How to Run a 2020 BQ Time at the 2018 New York City Marathon
An analysis of BQ times and pacing strategies in the New York City marathon.
In this post we use results from the last 11 years of the New York City Marathon (2006–2017, inclusive) — 488, 722 runners, including 303,394 males and 185,328 females — to explore how often participants achieve Boston Marathon qualification (BQ) standards, answering the following questions as we go:
- How often do male and female runners of different ages achieve their BQ times?
- When they achieve their BQs, how much margin do they typically enjoy and is this influenced by age?
- When runners fail to achieve their BQ time, how much do they miss by, and is this influenced by their age?
- How do the pacing patterns of those who achieve their BQ times differ from those who miss theirs?
- Are some pacing patterns (e.g. positive vs negative splits) more commonly associated with BQ times than others?
At the end of this article we will provide specific pacing advice, for male and female runners, across the different age groups, as to the pacing targets they should aim for in order to achieve their BQ time. This advice will be based on the typical pacing patterns of male and female runners who sucessfully achieve their BQ times in a given age-group on the New York course.
The New Boston Marathon Qualification Standard
With the 2018 New York City Marathon just around the corner many participants will be giving careful consideration to their desired goal-time, and their pacing strategy, as they aim to get the most from New York’s challenging course. Most participants will have their own personal goal-times in mind, and many may hope to achieve new personal-bests. Some may even hope to secure a much sought after qualification time for the Boston Marathon.
The Boston Qualification (BQ) time is one of the marathon’s best known, and most desirable, ‘good for age’ standards. It is based on the age and gender of runners. The chart below shows the new qualification standards (as of September, 2018) for the 2020 Boston Marathon, which must be achieved by runners on or after Saturday, September 15, 2018. These qualifying times are based upon each athlete’s age on April 20, 2020, the date of the 2020 Boston Marathon. The new standards lower the qualification times by 5 minutes relative to the previous standards.
Unfortunately, achieving one’s qualifying time does not guarantee entry into the Boston event, but simply offers the opportunity to submit for registration. In recent years, not all qualifiers who submit an entry have been accepted due to field size restrictions, and in such circumstances those who are the fastest among the pool of applicants in their age and gender group have been accepted; in this sense, rather than refer to BQ times, we should really refer to maybe BQ times or mBQs
While the above are the new BQ times, in the analysis that follows we will be analysing BQ rates over the past 11 years at the New York City Marathon, and we will do so using the previous BQ standards (5 minutes slower for each age group). The reason for this is that these would have been the goal-times targeted by runners who wished to achieve a (maybe) BQ status in the past. It would not be reasonable to analyse historial races against new targets. Of course, later, when we provide pacing tables to help runners achieve a BQ, we will do so by using the new BQ targets.
BQ Rates in New York
We define the BQ rate to be the proportion of runners in a given age group who achieve the BQ time associated with that group. In Figure 1(a) below we see the BQ rates for male and female runners, based on their individual ages. There is a small difference between the BQ rates of men and women, favouring women. One explanation for this might be found in recent research which suggests women are better at pacing themselves during the marathon than men; of course another explanation might be that the BQ standard for women is slightly easier than it is for men.
BQ rates also tend to vary with age, with a gradual increase across the age groups. For example, about 6% of runners in their 30s secure a BQ, compared with more than 10% of runners in their 60s. Does this suggest that BQ qualification is tilted in favour of older runners or are we witnessing the benefit of experience, in the sense that older runners will, all other things being equal, tend to be more experienced runners?
Notice the distinctive ‘saw-tooth’ pattern of the BQ rates in Figure 1(a), especially among older runners (>35 years old). This is indicative of systematically changing BQ rates within each 5-year age group. For example, about 11-15% of 45 year-olds achieve their BQ time, but over the next 4 years this rate falls to about 8%, for 49 year olds. It’s a similar pattern for other age groups. As runners age, qualification rates fall, and the difference between their group’s maximum and minimum BQ rates tends to increase.
As a general rule (save for the unusually large 18–34 group) the best BQ rates are associated with the youngest runners in a group, and the worst BQ rates are associated with the group’s oldest runners. In fact, the difference between the best and worst BQ rates tells us about how increasing age tend to impact performance. In Figure 1(b) we show the ratio of the best-worst BQ rates for each age-group. For instance, for 55–59 year-old male and female runners, the BQ ratio of 1.5 and 1.4 respectively. This means that in this age-group, the best BQ rates (for 55 year-olds) are better than the worst BQ rates (for 59 year-olds) by a factor of 1.5 and 1.4 for men and women, respectively. In other words, each additional year reduces the BQ rate by about 10% depending on the age-group. For ‘younger’ runners in the 35–39 and 40–44 age-groups the effect is less significant; the effect is exagerated by for the 18–34 year-old age-group because of it’s much longer age-span (17 years) and probably also because of the influence of elites.
