Introducing Expected Saves

Expected Saves can tell us if Gurpreet is over or under-rated

If you’re here, chances are you already know what Expected Goals is. If you don’t, the stat has taken the football analytics community by storm and countless articles have cropped up applying the stat to real life scenarios.

Here’s one about Burnley for example

I’ve been maintaining my own version of this stat for Indian Football. I’ve also gone one step further and created an Expected Assists metric.

Today, we’re probably going to squeeze that last juicy bit of goodness from Expected Goals and see what it can tells us about a Goalkeeper’s performance.

Expected Goals might seem like a complicated metric but the easiest way to understand it is as “Probability of chance being converted into a goal.” So, when we say a chance has 0.25 xG, we mean it has a 25% chance of being scored.

That also means there is a 75% chance that it was either missed or saved or blocked. If we only take shots that are on target and not blocked, we can calculate a new metric xSave as,

xSave = 1 – xGSoT,

where xGSoT = xG for all shots on target faced

Similar to how we add up xG values and then compare it to real life, we can also add up xSave values and compare it to real life. Here are a few charts that let's us see how xSave compares to regular saves in the Hero Indian Super Cup.

In general, the two charts are pretty similar. Two things to note - Shilton Paul drops away from top 3 in actual saves, hinting at under-performance. Ricardo Cardozo has made more actual saves than anybody else, indicating over-performance.

xSave alone doesn’t give us as much information as we would like. As you can see in the above graphs, in a knockout tournament, players that play more games will have higher xSave. So how do we scale this?

Saves to Expected Ratio = Sum[Saves] / Sum[xSave]

This number will tell us how many saves a keeper makes before he’s expected to make a save or “Saves per Expected Save”. Ideally, we should see goalkeepers who’ve done more than what was asked and kept the ball out of goal.

Immediately, we can see Kiran from Minerva, Cardozo from Churchill and Sanjiban from Jamshedpur who had some outstanding games for their teams come to the top of the list even though they only played one or two games. Gurpreet is 4th but that does not mean that the other 3 are better than him. In fact, it's a testament to his quality that he has such a high ratio even though he plays for the best team in the country. He's also faced more shots than the others and saved nearly everything!

The Hero Super Cup is a small sample size and this chart does not tell us much about the quality of goalkeepers in the country. However, it is a start and as I have demonstrated it is a powerful metric.

Keep track of the xSave standings by following @sgtsaltnpeppa on Twitter.