This is the second year in which I maintained the Boston Marathon Cutoff Time Tracker and the third year in which I did some analysis leading to a prediction. I wrote a reflection on my initial prediction here and on last year’s tracker here.
In each case, I was in the ballpark. But I wasn’t spot on. And, frankly, that’s to be expected. There are enough variables and unknown in this process that it’s really not possible to predict this down to the second with any kind of confidence.
But then, there was this year. The final projection on the cutoff time tracker was 5:13. This week, BAA announced the actual cutoff for the 2027 Boston Marathon: 5:17. That’s pretty darn close.
So I win, right? Not exactly. There were some lucky breaks that made this closer than it otherwise might have been, and after the number of applicants was announced I expected that it might wind up being a little lower (4:50 to 5:10). But for the third year in a row, this is absolutely proof that the general methodology is sound, and despite ongoing changes it can be expected to produce valid results.
Now that we’ve got some additional data, we can answer a few unanswered questions:
- Does the core methodology – predicting the cutoff based on the number of applicants – still work?
- Did the lottery significantly change the behavior of applicants?
- Did the downhill results index significantly reduce the number of applicants?
- Do re-qualifiers at Boston really apply to run the race again?
- Is the cutoff time likely to plateau or continue to rise?
Let’s get to it.
Does the Assumptions Underlying the Cutoff Tracker Still Hold?
In short, yes.
First, it did a good job of projecting the number of applicants. Set aside some of the other discussions around x-factors and untracked results (the lottery, Berlin, Tokyo, Sydney, etc), and the projection here was actually spot on.
Last year saw 33,249 applicants. The results I tracked showed a 1.69% increase in BQ’s over the previous qualifying period. That would yield 33,811 expected applicants. There were 33,649. That’s pretty damn close.
I had made the original projection based on an assumption that the number of accepted applicants would remain the same (24,300). Instead, it increased slightly – to 24,630. If we fed that into the original regression equation, we would have ended up with a projection of 4:59. Which would be an underestimate, but still within 20 seconds.
The chart above puts this in context. The blue dots are historical figures – based on the percent of applicants that were rejected in a given year and the actual cutoff in that year. The blue line is the linear regression equation based on that data.
The red dot shows the projected cutoff (5:13), given the known number of applicants and an estimated 24,362 acceptances. The green dot shows the actual outcome (24,630 acceptances and a 5:17 cutoff). Where the dotted line intersects the blue line is what the projection would have been if we knew the number of runners to be accepted.
So 2024 remains the biggest outlier. Every other year – even 2021, which was itself a strange year – fits the model relatively well.
Did the Lottery Significantly Change the Behavior of Applicants?
The short answer: no.
There was a plausible theory that the creation of a lottery for qualifiers who didn’t make the cutoff would make it more likely that runners with small buffers (under 1:00 or 2:00) would apply. Without the lottery, they had little hope of being accepted.
But if that were the case, then we would have expected to see a number of applicants that was inflated over previous years. Since the number of applicants wasn’t higher, the only other way this theory would hold is if the number of applicants yielded a significant over prediction. Instead, the cutoff time was 18 seconds higher than the model would have predicted.
For a more rigorous answer to the question, we can also look to see whether the conversion rate of the number of qualifiers in the tracker to the number of applicants changed.
Last year, there were 14,346 qualifiers in the tracker who had a buffer between 4:34 and 0. There were 8,887 applicants rejected. That yields an effective conversion rate of 61.9%.
This year, there were 15,147 qualifiers in the tracker who had a buffer between 5:17 and 0. There were 9,019 applicants who failed to make that cutoff. That yields an effective conversion rate of 59.5%.
Those two numbers are close enough that I’d wash it up to random variability. And unless there’s some confounding factor that is otherwise bringing that down, it’s safe to say that the lottery had no significant impact on the decision of qualifiers to apply.
One other lottery related question: were the 1,000 bibs net new bibs?
It turns out they were, or at least they weren’t taken from time qualifiers. It remains to be seen if the overall field grows by another 1,000 or if this is offset by other categories. Despite them saying that these bibs wouldn’t reduce the number of qualifiers accepted, I did kind of expect them to fudge that a little bit.
Now we know for sure.
Did the Downhill Results Index Reduce the Number of Applicants?
One of the other lingering questions throughout this entire qualifying period was just how much the downhill results index would change things.
Initially, I expected to see it nudge the cutoff time down slightly but not have a big impact. That was partially based on an expectation that REVEL Big Bear would return (which it didn’t). It also expected a similar number of runners to run these races. They didn’t.
A couple weeks ago, I took a look at the results and noted that there was a drop of ~1,800 net qualifiers from these downhill races. Now we know how many runners actually applied from those races: 240.
When I compared the results of the 2025 Boston Marathon to potential qualifying races, I estimated that ~2,000 of those runners made it in through a downhill race. If that number was accurate, then the drop to 240 is almost a 90% decrease. That’s huge.
