Accuracy tracker — seconds per day, logged over time
How to check automatic watch accuracy at home: set the watch against an atomic reference (time.is), log the reference time and the time your watch shows, then log again after 24–48h. Rate = (watch elapsed − reference elapsed) ÷ days. Two checkpoints give your first rate; more checkpoints reveal the trend. In short, this tab is a complete watch accuracy test for automatic and quartz watches — no app install needed.
Add a checkpoint
Results
Saved automatically in this browser (localStorage). Nothing is uploaded anywhere.
Per-position worksheet — the interactive ISO-3159 style test
Watchmakers test accuracy in five positions: dial up (CH), dial down (FH), crown up (6H), crown left (9H), crown right (3H). Keep the watch in one position for 24h between its two checkpoints (or paste a manual s/d rate), and the worksheet computes the mean rate, mean variation, greatest variation and the horizontal-vs-vertical delta — the same four metrics COSC checks.
Worksheet summary
Standards checker — is your watch within spec?
Enter a mean daily rate (use the tracker's overall rate via the button) for a quick watch accuracy check against published watch accuracy standards — COSC, METAS and more.
Published accuracy standards (reference)
| Standard | Mean daily rate | Notes |
|---|---|---|
| COSC chronometer (ISO 3159) | −4 / +6 s/d | Bare movement, 15 days, 5 positions |
| METAS Master Chronometer | 0 / +5 s/d | Cased watch; also 15,000 gauss anti-magnetism |
| Rolex Superlative Chronometer | −2 / +2 s/d | After casing, Rolex in-house standard |
| COSC quartz (ISO 10553) | ±0.07 s/d | Quartz chronometer at 23 °C |
| Typical healthy automatic | −10 / +20 s/d | Rule of thumb, not a certification |
| Typical quartz | ±0.5 s/d (≈ ±15 s/month) | Thermocompensated quartz far better |
Timegrapher reading interpreter
Paste readings from any timegrapher (or a watchmaker's report) — rate, beat error, amplitude, lift angle, beat rate — and get a plain-English diagnosis: healthy, magnetized, needs regulation, or needs service. This is also where you learn how to read watch timegrapher output: what timegrapher amplitude, beat error and timegrapher positions actually mean.
Quick guide to timegrapher numbers
- Rate (s/d): the same seconds-per-day the tracker measures — but instant.
- Beat error (ms): tick-tock symmetry. ≤ 0.5 ms excellent, 0.5–1.0 acceptable, > 1.0 regulate.
- Amplitude (°): balance-wheel swing. 270–310° healthy, < 220° = service, > 330° = overbanking risk.
- Lift angle: 52° default; 38° for Omega co-axial. Wrong setting = wrong amplitude.
- Magnetized watch on a timegrapher: very fast rate (often +minutes/day) with erratic, jumping lines.
BPH / Hz converter
Beat rate conversions for setting your timegrapher or understanding a calibre's spec sheet.
Magnetism guide
- Symptoms: sudden gain of minutes per day in a previously fine watch.
- Causes: phones, speakers, laptops, magnetic clasps and bag clasps.
- Compass test: hold a compass near the watch and rotate the watch slowly — if the needle follows, it is magnetized.
- Fix: a demagnetizer (a few seconds, inexpensive tool). Never regulate a magnetized watch first.
- Prevention: METAS-certified (15,000 gauss) and silicon-hairspring watches resist everyday fields far better.
Power-reserve estimator
Tachymeter reader
Time one kilometre (or mile) with the chronograph; enter the seconds — the scale math (3600 ÷ seconds) is done for you.
Methodology & honest scope
The formulas
- Seconds per day: rate = (watch elapsed − reference elapsed) ÷ days. Two checkpoints are the minimum; the overall rate is duration-weighted across all intervals.
- Per-position worksheet: simplified ISO 3159 — mean of the 5 position rates; mean variation = average |rate − mean| (COSC limit 2 s/d); greatest variation = max − min (limit 5); vertical-minus-horizontal delta (limit −6/+8 s/d). A real COSC test runs 15 days in a lab; this worksheet is the at-home approximation.
