Wait Time Theory

[hiimdoug]

It's been a few years since I've been to WDW, but I remember walking by Small World and the wait time showing something like 60 minutes, but the line did not seem that long so I got on and the wait was less than half what the wait time was showing.

My theory is that wait times work like a stock. During a crowded day, if Space Mountain's wait time was 15 minutes you'd jump right on (buy the stock), but if it was 120 minutes you'd likely hold off (sell the stock). Like a stock, each ride has a point where there are no more buyers, a wait time that doesn't seem worth it to anyone. Once this point is triggered and the buyers evaporate there is a quick selloff in the wait time, too quick in fact for even the wait time to be updated. This creates an opportunity to "buy" a ride at an undervalued wait time.

I'm curious if anyone else has noticed this and done any research on the issue. A way to test this theory would be to somehow obtain the wait time data throughout the day and see if it is cyclical and to see where the peaks are.

 

[supersnoop]

I think you've just described the entire purpose of the Lines app from Touring Plans.

 

[hiimdoug]

Right after I posted this I did a quick google search and found the Touring Plans Lines app.

I also found this article: https://www.andrewmunsell.com/blog/predicting-disney-world-wait-times-neural-networks/ which references this site: https://www.parkgeni.us/

 

[candygirl75]

I have a Math degree in C&O so I'm totally geeking out over this thread! Thanks for posting... really interesting information. The intro to the Unofficial Guide / Touring Plans actually talks about the math behind the plans and is why I trust their predictions so much. I'm using their planning tools heavily for our upcoming trip.

 

[Klayfish]

Yes, it's well known that posted wait times at WDW are over estimated. It's done for a reason...several reasons I guess. First, could you imagine the horrors and the outrage guests would have if the wait was posted at 30 minutes and someone actually had to wait 31 minutes. They'd probably file a lawsuit. So they tend to overestimate.

Second, especially at closing, they try to discourage people from getting into the line, so they can get the last guests through and get the ride closed.

Yes, the Lines app is pretty good, assuming guests are feeding it constantly. Otherwise, with closing time not included, I typically have found the posted time to be 10-15 minutes over real wait time...but that's just a general ball park. YMMV...big time.

 

[rg35]

The problem with the Disney wait times is that they're not updated often enough to be a 100% useful tool, just as a general guideline. Disney only updates their wait times when a CM hands a rider one of those red cards to carry with them through the queue. I have no idea how often they distribute them, but it's not very frequent.

Since Disney knows exactly how long it takes people to move through a queue because almost every guest has a magic band now, I wonder if they'll ever update the wait time mechanisms to make use of this constant stream of data.

 

[4HOLIDAYS]

This is why I love touring plan app! Wait times from people actually in line/just rode. So worth the cost.

 

[wilkeliza]

Well the way they get times is based on what someone waited who is no longer in line. So whoever was given that red card to update the official wait time could have indeed waited 60 minutes or they returned it at the end of the ride instead of the beginning or they perhaps waited 30 minutes in line for their family to join and then 30 minutes for the actual ride. There are human errors to the flik card system.

At the end of the night they do indeed inflate times to discourage people form getting in a line that may force then to have to pay over time and double back etc.

 

[Spencer Wright]

Yes, it's well known that posted wait times at WDW are over estimated. It's done for a reason...several reasons I guess. First, could you imagine the horrors and the outrage guests would have if the wait was posted at 30 minutes and someone actually had to wait 31 minutes. They'd probably file a lawsuit. So they tend to overestimate.

Second, especially at closing, they try to discourage people from getting into the line, so they can get the last guests through and get the ride closed.

Yes, the Lines app is pretty good, assuming guests are feeding it constantly. Otherwise, with closing time not included, I typically have found the posted time to be 10-15 minutes over real wait time...but that's just a general ball park. YMMV...big time.

LOL @ having to wait 31 minutes! I have found wait times to be a slight over estimation by about a third of the time, but overall pretty accurate.

 

[hiimdoug]

I have a Math degree in C&O so I'm totally geeking out over this thread! Thanks for posting... really interesting information. The intro to the Unofficial Guide / Touring Plans actually talks about the math behind the plans and is why I trust their predictions so much. I'm using their planning tools heavily for our upcoming trip.

