Alternative (Beta) Age Grade Calculators

Recently, I’ve been writing a series of articles on Medium exploring how to bets compare race results between runners of different ages.

Traditionally, the age grading system has been used to calculate an age graded time for the purpose of comparisons. This system uses an age factor to predict how much time you would slow down – compared to a younger runner.

I’ve been exploring two alternative methods of comparisons – percentiles and z-scores. I’ll link below to the original articles to more fully explain things, but here’s the gist of it.

The percentile shows what percent of runners in a given age group finish slower than a given time. So a percentile of 92% means that 92% of runners fail to meet that time – and the result is in the top 8% for that category of runner.

The z-score is a measurement of how far above or below a given finish time is compared to the mean finish time for that group. A result of -1 means that the time is one standard deviation below the mean.

You can use the calculator below to put in race results for a marathon and see how the age grading system, percentiles, and z-scores stack up.

Please note: This is a beta version of the calculator. I put it together quickly to go along with the Medium series, and I haven’t had a chance to fully implement things like error handling. Please enter a number for the age, and enter the finish time in the format of HH:MM:SS (i.e. 4:25:31). I’ll come back and spruce things up in the future, but for just know that if it fails to return a result it’s probably a result of bad input.

Also, please be patient. Once you click ‘Age Grade’ it will take one to two seconds to populate the table with the results.

Error:

Age Grade Results

Age
Gender
Time
Age Graded Time
Age Grade
Percentile
Z Score

Note: The age graded results currently incorporated in this calculator are based on the 2020 age grade tables. The percentiles and z-scores are also based on data from 2010 to 2019, so I used these for a more reliable comparison between systems. Once I update the percentiles and z-scores to use data from 2023, I will also update the age grade portion to use the new 2023 tables.

Methodology and Context

If you haven’t been following the series on Medium, here’s a little more context. The links below will grant you access to each individual article behind the Medium paywall.

In the introductory article, I laid out some of my issues with the current age grading system. In short, it’s a valuable tool and it’s better than nothing. But unless you’re at the very top levels of performance, there’s very little difference in individual performance. The system may also may advantage certain age groups over others.

In the next two articles, I lay out the sample of data I will use for my analysis and conduct a brief exploration of the data. My analysis is based on finish times from all American marathons with over 500 runners that took place from September to November, 2010 to 2019. Collectively, this provided a large enough sample to identify stable distributions for most age groups – although things get a little funny when you start getting into the 70-74 and 75-79 age groups.

In the next two articles, I laid out an argument for using percentiles to compare efforts and then conducted some analysis of how this system works. The system is not perfect, and it is hard to distinguish between efforts at the very top end of the distribution. However, it does seem to do a great job of distinguishing the regular efforts of above average runners. Some tweaking could help fine tune things.

Finally, this article lays out how z-scores could be used to compare results. Conceptually, I liked this. But once I dug into the data, I’m not all that satisfied with it. The way the distributions are set up, it clearly advantages some groups (mostly young women) over others. Although the mean finish times vary in a clear and predictable way, the standard deviations don’t – making z-scores a less useful tool than percentiles.

I will be collecting new data from 2023 to update these systems. Once I’ve done so, I’ll also update this calculator and spruce it up a bit. Please understand this is a beta version, and I rushed it out to help people reading the Medium series better understand how these things work.

2 thoughts on “Alternative (Beta) Age Grade Calculators”

  1. Nice calculator, but it doesn’t seem to accept ages greater than 79. In particular I get no results for ages 80 and &1.

    Reply
    • Good catch, Tony.

      The dataset I was working with didn’t have enough data on runners 80+ to reliably calculate percentiles and z-scores for them. So it can’t return those results.

      I can still calculate the traditional age grading values – I just need to fix the calculator to fail more gracefully. Once I update things with the 2023 data, I’ll also see if I can add percentiles and z-scores for the 80-84 age group

      Reply

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.