Before we overstate the impact, it’s worth pointing out that many of the milder downhill races changed their courses to come in just under 1,500 feet of net drop. So there are surely some applicants who qualified at those races. But even so, it’s still clear that the number of runners applying from a downhill race is far less than it used to be. Exhibit A is the huge decline in numbers from REVEL Mt. Charleston.
Does this mean it effectively cut the number of applicants down? Not necessarily. My past analysis showed that a good chunk of those qualifiers could probably have qualified on a flatter course. It’s quite possible – likely, I’d say – that at least some of the runners who were aiming for Boston simply shifted their attempts to flat races.
But that still achieves the goal. Even if this didn’t necessarily reduce the number of applicants and the cutoff time in a huge way, it put (almost) everyone on level playing field.
If there was a true decline in applicants from 2,000 to 240, then that’s worth about a minute on the cutoff. If you assume some of those a) still qualified on a mild downhill course without a penalty or b) still qualified on a flat course, it’s much less.
Hard to say exactly what that impact is. But it’s safe to say that steep downhill races no longer factor significantly into the qualifying picture.
Do Boston Runners Come Back to Boston?
One of the significant events in this qualifying period was the 2026 Boston Marathon. There was excellent weather, including a sustained tailwind, and the results were much better than in previous years.
This resulted in a net increase of ~3,000 BQ’s year over year. This year’s race had a record 15,835 runners who met their qualifying time for the next Boston Marathon.
More qualifiers means more applicants. And that means a higher cutoff time. Right?
At the time, there were some naysayers who just didn’t believe that a significant number of runners repeated Boston. I had already done an analysis of the conversion rate of qualifiers from individual races, so I was pretty confident that a significant number of these extra BQ’s would convert into extra applicants. Past data suggests that ~30% of re-qualifiers at Boston run the following year’s race.
Last year, Boston 2025 was the largest qualifying race in the field, yielding 2,981 applicants.
This year, Boston 2026 was also the largest qualifying race in the field, yielding 4,448 applicants.
The number of qualifiers at Boston went from 12,801 (2025) to 15,835 (2026), an increase of 23.7%. The number of applicants from Boston increased 49.2%.
I suspect there are two things going on here.
First, the greater number of qualifiers at Boston 2026 led directly to a greater number of qualifiers. Simple and straightfoward.
But there’s probably an ancillary effect of the better conditions at Boston 2026. Some qualifiers notch more than one qualifying time in a year, and runners at Boston are probably more likely than most to do so. Boston reports these numbers based on a runner’s best qualifying time.
As a result, some runners a) qualify at Boston but b) are attributed to another race. In a typical year, a runner is going to be slower at Boston than at a flat course. But the conditions were so good this year that many runners likely did better at Boston, further inflating the reported number of runners earning acceptance through the Boston Marathon.
In any case, I think this settles the debate: there are many people who are happy to return to Boston year over year if they have the chance. It may be one and done for some people, but that’s hardly a universal opinion.
What Does the Future Hold?
Let’s wrap up with a little prognosticating.
Unlike last year, there are no known major changes to the qualifying and registration rules. There’s nothing like the downhill results index, the elimination of the double dipping period, or the creation of the lottery to shake things up.
There are, then, two key variables to watch.
- Does the number of runners continue to increase?
- Does the number of qualifiers increase at similar races?
On the first count, I’d say this is almost definitely going to happen. CIM announced a larger field and London is having a double header. Grandma’s sold out in a matter of hours, and many other races have also sold out in record time.
Simply put, there’s no reason to think the growth of the sport will reach a plateau in the next year. The number of finishers increased 10% last year. Another 10% increase would mean many more qualifiers.
On the second count, this could come down to weather. Boston will likely revert to the mean and produce fewer qualifiers. I haven’t done the analysis on Berlin yet, but the weather was better and I’m sure it also reverted towards the mean. Aside from those two races, though, there were quite a few large non-Majors that had warm weather last year. More, I suspect, than usual.
This year, the number of finishers grew by over ten percent but the number of qualifiers grew by less than two percent. Part of that is no doubt the impact of weather throughout the year. If the weather is even marginally better this year, it’s quite possible we see another 5% to 10% increase in qualifiers.
Here’s a worst case scenario. The number of qualifiers increases 10%. The number of applicants also increases 10% (~37,000). The number of acceptances remains the same (24,630). The cutoff would be +/- 6:30.
That’s not a prediction. It’s just a vision of a possible future. And throughout the year, we’ll be able to track how likely that vision is to become a reality.
I’ll be taking a break from regular tracking for a few weeks, but I anticipate launching a new 2028 Boston Marathon Cutoff Time Tracker in November. Likely sometime between Election Day and Thanksgiving. Stay tuned.