How to Make the London Marathon Ballot Fairer

The ballot for the 2026 London Marathon Ballot has now closed with a record 1,133,813 people entering. This is a massive increase of previous years but, with the number of spaces available not increasing at the same time, the odds of getting in are getting smaller. The 2025 race had a record number of finishers at 56,640 and I have seen some articles optimistically claim this means you have a 5% or 1 in 20 chance of being successful.

However, once you remove the Championship athletes (1,000), Good for Age (GFA) (6,000) and all the charity and club spots, it is generally felt that there are somewhere between 15,000-20,000 spaces left for the ballot. For the purpose of this article I’ll make the assumption that there are 17,000 spaces available, so the odds of being picked this year are just 1.5% or 1 in 66, not the greatest odds!

Is the Ballot Fair?

Every year when the ballot results come out you’ll see a deluge of comments online claiming the ballot isn’t fair. “I’ve been trying for years without getting in and my neighbour got in first time”, “My friend got in two years in a row”. Personally I was lucky on my 13th ballot attempt (ironically!) yet my brother was successful on his one and only ballot attempt, is this fair?

As people we are notoriously bad at judging odds and probability. A lot of bookmakers would find themselves broke very quickly if we were! As such we look for patterns in any data and often draw the wrong conclusions.

In reality, it is most likely that the highest share of people in the ballot are doing so for the first time. Versus the prior year there has been an increase of 293,495 in the ballot, the majority of whom will be applying for the first time. On top of this, you’ll have many people drunk on the joy of watching the marathon on TV (or maybe something stronger) who throw their hat in the ring.

On this basis it would be fair to assume that about a third of the people in the ballot are applying for the first time (or at least the first time in some time) with a steady decline in numbers applying for the 2nd, 3rd, 4th time etc. To apply some maths to the argument I have assumed that for every extra year people have waited to be successful then the total number drops by around 1/3. This is all assumptions, but helps to understand and address some of the regular complaints.

“Loads of people get in first time” – If we assume that the ballot is completely fair and every person entering has the same chance of being selected, then using the numbers above, this has to follow as being true. If 1/3 of the people applying are doing so for the first time then 1/3 of people being selected would be in this group. If there are 17,000 spaces available then that would be over 5,600 getting in first time. They are just that lucky.

In fact in my model 80% of the people successful in the ballot would have been on their at most 4th attempt, so it will seem they are lucky. On the flip side however it follows that the vast majority of people who are unsuccessful are in this cohort, however they are less likely to be shouting about it being unfair than those on multiple failures.

“I’ve applied x many times and not got in” – As mentioned before, the current probability of being successful is about 1.5%, or inverted, the chances of being unsuccessful are 98.5%. The chances of being unsuccessful 2 years on the trot is calculated as 98.5% x 98.5% which is still a probability of being unsuccessful twice in a row of 97%. Using this simple logic, after 10 years trying it is still 86% likely that you will have never been successful.

In fact, at the moment, it would take 46 years for there to be a higher probability that you would have been successful at least once than not at all. Obviously the odds change every year with the numbers applying but if recent trends are anything to go by, then it’s only going to get harder, not easier.

“They got in two years on the trot” – One of the biggest views of a lack of fairness is when some people seemingly get in multiple years on the bounce. I am sure some of this is skewed by perception as I would imagine quite a few who do multiple years in a row are either taking charity spots or qualifying GFA etc. However, even if every finisher of this years race had to go through the ballot for next years race, of the 56k finishers, 1.5% would be expected to be successful again.

This is a not insignificant 850 runners, enough to possibly know someone and be irked by their luck. For comparison being lucky two years in a row on the ballot is less likely than getting 4 balls on the lottery, but still much more likely than getting 5.

Before any mathematicians start shouting at me, the above examples have been simplified for ease. There is a world in which all 17,000 are first timers, or returners. Likewise there are worlds where only people on their 10+ attempt are successful in a truly fair ballot. However at these volumes you would expect to tend to the average and that helps paint the picture.

