This is the direction I think the sport of volleyball needs to go – getting all skills, all contacts, all situations, and all performance evaluation into the single language of points. The same way the NBA has expected points per shot based on the location on the floor + number of defenders, the same way NFL coaches decide whether to punt or go for it on 4th down – volleyball needs to take a leap of faith into the world of statistics and get beyond Point Score% and Sideout% and other metrics that I would argue are relics of a non-rally scoring era.
In this dashboard, we evaluate players based on expectations and reality. What we mean by this is: what kind of situation was the player dealt? And what kind of situation resulted from the player’s contact? The attacker was given a great situation with an in-system 1-on-1 swing (high expectations)…but she waffled it out of bounds (really bad reality). We make these evaluations for each of the skills shown above: Serving, Receiving, Setting, Attacking, and Digging. (Blocking is excluded due to some nuance in getting block data in the same format as the other skills, since we evaluate blockers even if they did not register a touch on the ball, which makes it a little tricky).
All the specific breakdowns for how we evaluation each skill can be found in previous posts.
In addition to the specific input and output situations, we also account for the “Tier” of the team and it’s opponent. Tier 1 = RPI 1-10; Tier 2 = 11-25; Tier 3 = 26-50; Tier 4 = 51-100. We use the same formula for your opponent. Basically, we want to account for playing against better or worse opponents – and also we want to set higher expectations for teams in the top 10 than for teams that are just competing to get into the tournament. So admittedly, it’s not 100% apples to apples since we are holding better teams to a higher standard – but within the individual tiers, we can compare like to like.
This is honestly one of my favorites dashboards, not only because I’m stealing the bullet chart concept from Nate Ngo, who again, is also doing excellent work in this area (and is kicking my ass at chess currently), but mainly because it’s a quick and easy way to see who’s performing above and below expecation.
There are certainly multiple schools of thought about how to create these expected efficiencies, but I’d like to think this is a good start – and at least gets the ball rolling within the volleyball analytics community.
If you’re struggling to get your head around the efficiencies, especially the negative ones, you can always convert to a “Win Probability” style metrics by adding 1 and then dividing by 2. For example, if your reception expected value (eV) is .400, what that really means is (0.400 + 1) / 2 = 70% chance of winning the rally. If the expectation was at 0.300 (or 65% Win Probability), then that means the passer, on average, gives your team a 5% above average shot at winning the rally when she passes. Well done her.
Anyway, like most things in their nascent stage, this will be a work in progress – but I think it’s a cool way to see how players perform relative to expectation.