What Do the Number of Applicants for the 2026 Boston Marathon Mean for the Cutoff?

This morning, BAA announced the cutoff for the 2026 Boston Marathon: 4:34. Read a preliminary analysis of the data they released here.

This morning, the Boston Athletic Association released the number of qualified applicants for the 2026 Boston Marathon: 33,267. Which begs the immediate question – what does this mean for the cutoff time?

How Accurate Were the Projections?

In my final prediction, I predicted that the number of applicants would likely be between 34,000 and 35,000. I also stated that there was a broader range of possible outcomes – 33,000 to 36,000 – outside of which any other result was incredibly unlikely.

The final number – 33,267 – is below that tighter 34,000 to 35,000 range. But it is well within the broader range of 33,000 to 36,000. So I would rate this as: close but not perfect.

The original Boston Marathon Cutoff Time Tracker showed a 6.25% decline in the number of qualifiers in the sample. Applying this to last year’s applicant total results in a projection of 34,118 applicants. That’s 2.5% higher than the actual number of applicants. So again, close – but not perfect.

The second (experimental) dashboard I created to utilize conversion rate data to project the number of applicants projected 34,807 applicants. That’s 4.6% too high. A little less close, but still not wildly inaccurate. However, this does suggest that I might have gotten a little too cute with all the parameters. Sometimes simpler is better.

What Could Explain the Difference?

Given that the actual number of applicants was lower than the expected number of applicants, what are the possible explanations for the discrepancy?

First, it could just be random chance. A deviation of 2.5% is not large. Any given year, thousands of people make decisions about whether or not to apply, and it’s impossible to predict that outcome with extreme precision. If we could play this application period out 100 times, we’d probably get a range of applicants +/- 500 or 1,000.

But to the extent that there is something depressing the numbers, I’d suggest there are two possible reasons.

First, there’s been speculation that the current political climate will drive down international participation. If there would otherwise have been approximately 34,500 applicants, and 12,000 of them were from international runners, this could represent a 10% decline in international participation.

While the fact that the applicant pool remained so large indicates that there was definitely not a massive boycott – it’s possible that the numbers are down. We won’t know for certain until the registrant list is released in the spring, but until then we can chalk this up as a possible explanation.

Second, it’s possible that runners with a small buffer chose not to apply. In the last two years, about 10,000 applicants have had a buffer under 5:00. Throughout this qualifying period, it has been fairly well known that the cutoff time would likely be around 5:00. By the end of the qualifying period, I think the evidence was pretty clear that a 2:00 to 3:00 cutoff was all but impossible.

If 10% of those runners chose not to apply, that could also explain a large part of the decline. When the BAA releases the stats on the number of runners eliminated below that cutoff, we’ll have a better idea of how the distribution of this year’s applicant pool compares to last year.

Which explanation is correct will have an influence on the ultimate cutoff time. If the “missing” applicants are distributed throughout the full range of buffers (which should be true if it’s a general decline of international applicants), then the cutoff time will be relatively lower.

But if the “missing” applicants are clustered in the 0-3 minute range, that would decrease the number of applicants eliminated per minute of cutoff – and suggest a relatively higher cutoff time.

What’s the Cutoff Time Likely to Be?

The simple answer is that if the number of applicants was lower than expected, the cutoff time will also be lower than expected. Instead of a likely outcome of 5:30 to 6:00, we’re now looking at a likely outcome of 5:00 to 5:30 – with a high 4:XX being in the realm of possibility.

One way to look at this is to graph the number of rejected applicants in each of the last few years against the resulting cutoff time. See the visual below. Note that the y-axis represents the cutoff time in seconds. You can hover over a dot to get the specifics for that year.

There’s one significant outlier – 2021. That year saw the smallest field of accepted time qualifiers in a long time – and the highest cutoff time ever (7:47). Other than that year, there is a pretty good relationship between the number of rejected applicants and the cutoff time.

If there are 24,000 accepted applicants, there will be 9,267 rejected applicants this year. The dotted line marks that point on the graph – which intersects with the trend line at around 5:20. This trend line does overestimate the two most recent years (the dots to the right). So that 5:20 could be slightly overestimating things as well.

A similar way to look at this is to graph the percent of applicants that were rejected – instead of the actual number of rejected applicants. This eliminates a little bit of the messiness created by varying field sizes.

In this graph, the linear relationship between the percent of applicants that are rejected and the actual cutoff time is pretty tight. The 5:29 cutoff in 2024 shows the most deviation from the prediction. Otherwise, each year is within about 15 to 20 seconds.

Again, the dotted line represents the expected rejection rate for this year (27.85%) if the field size remains at 24,000. Here, the dotted line intersects the trend line at just over 5:00.

So if we base the projection on this relationship – between rejections and cutoff time – we get a projected cutoff time of around 5:00. Maybe slightly higher. Probably not any lower than 4:45.

What If the Field Size Isn’t 24,000?

This has been a common question all year, and everyone has their own pet theories about what will happen.

One guy argued that BAA would just expand the field size to include everyone who applied. I can pretty much say with certainty that that is not going to happen.

The most likely outcome is that they stick with last year’s field size – 24,000. Over the last decade, the typical sizes have been 23,000 to 24,000, with only a couple years at 22,000.

I’ve had people ask what would happen if they revert to 22,000. And I guess this is possible. But it would increase the cutoff time, and it seems like a silly move after last year’s increase.

I’ve also had people ask what would happen if they accepted more. Again, I guess this is possible. They surprised everyone by growing the overall field last year. But last year was the biggest field – outside 1996 and 2014 – ever. So unless they cut spots from charity runners and sponsors, I doubt they’re going to accept much more than 24,000.

But for the sake of argument, here’s the previous graph with the potential outcomes for each field size from 22,000 to 26,000.

What you’re looking for here is where the dotted line intersects the trend line. The bigger the number of expected runners, the lower the anticipated cutoff. If the field grew by 2,000, you’d be looking at about 4 minutes. If the field shrank by 2,000, you’d be looking at about 6:30.

The table below uses the linear regression to calculate the actual anticipated cutoff time at each place on the chart.

Again, I think the most likely outcome is that they stick with 24,000 accepted applicants. But it’s possible they veer one way or the other.

So How Does the Number of Applicants Impact the Prediction?

When I first saw the number of applicants, my initial take was that this would result in a slightly lower cutoff time – closer to 5:00 than to 6:00. After taking a closer look at the rejection data from the past decade, I think that’s a pretty solid prediction.

Assuming a) the field remains at around 24,000 accepted applicants and b) the distribution of buffers among the applicants remains similar to last year, than I’ll make an amended prediction of 5:16 +/- 20 seconds. So basically between 5:00 and 5:30.

If the number of accepted applicants varies in either direction, then the table above gives you an idea of what the likely outcome would be.