More often associated with professional sports teams, the use of advanced statistics has made its way into elite amateur sport, which probably couldn’t live without it.
The director of sports sciences at the Institut national du sport du Québec, François Bieuzen, is categorical: advanced statistics have become a must.
If we did not use them, it would be depriving ourselves of a means of performing, believes the director. It is a performance aid, in the same way as being well equipped or having physical trainers and mental performance trainers.
In elite sport, advanced stats are used a lot to calculate an athlete’s training load or injury risk over the next two days.
For example, in the data we take, there are jump heights on platforms, explains François Bieuzen. The athlete goes on a platform and we measure the jump height, which gives us an index of his neuromuscular fatigue or his neuromuscular performance.
One could say: “today is better than yesterday and that’s it”. Thanks to the analysis of advanced statistics, we can see if it is related to other events.
Each sport can adapt it to its reality.
For example, Snowboard Canada has used advanced statistics to better understand the relationship between tricks performed and judges’ scores.
By placing a sensor on the boots of the athletes, the federation collected the rotation data and thanks to an algorithm, it realized that the number of rotations is one of the parameters that most influences the score of the judges, but that there are nuances depending on the gender of the athlete and the side of the take-off.
The sensor also allowed Snowboard Canada staff to confirm that landing impact is a must for training load planning. On- and off-snow training plans were reviewed, and competition downhill strategies were developed with the statistics gathered.
The Canadian rugby sevens team has also used advanced statistics to establish the training load, but also to calculate the effort load according to the type of contact that the players undergo.
Where there is still work to be done is in performance prediction.
It is very difficult to predict performance, recognizes the director at the INS. We would very much like to be able to predict performance, not so much to select as to be able to evaluate what we are doing and know if we are on the right trajectory.
The Netherlands Speed Skating Federation has meanwhile developed an algorithm it uses to select which athletes will compete in the Olympics, with very positive results.
A similar situation is not currently envisaged in Canada.
The human, the conductor of advanced statistics
To all those who believe that advanced statistics dehumanize the work of the coach, François Bieuzen is not of this opinion.
Often, we start from the instinct of the coach, the athlete or a member of the team to say to ourselves: “there, we may have observed something, is the data that the ‘we can help us confirm or refute if we are in the right direction? Or if our feeling is justified or not.
” At no time are we selling advanced statistics as a decision that would be made in place of the staff or the coaches or the athlete. It’s really to have a more objective vision because we realize that we have a brain that retains certain events, which is very selective in certain things. Despite the experience and all the good will that we could have, it helps us a little bit to have a better decision. »
The human therefore remains the one who interprets the statistics and who uses his judgment, with all that that implies, good or bad.
An anecdote that occurred during the World Series of Baseball in 2020 demonstrates this well.
The Dodgers led the series 3-2 against the Tampa Bay Rays. But in Game 6, Rays starting pitcher Blake Snell delivered a stunning performance.
In fact, Tampa Bay was leading the game 1-0 in the sixth inning when Rays coach Kevin Cash decided to pull Snell because he didn’t want him to pitch in the third at bat.
Advanced statistics show that pitchers’ effectiveness decreases at this time.
Result: The other pitcher was knocked down and the Dodgers won the game and the World Series.
The coach also confessed after the game to regret his decision
because it didn’t work…
We didn’t invent anything
Athletes have been compiling data and statistics for a long time. The novelty is rather that artificial intelligence makes it possible to collect more data and compile it much more efficiently.
We are still on statistics as we did 20, 30 or 40 years ago, recalls François Bieuzen. Except that now we have powerful computers that allow us to go much faster to make calculations and manage large volumes of data.
Not all Canadian federations use advanced statistics in the same way. It is very expensive to collect and classify data. It also takes a human and a lot of time to analyze them.
It can also be difficult to harvest the basic material. Athletes have to fill out questionnaires, do physical tests or compile results and there must be enough numbers to be able to draw a conclusion.
The sports with which we have worked, it takes on average three, four, five and even six years before being able to really use the data, underlines the director of sports sciences. With speed skating, it took us four years to have data that we trust and know that it really helps to make decisions. Before, it was too fluctuating, it wasn’t solid enough.
Advanced statistics probably still have a lot to offer. François Bieuzen hopes that we will eventually be able to predict injuries over more than two days and predict the performance of athletes over time.
As accurate as sports data can be, there will always be exceptions that prove the rule. Talk to Kevin Cash.
A Quebec company has jumped into the world of sports data with both feet and is taking its place in this booming market.
Hexfit was born in 2015 from a need for health professionals to have a tool to manage the growing number of data while continuing to follow up with customers.
The software standardizes and centralizes sports data, makes calculations, curves and coaches, nutritionists or kinesiologists can access it.
A version of Hexfit was tailor-made for the INS, but the software is also used by sports teams, trainers in chains such as Nautilus Plus, Énergie Cardio and YMCAs. Professionals from around thirty countries have integrated the software into their daily lives.
It’s a bit like our dream to collect the data of a team or an athlete in high school and that his file follow up to the professional team, so that he has the same file foreverunderlines the president of Hexfit Étienne Dubois.
With the growing popularity of connected watches, step counters and sensors of all kinds, sports data is multiplying at a crazy pace.
coach n’a pas le temps d’aller dans chaque dossier pour voir si vous avez marché vos 10000 pas depuis une semaine, affirme Étienne Dubois. Un logiciel comme Hexfit peut lui dire : “tu as une cliente X qui n’a pas marché ses 10000 pas cette semaine”.”,”text”:”On a aujourd’hui une accessibilité à la donnée qui est plus grande, mais le coach n’a pas le temps d’aller dans chaque dossier pour voir si vous avez marché vos 10000 pas depuis une semaine, affirme Étienne Dubois. Un logiciel comme Hexfit peut lui dire : “tu as une cliente X qui n’a pas marché ses 10000 pas cette semaine”.”}}”>Today we have greater accessibility to data, but the coach doesn’t have time to go through each file to see if you’ve walked your 10,000 steps in a week, says Étienne Dubois. Software like Hexfit can tell him: “you have a client X who has not walked her 10,000 steps this week”.
The president is formal, Hexfit is one more tool for sports or health professionals.
It certainly does not replace the human and it is part of our vision, we do not believe that the human can be replaced in all this, there are too many aspects of motivation.