Technical scouting: data and context
I remember when I was 14 years old searching websites for football teams the player’s list of the first and young teams where I found Neymar at 16 years old.
At that moment not exists -or did not know- stats websites with players’ details such as transfermarkt, whoscored, fbref, and so on. I could only analyze the performance and projection by watching a video or live match.
Neymar was a wonderkid and I would have liked to have data to see the insights. Today, it’s possible through the data to make scouting of lot amount players optimizing time and discovery, this is part of Technical Scouting.
The real power of data is to change our relationship with the game. It can almost be said that the new technologies have sent to the storage room the old uses inherited from pure instinct, conjecture, and tradition. — Monchi
Technical Scouting use data to accomplish aims. The key differences with traditional scouting are:
- Data from scouts vs data from commercial companies
- Subjective vs objective judgments
- Written reports vs numerical reports.
The advantage is:
- League coverage
- Assessing “Intangibles”
- Bias
- Customization
- Information Quality
- Sensitivity to Context
- Resource Requirements
The type of data that are used are event data and tracking data. We can, in turn, classify them into standards measure (assist, goals), measure scouting (shot passes, key passes), measure context (pass received in final third), and expects (xG, xA). The measure depends on the target player.
How to use data to compare players in the scouting?
Well, we can use two techniques: Percentile Ranks and Benchmarking.
Percentile Rankings
For each measure, we can build the percentile. The percentile is a statistical measure that divides a distribution into 100 same parts.
In the next figure can we see a list of the imaginary players with the score data. Later, the list is sorted and assigned the rank (frequency table).
The percentile is 100/8 (players), and each percentile is 12.5, For example, Jack is in rank 7 and has P25.
A useful visualization to show the percentile method is radars where each measurement is assigned a percentile.
Benchmarking
The benchmark is the average of the comparison group. In the image can we see the combination between percentile ranking and benchmarks.
Tips for technical scouting by David Sumpter:
- Think carefully about what to measure
- Adjust as much as possible (Normalisation)
- Know your limitation
Both techniques allow to analyst help to scouting area and join with the scouts traditional in the make decision.
