2021 Disney World Crowd Calendar Retrospective: Park by Park (2023)

Last week we held a little contest between Disney and TouringPlans as we reviewed the resort-wide performance of our Disney World Crowd Calendar. Travelers may choose the overall time of their visit based on the resort crowd level, but they often choose which park to visit on which day by comparing crowd levels at the parks. So this week we’re back for more review, but instead of the resort score, we’ll be looking at the individual park scores. It won’t be as exciting because we can’t beat up on Disney (just kidding), but we’ll do a little park-to-park showdown instead.

There are two ways we could walk through this: park by park, or month by month. I’m choosing to go through time because it will help us see how park-level errors might be related to each other. Also, it will save me from writing “and then in August we see that COVID flattened attendance” again and again on each park analysis.

A little intro for the visuals: for each month below forecasts are on top, and the error (predicted – observed) is on the bottom. For quality control a miss in either direction is bad, so the error calendars are orange and purple to avoid implying that green is good. But just so you can have a little hint to remember which way the error is: the number is the predicted level minus the observed one. That makes crowd levels that are worse than expected (underpredictions) come out as negative numbers. Errors of only +/- a single crowd level are not colored at all since we consider those to be “good enough”.

There’s not a lot to see in January and February. Weekends in Epcot are consistently overpredicted throughout both months. The Epcot Festival of the Arts ran from January 14 to February 21, and that means that weekend crowds get a boost from locals. The missing crowds here might mean that the boost wasn’t as strong this year, but there are probably also some out-of-state visitors who might have taken a quick weekend trip in other years and decided to postpone travel in light of the post-holiday COVID surge.

That was a super-quick summary for January and February, but we’re going to spend alot of time with March and April. This is the first time we’re going to see some consistent trends and I want to really dig into them.

Question 1: Why don’t those underpredictions for Epcot show up in the resort calendar?

You might see that there’s some overprediction in the other parks, and assume that it’s averaging out if you add them together. There would be a lot of truth to that, and it’s why we can’t rely on the resort calendar to say how well we’re doing in the parks. But it’s not quite as simple as just adding up the park numbers and dividing by 4.

If you were totally obsessive very detail-oriented like me and you tallied up a few different days, you might notice something odd. On the 3rd Monday in March, the sum of the error from all the parks is -2, and the resort-wide error is 0. Two days later, on Wednesday, the sum of the error from all the parks is also -2, but the resort-wide error is -1. Whoa … what’s going on here?

The answer is in how the crowd levels are calculated. We start with a crowd level for each attraction, which is based on the wait-time distribution for that specific attraction. For forecasts this number comes from a prediction model and for observed levels it’s just what we saw in the parks.The resort crowd level is calculated by removing the highest and lowest predictions for each park and averaging the rest. Since the parks have different numbers of attractions, each park ends up having a different “weight” in the resort calendar. So you can see a general trend where over- and underpredictions may cancel out, but there’s no formula that you can use to go directly from the park levels to the resort level.

Question 2:Why do we see so much underprediction for Epcot?

Some of those misses are pretty severe, where we were off by 4 or 5 levels. This is Spring Break and Easter when crowds are usually very heavy, but to understand what happened with the calendar we need to remember where we were in pandemic time. More and more people were eligible for vaccination every day, but we were still a ways off from having a high percentage of the population fully vaccinated, and restrictions were still common in many places. Disney World was open, but attraction capacity was reduced, with rides loading every other row or every other car for social distancing. And overall park capacity — which is made up of not just attractions but also dining, shopping, and walking around — was severely limited.

Take a look at the chart below, which is a snapshot from April 28 that shows how far in advance Park Passes sold out for each date. All we care about here is the difference between the days that are gray, with passes available, and the days that are colored, showing that the park was completely booked.

(Video) One Family vs. The World, A Malcolm in the Middle Retrospective

You can see that for most days in this period all the parks are full except Epcot, and even Epcot is booked up for a few of them. We know that the resort overall was operating at 35% capacity during this period, but that capacity wasn’t evenly distributed among the parks. Looking at the crowd calendar, it turns out that we rarely observed a crowd level of more than 3 in the Magic Kingdom during this time. Since the park was at capacity (which we know because there were no Park Passes available), that means that the maximum crowd level achievable was about a 3 — our predictions of 4 and 5 over most of these weeks had no hope of coming true. We see a similar hard limit with Hollywood Studios, and though we don’t see it as clearly with Animal Kingdom, that’s probably because our forecasts for Animal Kingdom were lower to start with.

So we have a lot of people who might have been pushing our wait times up to 5 in Hollywood Studios, except that only “level 3’s worth of them” can get in because Park Passes are gone, and the same thing is happening in Animal Kingdom and Magic Kingdom. Where are all the people who couldn’t get into those parks going instead? Epcot! They’re going to Epcot, because Epcot turns out to have the lion’s share of that overall capacity, and that’s why wait times in Epcot were so much higher than our forecasts. It turns out that overpredictions and underpredictions compensating for each other in the resort calendar wasn’t accidental — it happened because we did a decent job of predicting the natural demand for attractions, but capacity limits forced people to distribute themselves among those attractions differently than we (and often they) had envisioned.

