How to assess fitness testing results. Atler Model for ranking.
Most of the coaches use norms or statistical analysis to rank testing results. Can they analyze data in a way that gives more actionable insights for your particular group of athletes? We think they can.
What’s wrong? Assessment challenges
There are 3 main issues:
- from a test report it is hard to say if a result is really high or low;
- norms are never 100% valid and don’t show particular differences inside of a group;
- hard to assess specific or customized tests results.
Having years of experience coaches can assess a result more or less right away. It even seems that anyone can look at testing protocol and say which results are the best and the worst. The question is how high or low a result supposed to be to consider it as significantly different? It is important when a coach is designing training program or defining significant athletic strengths and weaknesses?
Let’s look at the following Sprint results for football players. Is Kevin Masterson’s 4.21 high enough to be an insight?
It seems that it is — it is quite better than average. But we will get back to this example a bit later.
Norms may help here. Most of the professional coaches have their own norms and ranks. They appear after years of experience and help understand where an athlete is against others. It is important, but to get objective picture is hardly impossible — nobody has valid data. Moreover norms are valid only for particular country, age, sport, league and even position.
Normal Distribution and Standard Deviations
We believe statistical analysis may help. Many coaches use normal results distribution and standard deviations to analyze testing results. If you are not familiar with the terms — you may read about it here and here. In a word, it helps to distribute current set of results to groups below and above average. Results which are within ‘very’ and ‘far’ above average are 15.8% best of all results. Outstanding result for an athlete. The same if fair for results within ‘very’ and ‘far’ below average.
It does help you assess and compare athletes’ results. Most of all it helps to see outstanding results. If an athlete result is in +2e and higher zone — his or her result is in 2.2% of exceptionally great ones. Coach may have a strategic asset in that athlete.
What’s wrong with Standard Deviations?
Normal distribution has two issues when used for fitness results evaluation:
- there is no normal results range;
- accuracy depends on number of results.
For standard diviations of normal distribution results are either below or above average. Coaches always have results they need to improve or consider as competitive advantage. But do they really?
Let’s get back to our Sprint test. Gabriel’s 4.35 is below average. Hayden’s 4.33 is above. Difference is minor.
Can you consider to use Hayden’s above average as competitive advantage? Hardly so. Do Gabriel and Hayden supposed to have different speed training programs? May be, but probably no if you look just at testing results. In fact Gabriel and Hayden have modest speed characteristics.
We think that these analysis doesn’t give you actionable insights.
Efan and Kevin have 4.47 and 4.21 seconds respectively. That is significant difference, but for normal distribution they are still just below and above average. Same as Gabriel and Hayden.
Idea how Atler Model helps to improve situation will be in the next section.
Second issue with normal distribution is that you have to have enough results for it to be accurate. In a word, the more data you have — the more accurate is ranking. And vice versa. More about it you may read here, here and here.
There is not much you can do with that other than by taking bigger set of results. For example Atler doesn’t solve this issue once and for all, but has something to offer. It has option to analyze results taking into account all results you have for particular test type in the system. Lets say you have whole team testing results for three 2019 testings in Atler. To assess Kevin’s Sprint result it will take all athletes results from all three testings and build ranking model. The more data you have — the more accurate is analysis.
Atler Model
It tries to make evaluation the way that a coach has clear answer in every result. Is a result low enough that coach needs to fix it? Is it just normal? Is it high enough to be advantage?
It is based on standard deviations of normal distribution, but Atler Model considers 38.3% of the results in the middle as just normal. They are not low or high enough to consider them actionable. You neither need to fix it, nor can use it for competitive advantage.
Atler Model has 5 marks: Very Low (6.68%), Low (24.17%), Normal (38.3%), High (24.17%), Very High (6.68%).
Results outside of normal are something you should probably consider. You don’t want any athletes with Low results in Sprint compete in speed with those who have High results, do you?
Even more than that. You are getting more different marks. Results ranked with Atler Model are more diverse than results ranked with standard deviations.
Lets get back to our athletes:
Kevin has 4.21. That is High results. He is significantly faster than others. Efan’s result 4.47 is slow. In fact his result is in the slowest 31%. It is actionable to design specific training program and for competitive strategy.
Very Low or Very High results are only 6.68% of all. These results are outstanding enough and you have more of them highlighted. Standard deviations point out only 2.2% on each side.
Lets take a look at profiles built with Atler Model:
It is very simple to see if there is something you supposed to pay attention to.
We know that athletic characteristics are not everything in sports. But it is significant part. As in most of the other things each part is significant contributor to the final result. Atler Model designed to make internal results evaluation more insightful.
We want training and fitness testing to be more meaningful. We believe Atler Model helps coaches get more answers and actionable insights. You get this them as soon as you have entered testing results. With Atler Model development coaches can make better decisions designing training programs. Head Coaches can make better strategies.
We have short article explaining how Atler helps coaches and athletes.
