two black runners on blocks ready to race

case studies Ellida

overall improvements

  • 83 %

    runners met their goal time, beating it or getting within 60 sec
  • 3.8 %

    average error in race time prediction
  • 2.8 bpm

    average error in heart rate prediction

what was the challenge

runner finishing

and this is what we improved

  • Athlete profiling
  • Race prediction
  • Sport performance
  • Training planning

As seen in the flow chart, the initial 8-week training program for each runner was created by Ellida based on individual profiling, including objective measures (training log), subjective measures (fitness level, goals, and commitment), combined with domain expertise (philosophy by an world class coach). The first iteration of the training program was run through thousands of simulations, each with a slight tweak to the program, and the best simulated results, i.e. “Optimal schedule”, was provided to the runner. The training program is then continuously updated in a feedback loop with key metrics such as subjective ratings, compliance to the program, and training performance data.

Ellida accurately predicted both Race Heart Rate and Race Time, with an average error of just 2.8 bpm and 3.8 %, respectively. The accuracy held true across performance levels, from runners who ran the 10 km in 35 min to over 60 min with a very strong coefficient of determination, R2, of 0.81. What’s more, feedback from the runners indicated that they found the plans useful and enjoyable, and most were able to reach new Personal Bests.

would you like to learn more details?

Here is the detailed case study to download in PDF format.

In order to download it you need to leave us your email address.

RaceID Case Study