Every wait time on Park Queue Times is the figure the park itself publishes, read from the park's own live feed every few minutes while the park is open, and stored. Nothing is estimated by hand and nothing is timed with a stopwatch. Everything else on the site, the typical waits, the crowd levels, the reliability tables and the crowd calendar forecast, is built on that stored record. This page sets out how, using real figures you can check against the pages they came from.
Where a wait time comes from
Each of the 87 parks we track publishes its own queue times, usually to its official app and to the signs at the ride entrance. We read that same feed for all 3,705 tracked rides, every few minutes through the park's published opening hours, and the readings pause while the park is closed.
Each reading is kept. A ride page such as Nemesis Reborn shows how many readings sit behind it, 11,767 at the time of writing, and every average on that page comes from those readings alone, not from other rides or other parks.
The number we store is the posted standby wait. Where a park also publishes a single rider line or a paid queue-skip product such as Premier Access, those are tracked as separate signals and are not folded into the wait figure you see charted.
What the numbers on a ride page mean
A ride page uses a handful of terms, and each has one fixed meaning.
- Typical wait: the average of the ride's recorded readings over recent weeks. Nemesis Reborn shows 38 minutes.
- Peak: the level the queue usually rises to on a normal day, rather than a one-off spike. For Nemesis Reborn that is about 50 minutes.
- Longest recorded: the single highest reading in the window shown, with its date. Nemesis Reborn's is 85 minutes on Tuesday 1 September.
- Quietest hour and busiest hour: the hours with the lowest and highest average wait across the ride's last 28 days. Nemesis Reborn averages 24 minutes at 10am and 52 minutes at noon.
- Day of the week: the average across the last 14 days with a recorded queue, so it can move from week to week.
- Uptime: the share of the park's own operating minutes the ride was running, across the last 14 recorded days.
The windows matter. The hourly pattern is the last 28 days, the day-by-day rail is the last 14, and the month-by-month chart is the ride's all-time record. A headline that leans on one window can disagree slightly with a chart built on another. The "best time to ride" line is simply the gap between the quietest and busiest hour, 28 minutes for Nemesis Reborn, the same arithmetic behind the best time to visit guide.
How daily averages and the crowd level are built
A park's daily average is the mean of every reading across every ride the park publishes on that day. On the Alton Towers page that runs at 20 minutes across the last 60 recorded days, and 32 minutes on the busiest of them.
The crowd level percentage makes days comparable. As each crowd calendar explains, the number is how full the park is: how long the queues run on its main rides in the middle of the day, on one scale for every park. Where a park sells dated tickets, the share of its capacity sold is used instead if that is higher.
| Crowd level | Roughly what it means in queue minutes |
|---|---|
| 50% | about a 17 minute queue |
| 70% | about half an hour |
| 90% | about an hour |
The word beside the number is different. Quietest, Quieter, Typical, Busier and Busiest compare the day with that park's own history, in bands of 0 to 15%, 15 to 35%, 35 to 65%, 65 to 88% and 88 to 100%. A small park can therefore be at its Busiest on a day that is still short queues by a big park's standards. On the best time pages the same idea is applied by month and by day of the week: Alton Towers averages 27% in December and 77% in August, and Wednesday is its calmest day at 51% against 67% on a Saturday.
Months with fewer than 10 recorded operating days are left unranked rather than averaged from a handful of days, which is why a seasonal park can show no figure for an early or late month.
How breakdowns and late starts are counted
Each ride's open or closed status is recorded on the same cadence as its wait. The reliability page for a park turns that into uptime, time lost and a breakdown count, over the last 60 operating days. For Alton Towers that is 13 July to 10 September, 40 tracked rides, and an average uptime of 79.6%, which the page grades as patchy.
The counting rules are strict, because a naive count misleads.
- Uptime is the share of the park's own operating minutes a ride was running, averaged over the 60 days.
- A breakdown is a closure lasting 10 minutes or more. Shorter stops still count towards time lost, but a 2 minute reset is not a fault.
- Touching closure windows are merged into one closure. A ride shut all day arrives from the feed as that day sliced into the sampling interval, and counting each slice would report one bad day as a hundred breakdowns.
- A late start is a ride that was simply not running at opening. It is counted separately, not as a breakdown. Alton Towers had 1,011 in its 60 day window.
- Closures in the final hour are left out of the breakdown-by-hour chart, because those are rides finishing their day rather than failing.
