Fundamental to Stopper is the idea that GK coaches can use parents, team managers etc. to “crowd source” keeper game data. So how can you be sure that as a GK coach, you’re get an accurate snapshot of player performance during games?
It’s a valid question and one that we have frequently been asked as we develop Stopper. The short answer is that 1) any data is better than none; 2) the app is intended as a training aid for qualitative review as much as quantitative analysis; and 3) over the course of time, trends in player performance will appear regardless of individual game input errors.
Let’s look at those three points in detail.
1) Currently the only data for developing keepers is GAA.
Even at the highest levels of football, the only available data on keeper performance is GAA and SOT (Shots On Target) vs Goals, which aren’t metrics that translate directly to keeper ability. Stopper’s interface is designed to be so simple that even casual followers of the game (and parents!) can track key performance data points. Since keepers tend to touch the ball much less than other players on the pitch, it makes tracking the data manageable – in testing we have found that the correlation between data entry (same player, multiple observers) is 85+% and trending higher the more a user becomes comfortable with the interface.
2) Reflection and review is a critical part of Stopper.
Stopper isn’t designed to be definitive, but as a tool to help players develop their game. In that context, even a data input error can be an opportunity for coaching. For example: at the end of the first half, the keeper was incorrectly recorded as tipping the ball out to their right. In discussion afterwards, the player clarifies that it was actually a diving save to their left. That’s an opportunity to talk about the players decision making process – where was the attacker shooting from? What was the defence doing and did they organize their defence? Did they know whether the attacker was left or right footed? How where they positioned in anticipation of the shot?
3) Aggregate data compensates for data input errors.
When looking at longer-term performance data, Stopper is designed to return relative values rather than absolutes. For example: over the course of a season a player may successfully “complete” 250 instances of playing the ball out by hand, out of 300. But when it comes to goal kicks, they only “complete” 50 out of 200. Even allowing for a 15% margin of error, that data trend indicates that this developing GK is struggling with distributing from their feet relative to their hands (a trend that would likely be correlated by other data points).
As the above illustrates, game data isn’t a substitute for coaching – but it’s another tool to help GK coaches get a better understanding of how their players apply what they’ve learned in training, to actual game situations.