Could the Ballot be Fairer?

Whilst the above starts to delve in to maths behind probabilities, hopefully it is quite clear that even in a perfectly fair ballot, perceived unfairness can still be seen. So is the current ballot the right way of doing things and could there be an alternate?

Several years ago the ballot for the London Marathon had the benefit that after 5 unsuccessful attempts then you would be guaranteed a spot the following year. This was a great carrot for runners and made the whole thing seem “fair”. However, at that time it used to be the case that the ballot was limited itself so you had to be on the button to even get your application in before it closed.

This was great for eager beaver runners but much harder to bring more casual people who may not be aware of the open/close dates. Cynically it also limited the financial potential of the marathon so they moved to a time limited ballot that is unlimited in terms of how many can apply. We also today see so many more people applying for the ballot than in the past.

I searched the same figures for 10 years ago and that year there was a record 247,069 applicants. So in 10 years the numbers have increased more than 4.5x. Overall this means that in my model there would be 220,000 runners unsuccessful on their 5+ attempt, so giving them all a guaranteed spot would make the race massively over-subscribed. As attractive as this seems, realistically it’s a non-starter.

Another suggestion you sometimes see banded around is to ban anyone who ran the prior year. However, as I showed earlier, even if every finisher went back through the ballot, this would only account for around 850 extra spaces. In reality a lot of people are one and done so don’t reapply or they are getting in via other means so the true number of spaces unlocked is likely to be much lower than this. Besides, the majority would still be expected to go to first timers using the numbers shown. So, whilst it might make someone feel a little better, it won’t really change the odds of the person on their tenth attempt.

However, there is one change I believe the organisers could make to make it “fairer” to those trying year in, year out. It’s not a particularly complex or new idea either. I fact when I was getting ChatGPT to help me write the code for my model, I found it even has a name: weighted lottery with non-replacement.

In this model, you get an extra ballot entry for each year you are unsuccessful. Therefore if it is your first attempt, or you ran the previous year, you get one ticket. If you entered last year and were unsuccessful then you get two etc. The non-replacement but comes from the fact that if one of your tickets is picked and you are in, the rest are in effect made null and void.

The maths behind calculating your new odds with this are much harder so I instead built a model to simulate running the ballot multiple times using the spread of entries described earlier (ie 1/3 of the entries had 1 ticket, 2/9 had two tickets etc.). Ideally I would have run more times but my laptop is old and slow but already a picture starts to evolve.

Because of the vast numbers applying, the majority of those selected are still on their first or second ballot so there is no concern that it would completely alienate those having a punt. That said, the total share of those selected being from the first timers group would drop from around 33% to 11%. Maybe more interesting is the implied probability of being successful.

If you are in year one then, on average, the chance drops to just 0.5% but by year 3 the odds are the same as today and for every future year in the ballot, the odds start to improve. The chart does show the range when you run this as a real example rather than purely maths as the chances in the 20th year varied from 2%-16% across the various runs. Likewise, the strange “bump” after 17 years is due to the few times I ran the model rather than this being a sweet spot!

By the time you are at 10 years of waiting then then chance of getting in that year has increased all the way up to 8.8%. That said, the odds would still be against you, just less so than today. It would mean that the odds of being unsuccessful year after year would drop below 50% after 16 years, rather the current 46 years.

Will Anything Change?

Whilst it is enjoyable to think about what could change, in reality there is no reason for the organisers to change. When you are as over subscribed as the London Marathon currently is, does it really matter that you have several hundred disgruntled runners comparing their (lack of) success rates? They will undoubtedly be measuring success in the total number of applicants and in fact changing it to make it more favourable to those with multiple entries may discourage some people for giving the ballot a go.

Therefore I fully expect that come July when results come out, I will yet again receive the “sorry” email rather than the “you’re in” one. And if I am successful, I’ll be one of those that others hate for getting the golden ticket twice in 5 years. The ballot is never going to please everyone (in fact most will be disappointed) and I am just grateful that I was able to beat the odds to race in 2021.