +17.3% so far (its earlier +6.5% did not re-derive; corrected to +4.1%). Among the five settled rows the mirror holds: every bottom came in on a drained book (0.12–0.36), the one upward resolution on a book stacked to 4.67, the signal leading by up to 37 hours. From late Aug 21, 2026 the clock records each second’s true high and low — § marks those rows; their bid/ask cells are that second’s snapshot (ask ÷ bid gives the sampled second, not the peak).
| Date (UTC) | Raw | Norm. | Bid BTC | Ask BTC | BTC price | +24h | +3d | +7d | +30d | 30d max | R pre |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 2025-11-21 | 0.005776 | −173.14 | 330.58 | 1.91 | 81,924 | +0.9% | +4.0% | +9.0% | +5.3% | +13.0% | 3.46 |
| 2026-06-25 | 0.0264 | −37.94 | 923.20 | 24.33 | 58,175 | +1.6% | +1.9% | +5.2% | +8.8% | +13.6% | 2.96 |
| 2025-11-14 | 0.0377 | −26.54 | 1,166.30 | 43.94 | 95,824 | −0.4% | −0.7% | −14.3% | −7.2% | −15.7% | 0.23 |
| 2025-11-04 | 0.0506 | −19.77 | 387.81 | 19.62 | 99,602 | +4.4% | +3.2% | +3.3% | −7.8% | −19.0% | 0.23 |
| 2026-01-30 | 0.0605 | −16.53 | 468.82 | 28.36 | 81,585 | +2.4% | −5.3% | −20.7% | −18.3% | −26.1% | 0.14 |
| 2026-02-06 | 0.0647 | −15.44 | 902.45 | 58.43 | 60,168 | +15.9% | +16.5% | +9.0% | +11.2% | +21.9% | 4.36 |
| 2026-06-04 | 0.0661 | −15.13 | 1,121.81 | 74.12 | 62,192 | −0.1% | +0.7% | +1.4% | +0.4% | +8.0% | 1.85 |
| 2026-06-24 | 0.0707 | −14.15 | 954.61 | 67.45 | 60,250 | −1.3% | +0.8% | −0.2% | +6.3% | +11.0% | 2.96 |
| 2025-11-16 | 0.0735 | −13.60 | 1,007.48 | 74.06 | 94,088 | −0.2% | −5.1% | −7.6% | −7.0% | −14.1% | 0.23 |
| 2026-06-05 | 0.0735 | −13.60 | 1,031.79 | 75.86 | 60,433 | +0.5% | +5.6% | +5.4% | +3.6% | +11.1% | 1.85 |
Every row: a huge bid stack against a vanished sell side. Recoveries topped into stacked books (1.85–4.36), continued declines into drained ones (0.14–0.23) — knowable only afterwards. The June windows have since completed, both upward. Past patterns are not predictions.
BitClock came out of four months of pen-and-paper logging of the order book. Here is how the person who built it actually reads it — in his own words.
The clock is ambient. It sits on the lock screen and gets read the way you read a wall clock — dozens of incidental glances a day, zero taps.
Because extremes flash, the first question on any big move is retrospective: has the day already stuck its extreme?
A reading of 0.5 built on 900 bids vs 100 asks is not the same event as 0.5 built on 90 vs 10. That’s why the raw bid and ask totals stay on screen under the number.
He reads the ratio alongside price action and his other tools — never alone. The number tells you how the book is stacked; it does not tell you what to do.
Under Alerts, set a ratio threshold and BitClock notifies you when the book crosses it — above 1.0 alerts fire upward, below 1.0 they fire downward. If a threshold you enter doesn’t fit the display mode you’re in, the app says so on the spot and explains why.
That completes the ambient design: you don’t have to catch the flash. If the book ever empties toward the floor or stacks toward the ceiling while your phone is in your pocket, the clock taps you.
The whole point of the clock is a glance instead of a session. The wrist takes that to its logical end: the same reading, on your watch face, with your phone in your pocket.
Three surfaces: a watch-face complication — the live ratio as a number in a slot on whatever face you already run — a Tile one swipe away carrying the condensed clock stack (the ratio big, R-HOD and R-LOD, the BTC price), and a small companion app whose main job is relaying sign-in and data from your phone over an encrypted link.
Watch-face complications in the small accessory slots, and — on iOS 18 with watchOS 11 — the phone’s Live Activity appears in the watch’s Smart Stack, so a wrist-raise shows the reading without opening anything. It all ships inside the iPhone app.
There is no separate watch product and nothing extra to buy. The watch surfaces ship inside the BitClock phone apps: sign in on the phone, and the watch side comes with it. Add BitClock to a slot the way you add any complication when editing your watch face. It runs on the same account and subscription as the phone: the live ratio, BTC price and 24-hour stats show for everyone, and subscribers see the full stack — R-HOD, R-LOD and the depth pair — on the wrist too. A locked slot says so plainly: “SUBS ONLY”.
The ratio is the hero, and the smallest slots carry it alone — always a number, never a gauge. Larger slots and the Tile add the day’s extremes and the price: the same reading as the phone, ask depth ÷ bid depth within $2,400 of the price, shrunk to wrist size.
Both watch platforms strictly limit how often a face slot may refresh — minutes, not seconds. So the watch shows the latest reading it has, and when that reading is old it dims and shows its age (“· 2m ago”) instead of posing as live. The phone and the web dashboard are the live, once-a-second surfaces; the watch is the glance between them.
If it looks stale: a dimmed reading with an age stamp is the watch being honest, not broken — it catches up on the watch’s own schedule. The watch reads through your phone, so keep the phone app signed in and the phone nearby. And when the current second matters, go where it lives: the phone app or app.bitclock.net.
BitClock measures one thing: how the resting order book is stacked within $2,400 of the price, right now. It is not a signal service, it makes no predictions, and the record shows it spends most of its life in a quiet middle zone saying nothing at all. The interesting moments are rare — that is precisely what makes them interesting. And the first nine months on record were a single bear market — a 45% net decline containing three rallies; no bull-market data exists in the record yet.
Every statistic and every screenshot in this guide is real: the screenshots are unedited captures of the app, and every archive figure was recomputed from the full nine-month, 21.3-million-row record on August 8, 2026. The calibration zone shares were recomputed on September 5, 2026 from BitClock’s production record, window October 12, 2025 (first reading) through August 31, 2026, UTC (26,021,803 readings). On August 21, 2026 the record-book tables were recomputed from the production collector that succeeded the archive — the same feed the dashboard serves its customers, same formula, same screening, with the book’s own level counts as an added error check — and the June rows’ open 30-day windows were completed. One gap in the record: 2026-08-27, 10:51–15:32 UTC, when a 4 h 41 m hosting outage stopped collection entirely — nothing was recorded or displayed. On August 29, 2026 the record tables were extended with the first true-extremes-era entries (§). The combined record passed 26 million readings in late August 2026 and grows every second. Nothing here is illustrative, simulated, or estimated. For the full data story — every bracket, every extreme episode, and the methodology behind each claim — see the companion volume, How to Tell Time with BitClock.
Create an account at app.bitclock.net, sign in on your phone, and add the widget to your lock screen. That’s the whole setup — the clock does the rest.
Educational material. Not financial advice. BitClock describes the current and historical state of an order book; it does not recommend trades. Trading involves substantial risk of loss. Past patterns in order-book data are not a prediction of future prices.