Awesome! I run a global-macro investment research firm, so the idea of finding arbitrage anywhere always intrigues me, even in wait times at Disney.

The problem with the Disney wait times is that they're not updated often enough to be a 100% useful tool, just as a general guideline. Disney only updates their wait times when a CM hands a rider one of those red cards to carry with them through the queue. I have no idea how often they distribute them, but it's not very frequent.

Since Disney knows exactly how long it takes people to move through a queue because almost every guest has a magic band now, I wonder if they'll ever update the wait time mechanisms to make use of this constant stream of data.

The idea wouldn't be so much as to find the actual or true wait time that you would wait if you were to step in line at that moment. Instead, the idea would be to see if the inaccurate data that Disney in posting as the wait time effects the behavioral attitudes of those willing to ride the ride.

In other words the theory would be, does it make sense to actually chase longer wait times or wait time spikes, based on the idea that the masses are chasing after the lower weight times. Much like investing (although inverted because you're trying to gain time not waiting) you're looking for buy high and sell low.

For an example, lets say our park has 100 people in it and 2 rides Ride A and Ride B. 70 People are currently on a ride or waiting in line while 30 people are now trying to decide what to ride next. Assuming each ride has equal appeal to the 30 people the deciding factor will be wait time. If when the 30 people make their decision on what to ride next, ride A's wait time is at 60 minutes while ride B's wait time is at 30 min. Again, assume all things being equal between the rides, the 30 people looking for which ride to ride next will chose ride B.

Assuming equal distribution, at the time of the decision on what ride to ride from the 30, Ride A would have had 46 people on it, while ride B would have had 24. If all 30 chose ride B, then ride B will now spike to 54 people on it while ride A will remain at 46. If this logic is true, it actually makes sense for you to chose the longer wait line.

 

[AngiTN]

Yes, it's well known that posted wait times at WDW are over estimated. It's done for a reason...several reasons I guess. First, could you imagine the horrors and the outrage guests would have if the wait was posted at 30 minutes and someone actually had to wait 31 minutes. They'd probably file a lawsuit. So they tend to overestimate.

Second, especially at closing, they try to discourage people from getting into the line, so they can get the last guests through and get the ride closed.

Yes, the Lines app is pretty good, assuming guests are feeding it constantly. Otherwise, with closing time not included, I typically have found the posted time to be 10-15 minutes over real wait time...but that's just a general ball park. YMMV...big time.

Yes! Can you imagine some of the parents who would complain that you dared to make their precious snowflakes wait in a line! When the sign said it would be 10 min but it was 20, or whatever. You scared them for life by making them wait for something.

 

[Babe the Blue Ox]

Line Stacking - The appearance from the outside is that the line is enormous when, in fact, most of the people waiting are those visible. This crowd-control technique, known as “stacking,” discourages visitors from getting in line. (Stacking is also used at several Disney attractions during the hour before closing to ensure that the ride will be able to close on schedule.) Despite the apparently long line, the wait is usually no longer than if you had been allowed to queue inside.

 

[lentesta]

You may be interested in Duke University's routing theory work and Furman University's Math and the Mouse research. Both of those have produced papers on theme park routing using our (touringplans') data.

We've also recently run a machine learning competition at the U of Central Florida, asking their data science students to come up with new ways to predict wait times at WDW. We're still looking at those ideas.

Duke's work is additionally interesting because its goal is to provide a formal framework for the kinds of questions @hiimdoug asks. IIRC, their original paper was the base case, for making a choice between 2 rides with different popularity and capacity. They were extending it to N rides.

I talked a little bit about the "ride A or B" problem for WDW in my masters thesis. Here's a chapter from it - section 4.3.4 explains why short-term optimization can lead to worse long-term outcomes. It's one reason why the touringplans optimizer operates only on complete tours. Malandraki's routing stuff is generally well-written, and it's available on the web. (Malandraki was a lead research scientist at UPS; Dial worked for the Volpe National Transportation Lab at Harvard. I think they both got a laugh out of the application of their algorithm to Disney World.)

Other theory-related stuff we've worked on that involves Disney World:

Elon University's The Science of Happiness at Disney produces a wide range of term papers on various aspects of visiting WDW. Some of that work has been directly incorporated into the Unofficial Guides.

