Great explanation! One doubt - if a particular value is x STD's then what are the real-world applications for that piece of data? For instance, at 3:39, the athlete is 1.1 standard deviations more than the average, so how can we use this info?
Hi Joyan, that is a great question. It gives us a standardised score that represents how much higher or lower than normal it is for that athlete. Often people use a threshold of +/-1.5 as a flag or threshold to know it's notably higher/lower than normal. So if it is a wellness score for instance, it might show us for that athlete they are much higher/lower than normal. We might set up a dashboard that flags every athlete's z-score that's higher/lower than 1.5 z-score, so then although everyone is entering their own values, we can see on a single view who is flagging and we may want to explore those specific athletes further. Another application might be to compare a number of different tests in one view. So we can calculate the z-score of an athlete's test results compared to the rest of the team e.g., sprint time, jump height, body comp, strength etc. Although these are all very different numbers, we can use the z-score to see how the athlete compares to the group, and just how much higher/lower they are for each test, all on the same scale. Does that make sense? Maybe I'll do a video discussing this further!
Hello! Congratulations on your channel! It is very useful and clear! What is your opinion of using Z-Score to control the load, for example of the distance covered, of an athlete micro cycle to micro cycle? Is it really an alert if a player presents a Z-Score > 1.5 between two microcycles in his distance covered (for example)? Thank you
Thanks so much Sebastian! I think it comes down to how we interpret (and action) those z-scores. As always it is hard to know what threshold to use, but I do like to use +/-1.5. We might describe this as an "alert" but it is not necessarily a bad thing. The z-score is just describing what is going on in the data and can be a useful flag for us to investigate their training load further and look into the context as to why they've had a relative jump in load. Does that make sense?
Thanks for all your comments and feedback on this video! If you enjoyed this, you may also be interested in my video called Transforming Z-scores to T-scores & STEN Score: ruclips.net/video/16GZ1Ed2uVE/видео.htmlsi=LOeB4fH1LRMhEh0P
Thanks for your videos! In my case, in some cases I encounter a problem that to solve I always have to go to programming and I can't get it out with Excel. When we want to do this calculation but conditioning on the player's name, match day or any other variable/s, I cannot get the standard deviation of that set. Do you have any practical solution for Excel? Thanks again!
Hi, thanks for your comment! Yes, you could use some sort of combination of lookup function to extract based on a particular condition. Probably most commonly though, I use a pivot table to extract by player name or periodisation day, and calculate the z-scores from that. Maybe I should demonstrate that in a future video?
Hello, thank you very much for your videos. I would like to ask you a question. When the sample is less than 30 (example: 15 players), shouldn't a Student T-score be used, or is it the same to use a z score even if it is less than 30?
Best explanation of Z-scores I have come across for sport, good job.
Thank you very much Paul!
Great video! Found you through linkedin and z values are a hidden gem for most sport clubs data analysis.
That's great, thank you!
Great explanation! One doubt - if a particular value is x STD's then what are the real-world applications for that piece of data? For instance, at 3:39, the athlete is 1.1 standard deviations more than the average, so how can we use this info?
Hi Joyan, that is a great question. It gives us a standardised score that represents how much higher or lower than normal it is for that athlete.
Often people use a threshold of +/-1.5 as a flag or threshold to know it's notably higher/lower than normal. So if it is a wellness score for instance, it might show us for that athlete they are much higher/lower than normal. We might set up a dashboard that flags every athlete's z-score that's higher/lower than 1.5 z-score, so then although everyone is entering their own values, we can see on a single view who is flagging and we may want to explore those specific athletes further.
Another application might be to compare a number of different tests in one view. So we can calculate the z-score of an athlete's test results compared to the rest of the team e.g., sprint time, jump height, body comp, strength etc. Although these are all very different numbers, we can use the z-score to see how the athlete compares to the group, and just how much higher/lower they are for each test, all on the same scale.
Does that make sense? Maybe I'll do a video discussing this further!
@@globalperformanceinsights Got it! Thanks for such a detailed explanation and yes, would love to see a video on its applications!
Hello! Congratulations on your channel! It is very useful and clear!
What is your opinion of using Z-Score to control the load, for example of the distance covered, of an athlete micro cycle to micro cycle? Is it really an alert if a player presents a Z-Score > 1.5 between two microcycles in his distance covered (for example)?
Thank you
Thanks so much Sebastian!
I think it comes down to how we interpret (and action) those z-scores. As always it is hard to know what threshold to use, but I do like to use +/-1.5. We might describe this as an "alert" but it is not necessarily a bad thing. The z-score is just describing what is going on in the data and can be a useful flag for us to investigate their training load further and look into the context as to why they've had a relative jump in load. Does that make sense?
Thank you so much!@@globalperformanceinsights
Thanks for all your comments and feedback on this video! If you enjoyed this, you may also be interested in my video called Transforming Z-scores to T-scores & STEN Score:
ruclips.net/video/16GZ1Ed2uVE/видео.htmlsi=LOeB4fH1LRMhEh0P
Thanks for your videos!
In my case, in some cases I encounter a problem that to solve I always have to go to programming and I can't get it out with Excel.
When we want to do this calculation but conditioning on the player's name, match day or any other variable/s, I cannot get the standard deviation of that set.
Do you have any practical solution for Excel?
Thanks again!
Hi, thanks for your comment! Yes, you could use some sort of combination of lookup function to extract based on a particular condition. Probably most commonly though, I use a pivot table to extract by player name or periodisation day, and calculate the z-scores from that. Maybe I should demonstrate that in a future video?
@@globalperformanceinsights Thanks! I'll try with lookup combinations. In the meantime, I will wait patiently for your tutorial.
Well Done !!!
Hello, thank you very much for your videos. I would like to ask you a question. When the sample is less than 30 (example: 15 players), shouldn't a Student T-score be used, or is it the same to use a z score even if it is less than 30?
Hi, thanks for your comment! Yes, I believe for small sample sizes (typically
PERFECT!!! 😍😍