May looks a lot like January and February, with overpredictions on the weekends in Epcot (it’s Flower & Garden time now) and to some extent in the Magic Kingdom. There’s really nothing very exciting here, which to be honest is exactly the way we like it.

If we look closely, we can see that the max crowd levels have changed. Park Passes are still sold out every day in most parks, but now we’re seeing mostly 2-3 in Magic Kingdom, just 1 in Hollywood Studios, and also a max of 1 at Animal Kingdom. This is a good time to remind you that our crowd levels measure wait times and not the number of people in the parks. As we moved forward into the summer, vaccination levels rose and COVID levels dropped. Disney didn’t immediately respond by increasing overall capacity in the parks, but they did start to reduce social distancing and increase attraction capacity. Wait times (and our observed crowd levels) went down as a result.

In the early part of June our forecasts start to really climb as schools begin to let out. Most of the parks are still sold out every day, but we don’t see the massive underprediction for Epcot that we saw in March and April. Why not?

It’s because of the increase in attraction capacity that we saw in May. Guests were still getting blocked out of other parks and ending up at Epcot instead, but our higher forecasts for Epcot combined with the additional capacity to bring us closer to the mark. Now it’s the overprediction in the other parks that’s more extreme, because our predicted levels went up while the maximum attainable levels went down.

In early July we continue to see some overpredicting, but there are big signs that Disney has increased overall capacity limits. In late June more Park Passes were added for July, and on many weekdays passes are now available at all four parks. You can see this change in observed crowd levels as well, as we’re now able to see 5 in Hollywood Studios early in the month and 6 by the end of it. In mid-July we updated crowd levels for the rest of the summer to reflect these changes in attraction capacity.

The end of August usually sees lower crowds than the first two weeks, but not as low as the predictions you see here — COVID’s Delta wave is to blame for these exceptionally low crowds. Our forecasts for the last two weeks of August were pretty accurate whenever they were 1, but the reason we have so many 1’s on the calendar is that we responded to Delta by lowering our predicted levels in early August, and then lowering them again 12 days later. Those updates bought us smooth sailing all the way through September. September looks beautiful, doesn’t it? If only every month could look as beautiful as September.

(Video) The WEIRD World of Hey Arnold!

October definitely does not look as beautiful as September. The first weekend in October was Walt Disney World’s 50th Anniversary, and pretty much everyone expected large crowds in the Magic Kingdom and Epcot. There was so much expectation of high crowds that we broke our own rules and predicted 10’s at Magic Kingdom and Epcot even though our predictions based on wait times were nowhere near that high.

On the actual day the expected crowds did materialize; the Magic Kingdom was jam-packed and even closed to park hopping at one point. Epcot was also very crowded. But most of the people in each park seem to have been buying merch, eating, or soaking in the atmosphere, because the wait times were so low that the observed crowd level of both parks was a 1. The overprediction of +9 on this day comes out to be our biggest observed error of the year, which is not surprising because it is literally as wrong as you can possibly get on a scale that goes from 1 to 10.

I don’t think anybody really knew what the Anniversary crowds were going to look like after the first weekend. Our models showed something in the middle of the road, but after much lower crowds in the first week, we adjusted our forecasts downward for the rest of October. Unfortunately, that week turned out to be the calm before the Columbus Day storm. This is what we usually say about Columbus Day:

We do not usually see these Monday crowds keep going, and going, and going all the way through the week like the Energizer Bunny. We’re not sure what happened, but we think that it might have been families taking vacations that they had rescheduled from August. It was a rough 6 days for us on social media as we tried to decide whether we were seeing something temporary that we needed to ride out or an ongoing trend that we needed to adjust for, but on the 7th day we put the original forecasts back. And then everything was good. Kind of. Except for Hollywood Studios, especially in the last week. You know what? Let’s leave this dumpster fire of a month behind and move on.

In the first part of each month when crowds are lower we have a strong and confusing mix of hit and miss. Is there rhyme or reason? It’s hard to say. As we move through the Epcot Festivals we’ve arrived in Food & Wine season, COVID hotspots are continually changing, Lightning Lane queues have gone into operation, the weather has gotten more variable, starting in December some 5-11 year olds are fully vaccinated (which many families were waiting on for travel), and Annual Pass sales have resumed. We know that all of these things can affect both demand and demographics at Disney World, and trying to sort out how they’ve combined here is likely to be an exercise in futility.

In the heavy holiday seasons at the end of each month, we come back to a very familiar story. Park Passes are sold out in every park. We’ve predicted high crowd levels, but our forecasted levels are unattainable even though overall capacity has risen at every park; the result is overprediction by 3-5 crowd levels. Epcot is the only exception to this, and on December 19 and 20 we even see the underprediction come back at Epcot as it fills to its “brim” of 9 even though we’ve only predicted it at 7 and 6.

To compare park to park performance, we’re going to plot the number of days at each error level by park.