The BQ Margin
Focusing just on whether a runner achieves their BQ time is a rather binary matter; they either do or they don’t. But two runners might achieve the BQ standard, one with minutes to spare, while another might just barely make it. Likewise, when a runner fails to achieve their BQ time, how much they miss by can be informative. We call this their BQ margin, which is simply the number of minutes by which they come in above or below their BQ time, as a fraction of this BQ time. For example, the old qualification standard for a 50–54 year-old women was 240 minutes. If such a runner achieved a time of 216 minutes then they would have come in 24 minutes below their BQ time and their BQ margin would be 0.1 (24/240); they achieved their BQ time with a margin of roughly 10%. If instead their finish-time was 252 minutes, then they would have missed their qualficiation time by 12 minutes, and their BQ margin would be 0.05, indicating they missed by 5%.
Figures 2(a) and (b) show the BQ margins for runners who (a) achieve and (b) miss their BQ times. In Figure 2(a) we can see that when runners achieve their BQ times, whether male or female, they tend to do so with a margin of 0.05–0.1, falling slightly as runners enter their 30s and 40s, and increasing again in their 50s and beyond. In other words, in New York, not only do a greater proportion of older runners tend to achieve their BQ times, they tend to do so with more a margin.
In Figure 2(b) we see the corresponding margins for those runners who fail to secure their BQ times. This time the margin tends to decrease steadily with age. In other words, older runners tend to miss their BQ standard by an ever-decreasing margin, getting closer and closer to their qualification times. For example, unsuccessful 25 year-old males have a BQ margin of 0.42, meaning that, on average, they miss their 3 hour and 10 minutes standard by about 42% (roughly 80 minutes). In contrast, a 60 year-old male runner in New York misses their 3 hour and 55 minute BQ time by just under 30%, or just over 60 minutes. It’s a similar story for women except that they enjoy an even narrower BQ margin than their male counterparts.
One way to interpret this is that runners who succesfully achieve their BQ times do so in a similar manner, whether male or female, but for runners who fail to achieve their BQ times, men tend to miss by a larger margin than women.
One explanation for this difference between men and women is the better pacing of women acopmare with men. In the extreme this pacing difference translates into a well-documented higher rate of hitting the wall among men, compared to women; runners who hit the wall rarely, if ever, achieve their BQ times and because many more men hit the wall than women we can expect their times to suffer accordingly. It is worth noting, but not shown, that even when we account for this — by excluding all runners who hit the wall from the analysis — women still continue to enjoy superior BQ margins to men, when they fail to meet their BQ standards.
It is also worth remarking how in Figure 2(b) we can see a similar saw-tooth pattern as the one we saw for BQ rates, back in Figure 1(a). Not surprisingly, as runners age within their age group, those who miss their BQ target tend to miss by a greater degree than younger runners within the same age group. A similar pattern is less clear, but still present, in Figure 2(a), for those who hit their BQ times.
Negative/Positive Splits and BQ Rates
How should we pace our race in order to achieve a BQ time? Is a runner more likely to achieve a BQ time by running a positive split (a faster first half than second) or a negative split (a faster second half than first), or does it matter?
To answer this question we split our New York runners, based on their halfway times, into those who run a positive split and those who run a negative split. In Figure 3(a) we can see how negative splits are fairly rare. Only 10–12% of younger runners run a negative split, and this rate falls gradually with age.
What is interesting about New York is that there is little difference in BQ Rates for those running positive or negative splits (Figure 3(b)). In earlier analyses of Berlin and Chicago, we found that runners who ran negative splits were consistently and significantly more likely to BQ than those running positive splits (across all age groups). But in New York this pattern is far less clear, and if anything seems to slightly favour positive splits (for runners between the ages of 35 and 55). This may be due to differences in the topology of the various courses. Berlin and Chicago are flat and fast whereas New York is more challenging, from an elevation perspective, and includes some tough climbs in the second-half, perhaps changing the positive/negative split dynamic.
