Update: Now that BAA has released the number of applicants, I’ve published a brief follow up here.
Last week, I shared a prediction for the 2025 Boston Marathon cut-off time – 7:03. If you haven’t read that article, you may want to read that first to get some context.
Since then, I’ve gotten some feedback about the analysis and had some deep discussions about the data. This has helped me look at the data in a few different ways and consider some additional factors that could influence the final outcome.
I still think the methods are sound and that the cutoff will be in the same ballpark. But I’ve identified three factors that could either increase or decrease the actual cutoff time:
- The Berlin Marathon
- The Easter effect
- The distribution of qualifying times
I wrote up a thorough analysis of each of these factors and published it in Runner’s Life on Medium. If you’re not a Medium subscriber, you can use this link to get a special link to bypass the paywall for this article.
Or, you can read the short version below.
The Impact of the Berlin Marathon Results
When I originally analyzed the data, I had not collected the results of the 2022 and 2023 Berlin Marathon. They weren’t readily accessible in the sources I was using, and a relatively small percentage of qualifiers from Berlin actually follow through and run Boston.
But, Berlin is nonetheless one of the top five races that people use to qualify for Boston. The percentage is small, but the race is so large that it still yields a significant number of qualified applicants.
So I managed to source the results directly from the Berlin Marathon website and add them into my dataset.
What changed between the 2022 and 2023 Berlin Marathons?
The number of finishers increased by more than 20% and the number of Boston qualifiers increased by almost 30%. Once you account for the percentage of qualifiers that actually apply to Boston, there will likely be a net increase of several hundred applicants out of Berlin.
Individually, this isn’t a game changer. But it’s an additional datapoint that points towards: a) more finishers, b) more fast finishers, and c) a deeper cutoff time.
The Impact of Easter on the Number of Applicants
Something else I had not previously considered was that Easter is the day before the 2025 Boston Marathon. There are some people who have argued that this will have a large effect on the actual number of applicants.
Looking to the data, there are three recent years that could help us understand what might happen: 2022, 2017, and 2014.
However, 2022 and 2014 are outliers that don’t offer useful data. In 2022, the number of potential qualifiers and applicants was severely reduced because of the impact of COVID. On the flip side, participation in the 2014 Boston Marathon was at record levels because it was the year after the terrorist attacks in 2013. It’s impossible to disentangle to specific circumstances of those years from the general circumstance of the race happening the day after Easter.
Setting those two years aside, 2017 is the only “normal” race in recent memory where the race took place on Easter Monday. The number of applicants that year was lower than in both 2016 and 2018, despite a general upward trend.
If you assume that the entire decline in applicants that year was due to Easter, then the number of applicants that year was about 91-92% of what would otherwise have been expected. This of course ignores the potential impact of other explanations – like a reduction in the number of qualifiers from poor race conditions in the 2017 qualifying period (it was warm at the 2016 Boston Marathon).
To quantify the impact of the greatest possible Easter effect, I took the original model, reduced the number of qualifiers to 91.5% of what was measured in the sample, and graphed it along with the actual number of qualifiers in the 2024 and 2025 qualifying periods.
That lowered the curve to almost the same point as the 2024 qualifying period. It’s still slightly higher, and the predicted cut-off time in this scenario is roughly 5:30 to 5:45.
On the one hand, there is a distinct possibility that Easter could discourage some qualifiers from applying. If that happens, it would reduce the cutoff time.
However, this demonstrates that best case scenario (for those hoping for a lower cutoff time) is that this negates the dramatic increase in the number of qualifiers throughout the qualifying period and leads to a cutoff time that is similar to what it was last year.
There is nothing in the data to suggest that the effect would be great enough to reduce the cutoff time to below 5 minutes.
The Impact of the Distribution of Actual Qualifying Times
A third potential factor is the actual distribution of qualifying times – and buffers. For two reasons.
First, consider what would happen if all of the new qualifiers had a qualifying time of 1 minute under their BQ – verse what would happen if all of the new qualifiers had a qualifying time of 20 minutes under their BQ.
To calculate the cutoff, you essentially move from fast times towards slow times until you fill up the field. So if you added 20,000 applicants at BQ minus 20 – the buffer would be down around 20 minutes and we’d all be out of luck. But if you added 20,000 applicants at BQ – 1, then it wouldn’t really matter. You could add 100,000 applicants at BQ – 1, and if the field is full at BQ – 5:29, then it would have no influence at all on the cut off time.
I was curious to see if the increase in the number of qualifiers was spread evenly throughout the field – or it was larger in some areas than others.
The visual above shows the number of qualifiers graphed according to their buffer – from 0-1 minutes (the far left) to 19-20 minutes (the far right). The blue bars are the 2024 qualifying period and the purple bars are the 2025 qualifying period.
Although there’s an increase across the board, the increase is much larger from 5 to 12 minutes than it is elsewhere. There was a greater relative increase among the runners who were slightly faster than last year’s cut-off time – which would indicate a likely increase in that cut-off time.
Second, the data released by BAA when they announced last year’s cutoff time suggests that faster runners are less likely to apply to Boston than slower runners. When I compared the number of applicants reported by BAA to the number of qualifiers in my sample, the conversion rate was highest among runners with a buffer of 0 to 10 minutes, lower for runners with a buffer of 10 to 20 minutes, and lower still for runners with a buffer of 20 minutes or more.
I combined that data with the number of qualifiers in my sample to estimate the number of applicants across the range of possible BQ buffers.
The bar labeled 21 represents all runners with a 20 minute or greater buffer. The rest of the bars represent 1 minute (19-20, 18-19, down to 0-1).
