Last month, I wrote up a short analysis of both the number of women and the number of men finishing at specific fast times at the New York City Marathon.
The gist of it was this: There was a huge surge in the number of women finishing under 3:00 and men finishing under 2:40. At more modest, but still fast, times (3:30 for women and 3:00 for men), there was an increase. But the magnitude of the increase was much smaller.
One possible reason for this: the selection process for the 2024 New York City Marathon and, specifically, the way that non-NYRR time qualifiers were selected.
Previously, this was first come, first served. This year, the system prioritized the fastest runners in the group. This would have the effect of shifting the distribution of times to the left – and increasing the relative number of very fast runners within that group.
The flip side of things is: What happened at the other races this fall?
I’m currently working on an analysis of the Boston Marathon cutoff time for the 2026 Boston Marathon (expect that later this weekend), and for that project I’ve collected the results of about a hundred fall marathons in the United States and Canada. That dataset offers a good control against which to compare what happened at New York.
So let’s dive into the data and see what it shows.
What’s Included In the Dataset?
The sample used for this analysis includes marathons with more than 200 finishers in the US and Canada, between September 1 and the weekend before Thanksgiving. I’ve also included the Berlin Marathon.
In total, that includes 93 races for 2024 and 99 races for 2023. There were approximately 285,000 finishers at this set of races this year and 245,000 finishers last year.
I’ve also narrowed the field down to runners under the age of 40. These are the runners most likely to run the fastest times, and as you move up in age through the 40’s and into the 50’s, these fast times become increasingly rare. Focusing on the youngest runners offers a more stable view of the percentage of runners who finish at a given time and minimizes the effect of age as a confounding variable.
For the visuals, I’ve divided the finishers into three groups – the NYC Marathon, the Berlin Marathon, and all other marathons.
How Many Women Are Finishing at Fast Times?
Let’s start with the women’s side of things.
The initial analysis focused on the number of women finishing under 3 hours and under 3:30. I’ve expanded things a little bit to offer a broader picture, and the visual below denotes the number of women finishing under the following times – 2:50, 3:00, 3:10, 3:20, and 3:30.
These are cumulative, so everyone in the sub-2:50 bar is counted a second time in the sub-3:00 bar, and so on and so forth.
Hover over a particular bar to see the exact number for that benchmark time and the total number of finishers for that year and race.
When you look at Berlin, the number of women under 2:50 remains about the same, but there’s a large increase for each of the other four bars. There’s also an increase in the total number of finishers: 6,325 to 8,190.
The number of finishers increases by ~30% and the number of finishers under the benchmark times increases ~40-50%. So the two rise roughly in tandem. The general distribution is similar from one year to the next, although it shifts slightly to the left.
At NYC, there’s again an increase in the number of finishers – about 16%.
The increases in the number of fast finishers is much higher, but it levels out at the more modest times. It increases by ~350% at 2:50, 150% at 3:00, 70% at 3:10, and 35% at 3:20 and 30% at 3:30.
Then, at the bottom, you have the rest of the races.
The number of finishers increases by about 25%. But the number of women finishing sub-2:50 and sub-3:00 stays roughly the same. There’s a smaller increase at 3:10 (~15%) and then a ~25% increase at 3:20 and 3:30.
So the particular increase at New York City – especially at 2:50 to 3:10 – is definitely an outlier. The number of women finishing at those times increases far more than at the more modest times. And, despite the number of women at other races increasing at a typical rate at 3:20-3:30, there’s no increase at 2:50-3:00.
How Many Men Are Finishing At Fast Times?
And what about the men’s side?
Here, I’ve narrowed it down to four times – 2:30, 2:40, 2:50, and 3:00. Otherwise, the visual is set up in the same way.
At Berlin, there’s again a ~30% increase in the number of finishers. And when you look at the increase in the number of finishers at each benchmark, it’s in the 40-50% range.
So the increase in fast runners is slightly greater than the increase in all runners – but not by much, and the general proportions stay the same.
At NYC, there’s again a ~15% increase in the number of finishers.
But the increase among fast finishers is 150% for 2:30 and 2:40, 80% for 2:50, and 55% for 3:00. As with the women, the increase is much larger than the increase in finishers, and the increase is much larger at the faster end of the spectrum.
At the rest of the races, the number of men increases ~30% – so slightly more than the number of women. The increase at 2:50 and 3:00 is ~25%. And at 2:40 and 2:50 it’s closer to 15%.
Unlike with the women, there is an increase in finishers at the faster end. But it is smaller than the increase in finishers at the more modest end.
What Do These Numbers Look Like As a Percent?
Since the total number of finishers increased pretty significantly – across the board – it can make it a little harder to wrap your head around what the increases at each benchmark time mean.
Looking at it as a percent increase is one way to think about things. Another is to consider – what percent of the total field finished below a given benchmark?
So here is essentially the same visual as before – but instead of the number of finishers at each benchmark, the bar indicates the percent of finishers. So this standardizes the y-axis a little bit and makes it easier to compare across the chart.
At Berlin, the two charts look pretty much the same. There’s a slight increase at most of the bars, but they all go up in a similar way.
At New York, they all go up – but the left end grows a lot faster than the right.
The percentage at 2:50 goes from 0.14% to 0.54%. At 3:00, it goes from 0.58% to 1.24%. At the other end, the percent under 3:30 goes from 6.23% to 6.92%.
Then at the bottom, among the other races, the percent of women from 2:50 to 3:10 goes down a small amount, and it stays about the same at 3:20 and 3:30.
As an aside, it’s also interesting to note the extreme difference overall between Berlin and New York. About 3.5% of women at Berlin go sub-3:00, compared to 1.24% for this year’s outlier year at New York. At 3:30, the difference is ~15% to 7%. A good visual reminder of just how difficult the course is at New York City.
To round things out, here’s the same visual for the men.
At Berlin, there’s again a small increase across the board in the percent of runners hitting each benchmark time. Nothing out of the ordinary.
Meanwhile, at New York City, the percent of men goes from 0.30% to 0.68% at 2:30 and 0.98% to 2.15% at 2:40. There’s a more modest increase from 6.69% to 9.02% at 3:00.
And at the other races, there’s a minimal decrease at 2:30 and 2:40, and hardly any difference at 2:50 and 3:00.
So What Does It All Mean?
When you put this altogether, I think it supports the conclusion I came to in the previous analyses.
Something special is going on at New York – and that something special is likely related to the qualification process.
At Berlin, you have a slight trend towards faster times. Across the board, there’s a similar, small increase. This is the way you’d expect times to shift if there’s no special circumstances distorting the distribution.
At the other races, there’s an increase in the number of finishers – and a similar increase in the number of finishers meeting the benchmark times. But as a percent, they all stay roughly the same.
But at New York City, there’s a clear difference between what’s happening among the fastest runners – 2:50 to 3:10 for women and 2:30 to 2:40 for men. The drastic difference at the left end of the spectrum stands apart from the rest of the field.
Absent another explanation, I think the change in the process for evaluating qualifying times is the most logical answer. There could be other things at play pushing the field as a whole towards fast times – and this was a fast year at New York City.
But the qualifying times are a sound explanation for why the distribution is so distorted – and for why the number of runners at the fastest times increased so much compared to previous years.
If you have any other theories, I’d be interested in hearing them in the comments below. Moving forward, it’ll also be interesting to see if the further changes – excluding half marathons as an option for non-NYRR time qualifiers – has any further impact on this distribution in 2025.