- Timegrapher interpretation: thresholds from standard watchmaking practice — amplitude 270–310° healthy / <220° service; beat error ≤0.5 ms excellent / >1.0 ms regulate; lift angle 52° (38° co-axial); ≥ +120 s/d sudden gain suggests magnetism.
- BPH math: Hz = BPH ÷ 7200; beats/second = BPH ÷ 3600; ms per beat = 3,600,000 ÷ BPH.
What this tool cannot do
Honest scope: manual checkpoint tracking is not a timegrapher — it cannot measure beat error, amplitude, or positional behaviour, and reading the seconds by eye adds ~0.5–1 s of human error per checkpoint. The interpreter diagnoses from numbers you supply; it does not listen to your watch. No web page replaces a watchmaker's bench test for a sick movement.
Privacy
All computation happens in your browser. Checkpoints are stored in this browser's localStorage only — the page works fully offline once loaded.
Watch accuracy — frequently asked questions
How do I check the accuracy of my automatic watch at home?
Set your watch to an atomic reference such as time.is, write down the exact reference time and the time your watch shows, then repeat after at least 24–48 hours. Divide the difference by the number of days: that is your seconds-per-day rate. The Accuracy Tracker tab above does the math for you from two or more checkpoints — but be aware this is manual tracking, not a timegrapher: reaction time when reading the seconds adds a small error, so longer gaps between checkpoints give more trustworthy numbers.
How long should I test watch accuracy before trusting the result?
At least 24 hours between two checkpoints; 3–7 days is much better. A single 24-hour reading is noisy (temperature, position, and how fully the watch was wound all shift the rate). Log several checkpoints in the tracker and watch the 7-day rolling average — if it stays stable, the number is trustworthy. Per-position testing (the ISO-style worksheet above) needs one full 24-hour interval per position, i.e. about 5 days.
What is acceptable accuracy for an automatic watch?
There is no single law, but the benchmarks are clear: COSC-certified chronometers run −4/+6 seconds per day; a healthy non-certified automatic typically runs −10/+20 s/d. Under ±2 s/d is excellent, under ±6 is good, and anything beyond ±30 s/d deserves regulation or service. Quartz watches play a different game entirely — ±15 seconds per month is ordinary, and chronometer quartz is ±0.07 s/d.
Why would a watch run fast?
The most common reason why a watch would run fast all of a sudden — gaining minutes per day — is magnetization — the hairspring coils stick together, shortening the effective spring. Other causes: the watch was recently serviced and needs settling, the regulator was bumped, or amplitude is so high the movement overbanks. If your watch gained time gradually over months, it is more likely drying lubricants or wear. The timegrapher interpreter above distinguishes these cases.
What does it mean when my watch loses time?
Gradual loss usually means low amplitude — the balance wheel is not swinging fully, often from old lubricant, a weakening mainspring, or worn pivots. Loss that varies by position (fast dial-up, slow crown-down) points to pivot wear. A sudden large loss can be a slipped cannon pinion or a damaged hairspring. Consistent loss under ~15 s/d is often fixed by simple regulation; beyond that, service it.
How do I read timegrapher numbers?
Three numbers matter. Rate (s/d): how fast or slow the watch runs — the same figure the tracker computes. Beat error (ms): how asymmetric the tick-tock is; under 0.5 ms is excellent, over 1.0 ms needs regulation. Amplitude (degrees): how far the balance wheel swings; 270–310° is healthy, under 220° means service. Always set the correct beat rate (BPH) and lift angle first — a wrong setting makes every other number meaningless. Paste your readings into the interpreter tab for a plain-English diagnosis.
What is beat error on a watch?
Beat error is the difference, in milliseconds, between the tick and the tock of the balance wheel — it measures whether the balance swings symmetrically. 0.0–0.5 ms is excellent, 0.5–1.0 ms is acceptable, and above 1.0 ms the watch should be regulated (the hairspring stud or collet needs centering). High beat error also drags amplitude down, so fixing it often improves the rate too.
