Delaware Valley Orienteering Association

DVOA member Peter McLaughlin has worked on an analysis of our orienteering results, adjusting for age and gender, creating “age-graded” results.

See here for the results from Ridley Creek last November, and his explanation below - and feel free to reply with questions or suggestions.

Subj: Method for Comparing Running Speeds

over Race Distances of 3 km to 18 km

to World Road Race Record Speeds

This is the an example of

this concept, as applied to the four orienteering courses at

Ridley Creek State Park on 01 December 2024.

Method Overview

Objective:

The method presents a competitor's performance as a percentage

of the world record speed (%WRspeed), based on race length,

finishing time, age, and gender.

Process:

1. Developing the Standards:

World road race records for distances of 5 km, 10 km, and 15 km

and both genders are stored in an Excel sheet.

The records list ages and finishing times. The ages are then separated into five-

year Age Groups, ranging from 15-19, 20-25, 25-30, up to the maximum age.

The fastest finishing time in each Age Group is used to calculate

the world record speed in meters per minute for the distance and Age Group.

This WRspeed serves as the Standard speed for the Age Group.

For race distances other than the 5km, 10 km and 15 km distances,

each Age Group has a second-degree polynomial equation

that receives a race distance in the range 3 km to 18 km as input

and outputs the Standard WRspeed for the Age Group.

2. Comparing the Local Race to the Standard:

The local race speed, the Sample, is divided by the Standard WRspeed

of the corresponding Age Group and multiplied by 100

yielding the %WRspeed, the comparison to world record performance.

Discussion:

Each of the Age Groups, thirty-four in number,

has a regression equation which accepts input of race length

and generates WRspeed as output. The equations are stored on a sheet.

The regression model producing WRspeed shows an R 2 of 0.98 or greater,

indicating a good correlation between smoothed and raw data.

Currently, the application is confined to Excel spreadsheets, involving

a fair amount of manual work. Attempts to automate operation by

python programming were not profitable with existing resources.

For extensive real world work automation is indicated.

Questions, comments, corrections, etc., are welcomed,