What Do the Number of Registrants for the 2027 Boston Marathon Mean for the Cutoff?

Last week was the registration period for the 2027 Boston Marathon. I published this final projection for the cutoff time, anticipating a cutoff of 5:20 to 5:30. This was based on two different methodologies that yielded projections of 5:13 and 5:09, combined with an assumption that the actual outcome would err on the higher side rather than the lower side.

As they usually do, BAA provided an update the Monday following the registration week. There were 33,656 applicants. Previously, all of the projections were really based on predicting two things: a) how many applicants there would be and b) what cutoff time that would yield.

Now, we have one less variable and we can try to narrow things down. However, the addition of the lottery this year is a potential confounding factor that may still cloud the final picture a bit. When the dust settles, it’ll be worth doing some additional analysis to see just how much that did change things.

How Does This Compare with Previous Years?

Let’s put this number in perspective and see what it means for the cutoff, if we treat this year like previous years.

Over the past ten years, the number of applicants has increased dramatically. From 2015 to 2019, it increased from 25,000 to 30,000. That dropped in 2020 with the new qualifying times, stayed low throughout the post-COVID years, and then exploded in 2024.

In each of the last four years (2024-2027), there were in excess of 33,000 applicants. 2025 saw a record number (36,393). This year’s total (33,656) was the second highest and it was slightly higher than in 2026 (33,249).

The second important number for predicting the cutoff is the number of people who will be accepted. This is still unknown. Last year, it was 24,362, already higher than previous years. With the addition of 1,000 bibs for the lottery, I seriously doubt the number of accepted time qualifiers will increase further. It’ll likely be the same or slightly lower.

If we assume the number remains the same (24,362), then 9,656 applicants will be rejected. The number of applicants was only slightly lower than the projected number of applicants taken straight from the tracker (33,810), so the projection is about the same.

The regression equation, based on the percentage of applicants rejected, yields a projected cutoff time of 5:09.

Using that same equation, taking the number of accepted applicants down slightly to 24,000 would yield a cutoff time of 5:23.

Why Would It Be Higher Than Last Year?

One of the common themes I’ve seen in the discussions online is that a) the number of applicants is only slightly higher than last year, so b) the cutoff time should only be slightly higher. If last year’s cutoff time was 4:34, it should only go up a few seconds … right?

The flaw in that thinking is that it assumes that 4:34 was a typical outcome. But based on the data fed into the regression equation, it’s one of the lowest historical outcomes. See the visual below.

The diagonal line shows the predicted cutoff based on the percentage of applicants that are rejected. The current projection (5:09) is marked by the vertical line. Last year’s outcome (4:34) is marked by the horizontal line.

Of the ten years of historical data fed into this model, five of them have been higher than predicted and four of them have been lower than predicted. One was spot on.

Of the years that were lower than predicted, 2016 and 2020 were slightly lower. 2024 was far lower. And last year was about 25 seconds below the projection.

There’s inherent variability in any prediction and you’d expect future predictions to revert to the mean. So last year’s cutoff (4:34) isn’t the baseline. Last year’s prediction (4:59) is. And increasing the number of applicants slightly would then increase it slightly above 4:59.

What Could Be Different This Year?

This regression equation is based on historical data about the percentage of applicants rejected and the actual cutoff time. It assumes that there’s a relatively stable relationship between those two figures – and historically there has been. When you look at the graph above, 2024 is the only year that really deviates from the curve in any significant way.

But there are underlying datapoints that certainly influence the outcome. And if one or more of them is significantly different from previous years it could invalidate the entire model – or at least point in one direction or another.

One thing we know has changed is that the qualifiers are faster. When you look at the data from the tracker, there are more qualifiers with 5:00+ buffers and fewer qualifiers with 0:00 to 5:00 buffers. This points towards a higher cutoff time. When I combined this data with historical conversion rates for qualifiers, it suggested the field would reach capacity at 5:09.

There is one other potential change that could influence things in the opposite direction. It’s plausible that the conversion rate of low buffer qualifiers is higher than in previous years. The lottery gives someone with a 0:00 to 1:00 buffer a greater incentive to apply. By itself, that could inflate the number of applicants by a few hundred – up to a thousand or two.

If that’s the case, it would be fair to assume that the cutoff time will be lower than expected.

So What Does This Mean For the Cutoff?

Last year, I was pretty bullish on a high cutoff – higher than the projection. I got burned slightly with that one.

This year, I was still expecting things to come out on the high end. To be honest, I’m a little surprised the number of applicants was this low. I’d have expected at least 34,000. But a lingering unanswered question is how much impact the lottery had – and if it turns out it didn’t increase the conversion rate of low-buffer applicants this outcome is actually spot on with the projections.

But given the fact that the number of applicants didn’t come in higher than the projected number, I think it’s now unlikely that we end up with an outcome on the higher end of things. I think at this point, it’s safe to say that it’ll be closer to 5:00. I’d still be a little surprised if it’s (significantly) lower than 5:00, and if I had to peg it to a small range I’d say: 4:50 to 5:10.

This hinges on two unanswered questions.

How big of an impact did the lottery have? If the cutoff is significantly lower than 5:00, it’ll likely be because it turns out there was a huge impact from the lottery. We’ll be able to measure this after the announcement, assuming they release the breakdown of applicants by buffer as they usually do.

How many applicants are accepted? It’s plausible this number declines slightly. They could accept 24,000 applicants and still claim that the lottery spots were net new spots. It’s a small difference, but this would nudge the cutoff back up. It’s likely this remains the same or goes down slightly – which means there’s a chance of adding an additional 5 or 10 seconds to the cutoff.

Given the fact that it’s in the ballpark of 5:00 anyway, it’s also quite possible that BAA simply sets the field size at the point that they can keep the cutoff at 4:59 or lower. I think there’s a big psychological difference between a 4:XX cutoff and a 5:XX cutoff. And if the difference is a few hundred bibs, they might keep the field size high to prevent the cutoff from breaching the 5:00 mark – or let the field size wane slightly if the cutoff would still remain below 5:00.

Based on previous years, we can expect an announcement sometime next week. So stay tuned. And good luck.