What is a healthy amplitude for a mechanical watch?
For a freshly serviced Swiss-lever movement, 270–310° in the dial-up position is the healthy band. Below 220° the movement needs service — regulation cannot fix low amplitude. Above ~330° risks overbanking. Two caveats: amplitude is calculated from the lift angle, so use 52° (or 38° for Omega co-axial) on your timegrapher; and compare positions — a drop of more than ~60° from dial-up to crown-down signals worn balance pivots.
What is lift angle and why does it matter?
Lift angle is the number of degrees the balance wheel rotates while the pallet fork is unlocking the escape wheel — 52° for most Swiss-lever movements, 38° for Omega co-axial. Timegraphers compute amplitude from the lift angle you enter: if you enter 52° for a co-axial calibre, the displayed amplitude reads roughly 27% too high. Set it from the movement's published specification before interpreting amplitude.
How do I know if my watch is magnetized?
The classic symptom is sudden, large gain — often several minutes per day — in a watch that was previously fine, usually after contact with a phone, speaker, laptop, or magnetic clasp. Confirm with the compass test: hold a compass near the watch and slowly rotate the watch; if the needle follows, it is magnetized. Fix it with a demagnetizer (a few seconds, ~$10 tool) — never regulate a magnetized watch, the rate will be meaningless until it is demagnetized.
Can I regulate my watch without a timegrapher?
Yes — to regulate a watch without a timegrapher, measure the rate over several days with the tracker, open the case back, and nudge the regulator lever a hair in the +/− direction (or turn the regulating screws on the balance). Re-measure over another few days and repeat. It is slow but works — a timegrapher just compresses days of waiting into minutes. If the movement has no regulator lever (free-sprung balance), or beat error is high, leave it to a watchmaker.
Phone accuracy app vs. timegrapher — which should I trust?
They answer different questions. A timegrapher measures the balance wheel directly (rate, beat error, amplitude per position) in minutes — a phone watch accuracy meter app (Watch Accuracy Meter, Twelve O'Clock, WatchCheck) or this web tracker measures the watch's displayed time against a reference over days, which captures the real-world average but cannot see beat error or amplitude. Use the tracker for 'how accurate is my watch day to day' and a timegrapher (or the interpreter here) for 'what is mechanically wrong with it'.
What is the difference between COSC and METAS certification?
COSC (Contrôle Officiel Suisse des Chronomètres) tests the bare movement for 15 days in 5 positions: the mean daily rate must be −4/+6 s/d. METAS (Master Chronometer, used by Omega) tests the finished, cased watch: mean daily precision of 0/+5 s/d, plus resistance to 15,000 gauss of magnetism, power-reserve and water-resistance checks. METAS is stricter on rate and adds anti-magnetism; COSC remains the industry baseline.
What does power reserve mean on a watch?
Power reserve is how long a fully wound watch keeps running — commonly 38–80 hours for modern automatics. The estimator above converts the manufacturer's rated reserve and how wound the watch is into remaining hours. Note: an automatic only reaches full reserve if worn actively (~8–10 hours of daily wear); a watch sitting in a box slowly runs down, and the rate often worsens as the mainspring unwinds.
How do I read a tachymeter bezel?
Start the chronograph and stop it after the watch travels one unit of distance (usually 1 km or 1 mile); the seconds hand points at your speed on the tachymeter scale. The math is speed = 3600 ÷ elapsed seconds, so 45 seconds over 1 km = 80 km/h. The converter above does the same calculation and also gives mph.
What battery does my watch take, and how accurate is quartz anyway?
Most quartz watches use silver-oxide button cells (SR626SW, SR920SW, SR927W and similar — the exact watch battery reference is printed on the case back or in the manual) and any watchmaker can swap one in minutes. As for accuracy: ordinary quartz runs about ±15 seconds per month, high-accuracy thermocompensated quartz about ±10 seconds per year — roughly 100× better than a good mechanical. If your quartz watch is suddenly minutes off, the battery is the first suspect, not magnetism.