Pacing Strategy and BQ Rates
In our New York dataset we have access to the timings of runners at 5km (approx. 3.1 mile) intervals. This means we can calculate their pacing across each of the 5km (3.1mi) segments from the start to the 24.9 mile mark, and from the 24.9 mile mark to the finish-line. Moreover, for each runner we can turn their actual segment paces (e.g. 8 mins/mile) into relative paces, by dividing their actual pace for a segment by their average pace over the entire race. For example, if a runner runs the first segment at 7 mins/mile and their average pace is 8 mins/mile then their relative pace for the first segment is 0.82, indicating that they ran it 12% faster than their average pace. We can use these relative segment paces as a pacing profile, to analyse whether they ran a particular part of the race faster or slower than their average pace.
Figure 4 shows the pacing profiles of runners for New York — (a) all runners, (b) males, (c) females — separating those who achieve their BQ times (BQ Success) from those who do not (BQ Failure). Each individual line corrsponds to a different age-group with the solid lines corresponding to the 18–34 year olds and the dashed lines to 65–70 year old runners; the paler, thinner lines correspond to each intermediate age group.
The main message to draw from these graphs is that the pacing profiles of those who achieve their BQ times are less varied than those who do not. For example, in Figure 3(a), runners who achieve their BQ times tend to start the race 2–3% faster than their average pace (relative pace of approximately 0.97–0.98) and they tend to finish about 8% slower (in the penultimate 3.1m segment, before the final ‘sprint’ finish). By comparison, runners who do not make their BQ times tend to start 5–8% faster but finish about 12% slower, again in their final 3.1 mile segment, before the ‘sprint’ finish. It’s a similar pattern for men and women, as shown in Figures 3(b & c). Curiously, non-BQ’ers tend to manage a faster sprint finish than those who BQ, suggesting that they have more energy to burn as they crossed the line, and perhaps indicating that they did not pace themselves to ‘give it their all’ on the course.
Pacing a BQ Attempt in New York?
How might all of this help someone to pace a BQ attempt in New York? Is there a set of recommended paces to run a successful BQ that is tailored for New York? In fact this is quite straightforward to determine, by calculating the average segment pacing of those who have (narrowly) achieved their BQ times in New York. In doing this we can produce a set of paces that reflect how runners BQ on the New York course, as opposed to using a simple one-size-fits all pacing model.
In the tables that follow we present these paces for men and women, across the different age groups, for positive and negative splits. Importantly, we calculate these pacing tables based on the new 2020 BQ standards with reference to historical runners who achieved these new standards on the New York course.
As mentioned above, the New York race results are timed every 5 kms (3.1 miles) and so these pacing charts reflect pacing for each 5 km (or 3.1 mile) segment. In the tables below, we are use minutes per mile pacing, and the segments are labelled using miles; the corresponding tables for kms and mins/km pacing are included as an appendix at the end of this post.
To use these pacing tables simply find the table and row that corersponds your gender, preferred split type (positive or negative), and age group, and note the paces for each of the race segments. For example, for a 45–49 year old male, aiming to to make their BQ time (3 hours and 20 mins or 200 ) with a positive split, then use the 45–49 row in Table 1. This recommends running the first 3.1 miles at 7m 16 per mile, then speeding up marginally over the next two segments before gradually slowing from the 9.3 mile mark to 7m 19s and then continuing to slow to finish in 8m 9s, which will get the runner over the finish-line in just under 3 hours and 20 minutes. For the same runner, if they wish to instead run a negative split then the recommended paces are in the corresponding row in Table 3, with female positive and negative splits in Tables 2 and 4.
One minor point worth making, for the sake of accuracy, is that in computing these average paces we have focused on those runners who narrowly achieve their (2020) BQ times, within a 1% margin. This ensures that the paces match a BQ time that is just within the qualification standard. Had we included all BQ qualfiiers then we would have produced pacing tables with more ambitious times, which are probably less useful, and certainly more risky, for most runners. It is also why finish times tend to be a minute or so faster than strictly required; so there is a safety marging in-built to these pacing tables. In any event, if a runner wishes to target a more ambitious BQ time then they can always borrow the pacing from a younger age group.
Running a BQ is far from straightforward and most don’t manage it. Those who do, achieve it by pacing themselves carefully — without starting too quickly to avoid finishing too slow.
The BQ pacing tables above provide concrete pacing recommendations to achieve a BQ time that is tailored for New York. If they help you please let me know!
Once this year’s New York marathon finishes it should be possible to compare those targeting the new BQ standards to previous years under the rule of the old standards. The new standards are objectively tougher. What will this mean for BQ rates and margins?
Appendix — Kilometer Pacing
The following tables have been converted to metic, using kms and mins/kms pacing. They can be used in the same way, and target the same BQ times, as the pacing tables above.