Predicting The 2024 Olympic Winning Times

This article originally appeared in the 2024 Olympic Preview edition of SwimSwam Magazine, courtesy of author Daniel Takata. Subscribe to the SwimSwam Magazine here.

This summer, the Olympic Games will take place in Paris. And, as usual, there is a lot of expectation for great performances, legendary races and, of course, world records. Many will try to guess who the winners and medalists of the swimming events will be. As always, some favorites will confirm favoritism and some underdogs will appear. Obviously, it is very difficult to predict who they will be.

But we can use quantitative methods to make another type of prediction. Based on recent results, what are the estimated times needed to win the races in Paris? And which records are most likely to be broken?

It doesn’t take a genius to realize that there are events in which the world record is highly unlikely (for example, the men’s 200 freestyle and the women’s 200 butterfly) and others in which the record is expected, given the evolution in recent years (e.g., women’s 400 freestyle, a race that will feature the last three world record holders).

But it is possible to quantify this probability using statistical methods.

Estimating the improvement in Olympic years

To get an idea of the times needed to win the gold medal at the Olympic Games, we need to keep one thing in mind. Historically, there is a significant improvement in athletes’ times in Olympic years. This statement may sound obvious, and it really seems to be the impression that many people have, but it is necessary to analyze the data. And the data shows exactly that.

Let’s look at the evolution of the men’s 100 freestyle over the years. The graph below shows the average time of the fastest 100 swimmers in each year since 1990.

The decreasing pattern is evident, as we would expect. But we can observe another interesting aspect. Every four years, it is possible to notice a dramatic drop in the average time, which is not a coincidence since it corresponds to the Olympic years. The only exception is the period 2008-2009: The times decrease from 2007 to 2008, and decrease even more from 2008 to 2009, which is unusual, since 2009 was not an Olympic year. Of course, the hi-tech suits were the cause.

The same pattern is observed in all other events included in the Olympic program. This needs to be considered when predicting swimming performances for 2024. And, to make this prediction, the statistical method considers the distribution of times of the 100 fastest swimmers each year, in each event.

We use a branch of statistics called Extreme Value Theory (EVT), since we are modeling the extreme performances of athletes (fastest times in the world). We took the times of the 100 fastest performers in each year, in each event since 1990, and used the EVT to determine its probability distribution. There is a very powerful mathematical theorem called Generalized Pareto Distribution (GPD) that shows that this distribution is always the same.

Analyzing the data

We can project the shape and the parameters of the distribution for the next few years, considering the evolution of the events and, of course, considering an even greater evolution in an Olympic year. Now we can simulate from this distribution a possible scenario for the top 100 fastest performers in 2024 and take the fastest time — that would be one single prediction for the fastest time in the world in 2024, which usually corresponds to the Olympic winning time. By simulation, let’s say, ten thousand scenarios, we can compute the average of all number ones in each scenario. That would be the final estimate for the fastest time in the world this year.

By doing this, we can also compute an interval for the fastest time in the world in 2024, let’s say, with 95% of confidence. Also, among all scenarios, we can count on how many of them the world record is broken, and so we can estimate the probability of it.

The results are in the following table.

Women’s events

Men’s events

As the predictions are based on the evolution of swimmers in recent years, it is natural that they are at times corresponding to what we have had in recent seasons. For example, it has been a few years since any swimmer has threatened the women’s 100 breaststroke world record – since Lilly King set the world record in 1:04.13 in 2017, the fastest time in the world has been half of a second slower, so the world record is not expected to be broken in 2024.

On the other hand, last year Ahmed Hafnaoui and Bobby Finke came very close to the men’s 1500 freestyle world record. Therefore, there is a reasonably high probability that long-standing Sun Yang’s world record from 2012 will finally be broken.

The results also reflect what we saw in 2023 in terms of world records. There were far more world records set in women’s events than in men’s events. Accordingly, we can see that the chances of world records being broken, in general, are greater among women.

It is interesting to observe the intervals for the times, but will they reflect the real results in Paris?