Do race finish time predictors work ?

I’m a big believer in trusting in your training, put in the hard work and you’ll be rewarded. I have said it many times on this blog running is an honest sport, this of course is a two edged sword with very few examples of running faster than expected times off poor or no training. Of course returning from injury can sometimes be a blessing in disguise because you are forced to take some time off and this, combined with good training before and after the injury, can surprise you with better than expected times. Which leads me to the point of this post, predicted finishing times for longer races from shorter races.  The main example I use is my current half time times two and add 10 minutes, this  will normally be pretty close to my marathon time. e.g. if I’m running 1:20 for the half my marathon time is around 2:50 (1:20 * 2 + 10 minutes)

My current half PB is 1:15:00 but this hasn’t translated to a 2:40 marathon ( I managed 2:41:41 ) mainly due to the half time being one of those races when ‘the magic happens‘ and you out perform what you thought you were capable of. Saying that to get to within 2 minutes is pretty accurate. When I ran my marathon PB , 2:41:14 my half time was 1:16:24 which was closer. You can use a 10k or even a 5k time as an indicator but of curse the margin for error is probably larger as you are multiplying the finishing time by either 4 (10k) or 8 (5k). There are various websites where you can ply in your predictor race and your goal race and get an estimated finish time.  A good example is the McMillan website ( ) or Runners World ( )

The Runners World race time predictor has bee rebuilt apparently (see below)  It seems most race predictors are overall generous with their predictions and the work by Andrew Vickers paints a gloomier picture by also adding in weekly mileage. Common sense really as those runners running higher mileage would be better prepared , cardio-wise if nothing else, for the longer races. Most, if not all, race predictors never ask for weekly training mileage, strange this has been over looked.

As always I have an amusing story reference race predictors. My good friend Mike Kowal wasn’t running as much as he should have been due to injury (we use to call him ‘sick note’ as he was injured so much. There was the amusing story where he got injured at the sports massage clinic because the physio left the heat pack on and told Mike to call out if it got to hot. Not wanting to seem a wimp Mike kept quiet and ended up with second degree burns ! You get the picture…)  Anyhow because Mike hadn’t run a half , 10k or even a 5k he started to predict his marathon finishing times off even shorter distances, At one point I’m sure he had it down to a 10 metre sprint. This is not to be recommend for predicting a marathon finish not matter what magic he used, it is to be noted Mike is an Engineer and probably developed some devilish formula to show a sub3 prediction of a 4 paces sprint. Funnily enough he did go sub3 on his first marathon but has yet to repeat the magic on subsequent outings.

All joking aside can you use a race predictor and rely on the predicted time to set your pace for a marathon ? Remember the marathon runner who slows the least normally wins (or achieves their goal). Predicted time and the resulting race pace is pivotal to success when it comes to running a marathon so get the predicted time wrong and the race pace will also be wrong,  which equates to failure. I like the idea of entering two previous results in the Runners World predictor because it adds in experience which is so important to race finish times. I can predict to within +/- 90 seconds my marathon time these days and this is all down to experience. With over 60 marathons and ultra’s under my belt I can predict my race pace very accurately and have only hit the wall twice, once in my first marathon by not preparing and running a time beyond me and once when I was struggling mentally and talked myself into failure. (I have covered the mental part of marathon running in various posts, so important and so under estimated.) With experience and good quality training comes consistency. 

I used the Running World Race Time Predictor and added in my best 10k and 5k times and adjusted the weekly mileage until I got a sub 2:40 marathon time. I had to get two 160 km a week (100 miles) to get my predicted time under 2hrs 40 minutes for the full marathon. Probably about right truth be told but it never asked for my age which would surely be another factor that would be pivotal to the accuracy ? I’m 51 in a few weeks and assuming all the variables where the same I would expect a 35 year old runner, for example, with the same race times  to out perform me. Maybe I’ll try and run a 2:38:30 in June this year, hell the computer says ‘Yes’ so what could possibly go wrong ?


160km a week, piece of cake.
A Mike Kowal 10metre marathon predictor run with the man himself. I think he predicted a sub 2hour marathon off this one?


  1. Mike | 13th Jan 18

    Due to injury at the time (as usual) I had to start running short races first, starting with 3k, then 5k, 10k, 15k, half marathon. I always ran the longer races better than McMillan race predictor based on the previous shorter distance race times. So being an engineer I just assumed a linear progression based on pace and that I would always outperform the McMillan predictor, This actually worked all the way up to my marathon time, and I think the main reason why is that I was confident that my prediction was more accurate. So mentally I was comfortable with my predicted pace and either achieved that pace or bettered it. Using this new Runners World race predictor based on my race times and km run per week at that time my marathon time should have been 3:11 but I ran just under 2:58. McMillan was far more accurate for me, I think around 3:01:30 based on my HM time. So my advice is to find race predictor calculations that are more accurate for you (we are all different) with a margin over or under extrapolated based on recent race times. The important thing is that the race times need to be recent, no point going back a few months, has to be no more than 2 or 3 weeks before, and then have confidence in your prediction to be able to achieve it on the day!

    • | 14th Jan 18

      Thanks Mike, nice reply. Totally agree reference finding an indicator predictor that matches your times, which of course is where the difficulty arises. Running is so personal and individual a ‘one size fits all’ predictor is never going to accurately map individual performance but it can serve as a litmus test along the training journey.

Leave A Comment

Your email address will not be published. Required fields are marked *