Ideally we want these distributions to be hill-shaped (the geek term is “unimodal”), and the good news is that they are. We’d like for them to be centered around 0, meaning that more days were dead on target than any other number, and again they mostly are — only the distribution for Magic Kingdom misses with 1 as the most commonly seen error level. The measurement that I used for ranking is On Target prediction, which is the percentage of days with prediction error between -1 and 1. Every single park has better than 50% On Target performance (yay!), but Animal Kingdom is the clear winner at 66.6%. You might feel like these results are surprising given how much time we just spent talking about where the calendars were struggling with accuracy, but remember that there were quite a few months when the calendars performed well — it’s just that we zipped past them quickly in the analysis.

I joked about beating up on Disney last week in the resort-wide calendar contest, but this week it really feels a bit like Disney was beating up on us. Of course, in reality Disney was just responding to changes in the pandemic and in some cases — like with plummeting attendance in August and September — they had as little control over external forces as we did over their operational changes.

You might wonder why we didn’t simply adjust our forecasts based on what we were seeing with capacity limitations. That is a terrific question and we would have loved to be able to do that. The problem is that the park crowd levels are built starting from the attractions and working up. To fix the mismatches we saw between our crowd levels and the park capacity, we would need to build new models and rescale crowd level distributions for each attraction in the affected parks. Both of these processes need a pretty large sample of days to work with, more than we could reasonably expect to collect before capacity changed again.

(Video) Panzer Dragoon Series Retrospective - A Complete History and Review

Our very first crowd calendar update in 2021 was titled “Post COVID-19 Crowd Calendar Update“, and discussed our prediction that as vaccinations rolled out Disney would begin to undo COVID-related capacity and operational changes and be mostly back to normal by summer. Obviously it did not work out that way, and we definitely hope for less of a rollercoaster ride in 2022. Our calendars performed pretty well this year despite the challenges, and here are our suggestions to help you get the best experience with them.

  • Check crowd-level predictions shortly before your trip, since we make updates on a regular basis. Subscribers can use the crowd tracker and get notified automatically when we make updates to days they’re following.
  • Take the time to look at the predicted attraction wait times for the park you plan to visit. Sometimes we get complaints from people who are in the parks on a level 3 day and are surprised by a 40-minute line; they don’t realize that what they’re seeing might be par for the course at level 3. Having realistic expectations can be the key to a great day.
  • If you use a touring plan, reoptimize it every now and then during the day. Once the park opens our prediction is frozen in history, but we update our day-of wait time estimates frequently based on what we see in the parks. Even if our predicted crowd level turns out to have been way off, reoptimizing will help you to have an optimal day.
  • Be careful with the Epcot crowd level. We don’t expect to continue seeing the underprediction issues with Epcot that we saw in the spring this year, but our calendars are based on wait times and Epcot has a lot of Festivals that can fill the World Showcase with people who are not riding any rides. If there’s a Festival going on then the difference between our prediction and what the crowd “feels like” will be most noticeable on weekends and in the late afternoons and evenings.

Are you planning a visit to Walt Disney World this year? What do you think the crowds will be like? Will all the parks return to 100% capacity soon? Let us know in the comments!

You May Also Like...

  • Universal Crowd Calendar Update for March 2021

    Wait times at Universal Orlando Resort are up, especially at the Universal Studios Florida park. We have seen crowd levels at…

  • WDW Crowd Calendar update for late 2021

    (Video) Castlevania Series Retrospective - A Complete History and Review

    ThisCrowd Calendar update just affects November and December 2021. The prior predictions were based on data prior to 2019. We expect…

  • February 2021 – WDW Crowd Calendar Update

    Crowds have mostly been low so far for 2021. We expect the week of President's Day and Mardi Gras will be…

  • Disneyland Crowd Calendar Update for November 2021

    Happy Thanksgiving Everyone! Well, almost. We are anxiously waiting to see how many people head to Disneyland Resort for the holiday.…

    (Video) From RollerCoaster Tycoon to Planet Zoo | The Death & Return of Park Builders


1. The Story of Super Mario World | Gaming Historian
(Gaming Historian)
2. Mad Tea Party | Magic Kingdom (Ride Through)
(Leal Legacy)
3. A Bully Retrospective : Rockstar's Neglected Son
4. OFF: A Retrospective (Feat. @NezumiVA)
(Sagan Hawkes)
5. One Last 3D Mario Retrospective | Jenny Geist
(Jenny Geist)
6. How To Plan a Walt Disney World Vacation!
(The Vacationeer)
Top Articles
Latest Posts
Article information

Author: Patricia Veum II

Last Updated: 12/02/2022

Views: 6521

Rating: 4.3 / 5 (64 voted)

Reviews: 95% of readers found this page helpful

Author information

Name: Patricia Veum II

Birthday: 1994-12-16

Address: 2064 Little Summit, Goldieton, MS 97651-0862

Phone: +6873952696715

Job: Principal Officer

Hobby: Rafting, Cabaret, Candle making, Jigsaw puzzles, Inline skating, Magic, Graffiti

Introduction: My name is Patricia Veum II, I am a vast, combative, smiling, famous, inexpensive, zealous, sparkling person who loves writing and wants to share my knowledge and understanding with you.