Are Tokyo Disney crowd and wait-time forecasts accurate? Verified results and precision explained
Yes, you can plan with confidence. With the forecast we publish the day before, only 10 in 100 days turned out busier than forecast. Picking days by last year's crowds gave 15 to 17 in 100, and other forecast sites gave 21 to 23.
Who this is for:Anyone wondering whether they can rely on the crowd calendar and wait-time forecasts to plan their visit
Contents (8)
Is the crowd calendar accurate?
With the forecast we publish the day before, only 10 in 100 days turned out busier than forecast. For example, on 90 in 100 days marked "Moderate," it was actually "Moderate," "Fairly quiet," or "Quiet." The worst case is expecting a quiet day and finding crowds, so days that may tip one level busier get a ▲ on the calendar. We redo the forecast every night, up to a year ahead. Narrow your dates early, then check again as your trip gets closer.
Sources:[1]Offpeak Parks data
How accurate is the AI that forecasts wait times?
The official app shows the wait at this very moment. A 40-minute wait can be 60 minutes by the time you walk there. Offpeak Parks' AI takes in the latest waits and forecasts what comes next. The average gap between forecast and actual waits was about 6 minutes up to 1 hour ahead, about 8 minutes for 1 to 3 hours ahead, and about 9 minutes beyond 3 hours.
Sources:[2]Offpeak Parks data
Lower is better. Checked across all attractions in October 2025 and January, April and July 2026.
What does the AI that plans your route do?
Pick the rides you want, and the route planner builds an order that keeps your time in line short. We compared it on the same days' actual waits with heading to the most popular ride at opening and then riding in order of longest wait. Across the 22 model courses we publish, the planned route let you ride more on 245 of 271 days. For example, Enchanted Tale of Beauty and the Beast had a 2 hr 40 min wait first thing in popularity order. The planned route rode it in the 3 p.m. hour with a 50-minute wait (September 26, 2026).
Sources:[3]Offpeak Parks data
- STEP 1Forecast the day's waits
For each ride, it forecasts how waits rise and fall from morning to night.
- STEP 2Build the shortest-wait order
It keeps your ride priorities, meals and show times, and rides when lines are short.
- STEP 3Re-plan after every ride
After each ride, it rebuilds the rest of the order with the current waits.
How does it compare with other approaches?
How does it compare with other approaches?We compared on the same days with the same answer key: each day's actual crowd level, worked out from actual waits. The five other forecast sites are kept anonymous as Sites A to E.
Sources:[4]Offpeak Parks data
| What we compared | Offpeak Parks | Compared with |
|---|---|---|
| Days busier than forecast (Disneyland) | 10 in 100 days | Going by last year: 15; Site A: 21 |
| Days busier than forecast (DisneySea) | 10 in 100 days | Going by last year: 17; Site A: 23 |
| Six comparisons with five other sites, by period and park | Fewest days busier than forecast in 5 of 6 | Sites A to E |
| Gap between forecast and actual waits (all lead times) | About 8 min | Assuming the current wait holds: about 14 min |
| Rides per day (22 model courses) | More rides on 245 of 271 days | Riding in popularity order |
📅 Forecast vs. reality, week by week
Forecast vs. reality, week by week
Since Wed, Sep 30, 2026 every nightly forecast has been kept as issued (never rewritten). At the end of each week we compare it with the measured crowd level (1–5) of each day in both parks.
Last week (Nov 2–Nov 8, 2026): within one level 14 of 14, using the forecast issued 7–9 days ahead. Since Wed, Sep 30: 75 of 80 within one level.
| Week (Mon–Sun) | Forecast issued | Within one level | Quietest-day call | Since Wed, Sep 30 |
|---|---|---|---|---|
| Nov 2–Nov 8, 2026 | 7–9 days ahead | 14 of 14 | 2 of 2 parks | 75 of 80 |
| Oct 26–Nov 1, 2026 | 7–9 days ahead | 11 of 14 | 1 of 2 parks | 61 of 66 |
| Oct 19–Oct 25, 2026 | 7–8 days ahead | 14 of 14 | 2 of 2 parks | 50 of 52 |
| Oct 12–Oct 18, 2026 | 7 days ahead | 13 of 14 | 2 of 2 parks | 36 of 38 |
| Oct 5–Oct 11, 2026 | 5–7 days ahead | 14 of 14 | 2 of 2 parks | 23 of 24 |
| Sep 28–Oct 4, 2026 | 0–4 days ahead | 9 of 10 | 2 of 2 parks | 9 of 10 |
One comparison = one park on one day (7 days × 2 parks = 14). “Within one level” means the forecast and the measured crowd level (5 levels) differed by at most one level. “Quietest-day call”: the day we forecast as the quietest of the week was among the two quietest measured days. Rows appear once at least 10 of the 14 comparisons are available. Next update: Mon, Nov 16, 2026. When quoting, please link to this section (#weekly-check).