If you assume the field remains the same size – 22,000 spots for time qualifiers – it will fill up near the 7 to 8 minute mark. Notice that there’s a significant increase in the number of applicants with a 20 minute buffer, as well as relatively large increases at 10, 9, 8 ,7, and 6 minutes.
The projected number of applicants at last year’s cut-off time – 5:29 – is 24,687. Which again points to the fact that the cut-off time will end up being significantly lower than 5:29.
Finally, just note that the group of qualifiers with 20+ minute buffers includes all age groups. So while there are a significant number of young men running 2:40 or less, the majority of these runners are 35-59, and the largest increase (more than a third) is among runners 35-39. It’s split fairly evenly between men and women.
So What Does All This Mean?
Taken as a whole, this evidence suggests – pretty strongly – that the cutoff time will be at least 5:30. And it will quite likely be in the 7 to 8 minute range.
The original prediction was based on a large quantity of data – ~500,000 individual results across more than 250 races per qualifying period. Both the Berlin data and the distribution of qualifiers’ buffers offer additional evidence to support a conclusion that the cut-off time will be steeper than last year.
The Easter effect does offer a potential alleviating effect. But it is very unlikely that the effect is strong enough to reduce the cut-off time below last years.
I haven’t yet collected the results from this week’s last chance races, but a quick peak at them points towards the same trend – more finishers and more qualifiers. There aren’t enough finishers to have a huge impact one way or the other, but it’s more evidence in favor of the original prediction.
So … how are you feeling about your prospects?
For the record, I just registered and submitted my 3:08:31 – with a qualifying time of 3:10:00 (M40). I don’t expect to get in, but if there’s some kind of miracle I’m not passing up my chance to run Boston.
Very good analysis
I have a question, when will the cutoff time be revealed?
Typically they’ll announce the number of applicants at the close of the registration period (Friday). Then, they’ll take two weeks or so to verify the qualifying times of those applicants and calculate the cutoff.
Based off last year (registration ends 9/15, cutoff announced 9/28), they’ll probably announce it by 9/27, maybe a day or two earlier if verification goes smoothly.
Brian,
Thanks for your analysis. I think your model is probably the closest of all the ones currently circulating, primarily because you make an attempt to account for these factors.
I do think that there is one significant factor that has not yet been appropriately factored in yet.
And that is the increasing commercialization of the “Six Star” program and exploding international interest in running Boston, which can’t be fully captured by sampling only Berlin and a handful of the other big international races
Your dataset is huge, but I think it’s feasible that a surge from outside of the U.S./Canada/top international races could add, I don’t know, another minute or two to the numbers. I don’t know how anyone can model it appropriately, but I’m convinced that it’s real based on the continud proliferation of comments I see from international runners about running Boston and the growing global worldwide running phenomenon in general
For that reason, my estimate has been more on the order of 9-10 minutes as opposed to 7-8 minutes (and I certainly don’t buy that the cutoff has any chance of being near 5:29 again for all the reasons you laid out)
Another hint about a more significant cutoff than predicted is the change in Chicago corral times. Huge changes to get in the faster corrals. It seems average finish times at the big races are significantly faster in some cases. In terms of qualifiers your chart of increase in applicants projected by each minute in “buffer” is probably the most useful graph I’ve seen, but I wonder if it understates the case on a global basis when I look at the Chicago corrals
Any thoughts on this? Thanks again
I think it’s a good observation. The number of international runners was up last year, compared to previous years, and if it’s up again this year it’ll be a strong indication that there is greater demand from overseas runners.
If that is the case, it would be possible to pull in some additional large overseas races. Comparing the entry list of the 2025 Boston Marathon to qualifying times at some of those marathons will help quantify how likely it is those runners will want to come to Boston. I think there’s a danger in just including a ton of overseas races, but if you can weight them appropriately it makes sense.
It is definitely a potential source of additional applicants, though. Globally, there are likely ~2,000,000 finishers per year. The North American-centric sample I’m working with is only a quarter of that. The other 1.5 million aren’t as likely to run Boston, but even a small shift in the percentages could reshape the field significantly. Add in a few thousand additional overseas applicants, and you’re going to have a large impact on the cut-off time.
I haven’t looked into this but someone told me the weather at Boston this year was bad for running and there were less qualifiers. If true, is that factored in? I qualified with a 7:40 buffer at Chicago, so looks like I’m in the line with your predictions. Would suck If I miss it a second year in a row again by a minute or less. Btw, I noticed over 600 people in my age group also qualified. That feels like a lot..but I don’t know how many of those people apply.
Hi Carlos. Yes, the weather at Boston was warm this year, and there were significantly fewer runners who ran qualifying times there. Boston is also the number one race where runners qualify – so a lot of people think that means there are fewer qualifiers this year.
But across the board there are way more qualifiers at other races. The increase at other races – Chicago being the biggest of them – far exceeds the decrease in the number of qualifiers at Boston.
I talked more about it in this other article, but yes, that was part of the analysis.
Great analysis-my 7:23 buffer looking more shaky than I thought.
I think you analysis is just super. I have a 15 minute and 30 second cushion so I hope that will suffice. From what you write, I should be good to go…..?
Yup, you should be good to go. Good luck, and enjoy!
Brian
What do you think now that the applicant total has been revealed
Hi Andrew. You can read my latest article for some more in depth thoughts – but the reported number of applicants matches up quite closely with the prediction in my model. This makes me a lot more confident in the original prediction.
So I still think the cut-off will be right around 7, but the potential margin of error is now much lower. It could be slightly higher or lower, but I’d be surprised if it was outside the range of 6:30 to 7:30.