Uptime is then graded on fixed bands: Excellent at 95% and up, Good at 85 to 94%, Patchy at 70 to 84%, Poor under 70%. These figures include weather closures, so water rides and exposed coasters lose uptime in bad weather even when nothing is broken.
The crowd calendar forecast and how it is scored
The crowd calendar for each park runs 9 months ahead. Past days on it are measurements, not forecasts, and are never revised by the model. Future days are produced by Almanac, the forecasting model described in full on the almanac page, and refreshed weekly.
Almanac reads four things: the park's own recorded wait times, the calendar (weekends, school terms, public holidays and time of year), the weather forecast for near-term days, and its own track record, checked against what actually happened.
Accuracy is measured 6 weeks out, only on real days the model was never trained on, and is defined as 100 minus the average miss on a 0 to 100 busyness scale. The current published edition, 1.2.0, scores 87.3% across the 80 forecast parks. It is compared against the two simplest ways to guess a crowd.
| Park | Almanac | Last-week guess | Seasonal average |
|---|---|---|---|
| All parks averaged | 87.3% | 85.4% | 84.3% |
| Alton Towers | 86.5% | 83.3% | 83.2% |
| Thorpe Park | 85.5% | 72.0% | 91.1% |
| Paultons Park | 88.9% | 86.0% | 66.1% |
| Legoland Windsor | 70.9% | 56.3% | 56.9% |
Read the table honestly. Almanac is ahead on the all-park average and at most parks, but at Thorpe Park a plain seasonal average scores higher, and at Chessington a last-week guess does, 88.1% to 87.2%. Legoland Windsor is the hardest UK park to forecast.
Every edition has been scored the same way: 82.9% for the first edition in February 2026, 85.8% in April, 86.4% in June and 87.3% in August. A rebuilt engine, version 2.0.0, began forecasting on 28 August 2026 under a stricter rule that it may only use what was knowable on the day it made the call. Its accuracy will be published on the almanac page once enough of its forecasts have been scored, and not before.
What these numbers cannot tell you
- Posted waits are what the park publishes. They are the park's estimate, not a measured walk-through. Parks occasionally over-estimate queues by 5 to 15% to manage expectations, so the real walk can be shorter than the sign.
- Other queues are separate. A virtual queue return time is not a standby wait and is not the number charted. Single rider and paid queue-skip lines are tracked as their own signals where the park publishes them.
- A 0 can mean closed. Some parks send a plain 0 when a ride is shut rather than a closed flag. The live page shows the open or closed status beside the wait, with the reason where the park provides one, and that status is the one to trust.
- Feeds go dark. If a park's feed stops, the live page goes stale and that day's readings are thin or missing. Days like that are why the month pages insist on a minimum number of recorded days before ranking anything.
- Opening hours come from the park and can change. The hours page shows what the park has published, and the park's own site is the final word before you travel.
- Forecasts can be wrong. An 87.3% average still means some days miss by a wide margin, which is why the accuracy ledger is published rather than assumed.
Free ride alerts fire when a wait drops below a threshold you set, and Park Queue Times Plus adds per-park cheat sheets built from the same recorded history. The compare tool and the FAQ cover the rest.
Questions people ask
Are the queue times measured by people standing in the line?
No. Every reading is the wait time the park itself publishes, read from its live feed every few minutes while the park is open. It is the same figure shown in the park's app and on the ride entrance signs; the site records it rather than measuring anything independently.
Why does a ride show a different quietest hour in two places on its page?
Because the two figures use different windows. The hourly chart is the ride's last 28 days, the day-by-day rail is its last 14, and the month chart is its whole record. Each is labelled with its window on the page, and a headline can lean on one while a chart uses another.
What does a 60% crowd level actually mean?
It is how full the park is on one shared scale, where 50% is about a 17 minute queue on the park's main rides in the middle of the day and 70% is about half an hour, so 60% sits between the two. The word beside it, Typical or Busier for example, is judged against that park's own history, not against other parks.
How far ahead does the forecast go, and can I trust it?
The crowd calendar runs 9 months ahead for 80 parks, refreshed weekly. Its accuracy is scored 6 weeks out on days the model never saw, and the current edition averages 87.3%, ahead of a last-week guess and a seasonal average across all parks, though not at every single park. The almanac page publishes the full scorecard.