As of Fri, Nov 13, 2026. A new row is added automatically every Monday.
📊 Monthly comparison with other forecast sites
Month by month: Offpeak Parks vs. other forecast sites
Since Thu, Oct 8, 2026 we save the forecasts other sites show every morning and score them against ours with the same measures. Each row adds one month, automatically.
| Month | Compared with | Days (both parks) | Match rate: Offpeak | Match rate: other site | Quieter-day pick: Offpeak | Quieter-day pick: other site |
|---|---|---|---|---|---|---|
| October 2026 | Site A | 48 | 54% | 35% | 81% | 51% |
| October 2026 | Site B | 48 | 54% | 37% | 81% | 59% |
| October 2026 | Site C | 48 | 54% | 29% | 81% | 64% |
| October 2026 | Site D | 48 | 54% | 45% | 81% | 67% |
| October 2026 | Site E | 48 | 54% | 34% | 81% | 63% |
Match rate: share of days where the forecast hit the measured crowd level exactly, after mapping every forecast to the same 5 levels. Quieter-day pick: of two days with different crowds, how often the forecast picked the quieter one (50% is a coin flip). Offpeak uses the forecast issued the night before; other sites use the value on their site by the morning of the day. Other sites are anonymized; the same letter is the same site.
As of Fri, Nov 13, 2026. A new month is added automatically.
How to use forecasts effectively
- When choosing dates
- Star 2-3 candidate dates on the Crowd Calendar to compare
- Narrow down dates early without waiting for ticket releases
- As your visit approaches
- Build an itinerary with Route Planning to optimize ride order
- Re-check 1 week before as models update nightly with fresh weather data
- On the day of your visit
- Check Real-time Waits for current lines and hours-ahead forecasts
- Recalculate route order after each major attraction
Star candidate dates to let Suki highlight the best options.
Compare candidate dates on Crowd CalendarRead next
This is guide 16 of 33 under “Plan the trip”.See them all
Frequently asked questions
Q.Is this the official crowd forecast?
No. Offpeak Parks is not affiliated with Tokyo Disney Resort. We make our own forecasts from past wait-time records.
Q.How do you make the forecasts?
An AI trained on about 11 years of wait-time records redoes the crowd forecast every night, up to a year ahead. On the day, it forecasts waits while taking in the latest ones.More in "How accurate is the AI that forecasts wait times?"
Q.When should I check the forecast?
It's fine to narrow your dates months ahead. Check again as your trip gets closer; near the day we also use the weather forecast.More in "Is the crowd calendar accurate?"
Q.What about weather and rain?
Days far ahead are forecast from holidays, school breaks and ticket prices. Near the day and on the day, we redo the forecast with the weather forecast and current waits, so it keeps up if rain changes the crowds.More in "How accurate is the AI that forecasts wait times?"
Q.Is it more accurate than other crowd forecast sites?
We compared with five other forecast sites six times, by period and park. We had the fewest days busier than forecast in 5 of the 6.More in "How does it compare with other approaches?"
Tools:Live waitsBuild a routePacking list
Who writes this and where the numbers come from:About Offpeak Parks
Sources for this article (4 sources, checked October 20, 2026)
Checked on October 20, 2026. Prices and rules can change, so check the official site just before you go.
- [1]Offpeak Parks dataOffpeak Parks crowd forecast check (day-before forecasts, April 6 to October 4, 2026, both parks)Compared the day-before forecast with each day's actual crowd level from actual waits, and counted the share of days busier than forecast. Going by last year's crowds was counted the same way.
- [2]Offpeak Parks dataOffpeak Parks wait-time forecast check (October 2025 and January, April and July 2026, every 15 minutes)Forecasts used only what was known at the time; we averaged the gap from actual waits. The comparison is assuming the current wait holds.
- [3]Offpeak Parks dataOffpeak Parks model course replays (22 published courses, 271 days, September 20 to October 11, 2026)On the same days' actual waits, compared the planned order with riding in popularity order (most popular at opening, then longest wait first).
- [4]Offpeak Parks dataComparison with five Japanese crowd forecast sites (April to October 2026)Counted the share of days busier than forecast, with the same answer key and period.