Middlesex County College uses our data to teach basic statistics with real-world applications. (I love the idea that the extra walking cancels out all the extra calories one eats in WDW.)

In the popular press, touringplans is a case study in Numbers Rule Your World by Kaiser Fung. There's less math in there, but a good explanation of why Disney World waits are generally predictable, and the general theories behind all of it. That includes the concept of balking times, which the OP refers to.

 

[hiimdoug]

You may be interested in Duke University's routing theory work and Furman University's Math and the Mouse research. Both of those have produced papers on theme park routing using our (touringplans') data.

We've also recently run a machine learning competition at the U of Central Florida, asking their data science students to come up with new ways to predict wait times at WDW. We're still looking at those ideas.

Duke's work is additionally interesting because its goal is to provide a formal framework for the kinds of questions @hiimdoug asks. IIRC, their original paper was the base case, for making a choice between 2 rides with different popularity and capacity. They were extending it to N rides.

I talked a little bit about the "ride A or B" problem for WDW in my masters thesis. Here's a chapter from it - section 4.3.4 explains why short-term optimization can lead to worse long-term outcomes. It's one reason why the touringplans optimizer operates only on complete tours. Malandraki's routing stuff is generally well-written, and it's available on the web. (Malandraki was a lead research scientist at UPS; Dial worked for the Volpe National Transportation Lab at Harvard. I think they both got a laugh out of the application of their algorithm to Disney World.)

Other theory-related stuff we've worked on that involves Disney World:

Elon University's The Science of Happiness at Disney produces a wide range of term papers on various aspects of visiting WDW. Some of that work has been directly incorporated into the Unofficial Guides.

Middlesex County College uses our data to teach basic statistics with real-world applications. (I love the idea that the extra walking cancels out all the extra calories one eats in WDW.)

In the popular press, touringplans is a case study in Numbers Rule Your World by Kaiser Fung. There's less math in there, but a good explanation of why Disney World waits are generally predictable, and the general theories behind all of it. That includes the concept of balking times, which the OP refers to.

Wow, thanks for this response. I'll have to read through the research. I guess it really shouldn't be much of a surprise but, I'm amazed at home much time has been spent and the depth of research that has gone into cracking this code. Very neat.

I also went ahead and download and subscribed to the touringplans Lines App. Looks intriguing. I'll see how it plays out next weekend =)

 

[laura2287]

Great thread. Looking forward to reading some of these resources. Working in the research industry, this is all so exciting to me!

 

[hakepb]

The problem with the Disney wait times is that they're not updated often enough to be a 100% useful tool, just as a general guideline. Disney only updates their wait times when a CM hands a rider one of those red cards to carry with them through the queue. I have no idea how often they distribute them, but it's not very frequent.

Since Disney knows exactly how long it takes people to move through a queue because almost every guest has a magic band now, I wonder if they'll ever update the wait time mechanisms to make use of this constant stream of data.

Actually, I think Disney gets relatively frequent, useful data...then posts different times.
For a few months, easywdw found access to Disney's actual wait times and had it on his Wait Times page, plus the regular, posted times. That real time data helped me find a few additional rides ... but I also saw times where it was too low. (Like how PP might be low during a parade, then gets a large rush of guests after the parade)

 

[dreiwhit]

Of course, sometimes the standby times are bang on. I rode Pirates once and the wait time was 45 minutes. It took exactly 45 minutes before we boarded a boat.

Another time, Enchanted Tales with Belle had a 25 minute wait. At the 35 minute mark, and with a big line still in front of us, we gave up and left.

 

[DVCcurious]

where are they?

I think they're talking about us

 

[Doug S]

I've been collecting wait times from Disney since spring this year and every time I try to build a model I end up realizing why I just give my money to Touring Plans. I have no stats degree and their numbers are pretty accurate anyways. It would be fun to figure out how to build one, but it's a bit overwhelming if you don't understand how to build a prediction model.

 

[PDXmouse]

So several of you stated Touring Plans does a good job. Dies that mean I should trust their assumptions about my touring plan that I'm building for January? In many cases it has 0-5 minute waits on almost everything (for the most popular I plan to get FP's.)