In team sports it is now commonplace to have GPS wearables and different camera systems to track specific metrics such as distance covered, high intensity runs, etc. This has become a great tool in the ability to monitor a team of athletes much more closely than what was once not possible. With this GPS data we can look at past training and match days and see the accumulation of load for each athlete. This allows us to make better informed decisions on player readiness as well as potential over and undertraining scenarios.
While we look at past data to see the current load an athlete is under, wouldn’t it be great if we could see into the future and plan and predict player load? As long as you have a big enough data set of drills, you can do this as well. Predicting player loads can allow fitness coaches a way to have an estimate of how much load the team will be under days or weeks into the future. This way of planning may also help achieve the specific training goals for certain days.
I first saw predicting player loads from Red Bull’s Fitness Coach Tony Jououx. I’m not sure if this is exactly how he did this, but using this idea, I want to share an easy way to build this out as well as some potential limitations in using this method.
To build this out on excel:
1. Export all of the drills from your database with the duration of the drill and the important data columns you wish to have included in the predictor.
2. Calculate per minute values for each physical stat (= Stat / Minutes Played). An easy way to do this is to create a table in excel (Ctrl+T) and when you calculate one, it should auto calculate the remainder of the column for you.
3. Average the entire data set for each specific drill and then create a another table specifically for drill averages.
4. Create a drop down list of drills. In an open cell click “Data Validation” . In the drop down bar select “List”. Next, in the “Source” box, you’ll have to reference the drill names (in this case A2 to A11). If you click and drag over all of the drills it should something like this. If you need more help, click HERE.
Now you should have a drop down menu that allows you to choose the drill.
5. Now to create cells that will give us the correct output for each physical stat that changes with the duration input. For this you’ll need to use the VLOOKUP function.
The VLOOKUP formula would look something like: =VLOOKUP($A$14,TBLssg,3,FALSE)*$B$14
- $A$14 is the drill referenced
- TBLssg is the entire table
- 3 is the column referencing (Drill is column 1, Distance is column 2, Distance/Minute is column 3, etc.). For this we want to reference the ‘Stat’/Minute column since we’ll have a minute multiplier.
- FALSE is looking for an exact match.
- *$B$14 is the minute multiplier.
($ represents a fixed column or row – this allows you to drag the formula without changing the referenced cell)
You’d carry this same formula out for all stat columns only changing the “3” (column referenced) to whatever column you want the function to reference. So for HI Distance I would type “5” and for HI Runs “7”, etc.
If done correctly (and I explained in a way that you could follow), you should have two cells that control the output. A drill dropdown menu and duration. Changing either one, or both, will allow you to have a good idea of what physical output players will have.
While this is certainly better than using no predictor at all, there are some limitations in using this:
- Game Conditions – This is one of the biggest limitations to using this. As an example, the team could play two sets of eight minute halves of 9 v 9 with the same field dimensions three weeks in a row in training, but each week change the conditions of the game to fit the tactical plan for the week. This would undoubtedly change the physical outcomes of the SSG. In another example, the coaches could use the same conditions, but one week there is a lot of starting and stopping because the coach is providing feedback during the game and the next week the game flows with very little start and stop. Changing the conditions and the start/stop pattern of play can greatly alter load of that drill.
- Athlete Position & Effort – To go along with the previous point, an athlete’s position on the field as well as the specific formation being played during SSG will also alter the load on each player and with this idea an average only goes so far. We notice that depending on position (and effort), there are wild fluctuations from the top to the bottom of a specific stat and averages certainly don’t tell the entire story.
- Pitch Size – If the pitch size changes for the same SSG, the the physical output can change as well. If the coach uses multiple pitch dimensions for the same SSG, an easy way to fix this is to attribute pitch sizes to the SSG.
You could do this for any warmup, passing, or possession drill as well, and if anything, it can help to pair certain drills together to meet the physical goals of the day. Of course, the more data you have the more accurate these should be (having 10 samples of a drill will be a bit more accurate than only having 2 samples) so naturally, the coaching staff would need to recycle drills for this to be somewhat effective.
Although there are some limitations here in using this method, this is probably better than waiting to see the data after the session is totally over.
If you have any additional questions on this, please comment below.