Do Service Errors Matter?

Interactive Dashboard:

https://public.tableau.com/profile/chad.gordon#!/vizhome/ServiceErrors/Service_Errors

I’d argue, no. At least not in the way that most people think?

The core issue here is that given a specific input, we see a wide range of outputs. As you can see from the rainbow..ish blob on the right, at any given range of Service Error Percentage, there’s quite a variety of Expected Outcomes.

On the left, we look for changes in the trend based on Wins or Losses. Unfortunately, the trend is unyielding and suggests consistently that missing more serves…is bad. This simultaneously passes and fails the eye test. Obviously if we miss too many serves, we’ll have given away too many free points to win the set. But on the flip side, coaches are always saying that if we serve too easy, the opponent will be in-system and sideout relentlessly. I’d like to sit on the fence here and just say: I hear you.

I’m not suggesting we chip balls in underhand, nor am I promoting wildly aggressive serving strategies. I’d just like to point out that many of the voices in this sport grew up and/or were heavily influenced by those who played in the sideout scoring era – and I always wonder how much this influences their thinking. Gotta think it would be tough to takedown Karch on the big court in the 1993 AVP KOB in Daytona Beach, Florida if you’re just lobbing in freeballs from the service line. But I disgress.

In the dashboard above, we skip over Ace to Error ratios, Passer Rating vs. Error %, and even FBSO vs. Error %. Instead, we skip right to the good stuff with Point Win Probability (PWP). Here we look at all the various outcomes from the serve: quality of pass, location of next touch, is it an overpass (if so, was it attackable or just a freeball), etc. For each unique combination, we look on average how often all teams (in the 7500+ matches of data) won the rally. That way, each unique outcome carries with it a probability of winning, thus Point Win Probability (PWP). We then use these values to determine the expected value (here: expected sideout %) of each team, in each set included in the dashboard.

The point is that you can be really effective in a variety of ways. There’s no right or wrong philosophy. Some of John Speraw’s successful teams are notoriously high error from the service line – while if you watch Hugh McCutcheon’s Minnesota squad, they really adhere to the matra of: “hit a good ball, in the court, often

You could of course argue that in the aggregate, any effects from weaker teams might wash out the results. But here is just the top 10 RPI teams in sets where they are playing another top 10 RPI team. Seems to stay true to what we’ve seen. You can win and lose at a variety of SE % levels – but you can also hold your opponent to low Expected Sideout numbers even if you miss 5-10%.

Anyway, I think service errors get let off the hook too often. For most of us, not playing at the absolute highest level of international competition, maybe we need to be a little more mindful of keeping the ball inbounds?