How to evaluate serving?
Who hits it the hardest?
Who misses the least?
Whose float...floats the most?
My thesis: do you lower your opponent's likelihood to win the point?
Because all these other things, they're just proxies for what we consider "good" serving
But really, rallies are just tug-of-war
And my job is to get the momentum on my team's side
So we will talk in terms of points
A few definitions:
Exp Exp Point Score
Expected Expected Point Scoring is, given the opponent you are serving against, how well do they pass on average?
This refers to the quality of the passes
and we use the value of those passes to an average team
This way, we're excluding the power of the offense -
and instead, isolating the actual reception and how much value it would bring to an average squad
An EEPS of 40% here means that you are expected to win 40% of the points when you serve.
Actual Exp Point Score
Actual Expected Point Score is, given the actual receptions you created as a result of your serve, what is their value to an average team?
So again, we take the quality of the passes
And the value of those passes, to the average team.
Again, so we can remove the skill of the attackers, we look at the value relative to "average"
An AEPS of 45% here means that given the receptions your serving created, you are expected to win the rally 45% of the time
Point Score Over Expected
Now we just take the difference between Expected and Actual from above ^
PSOE = Actual - Expected. This is the value the server has created, above/below expectation
(Naturally, positive = above expected // negative = below expected)
In our example, AEPS - EEPS = 45% - 40% = +5%
This means that our server is adding 5% to our expected point scoring when he/she serves.
When converting between Expected Value and Efficiency:
+5% in point scoring = lowering opponent offensive eff by 0.100
so if your opponent usually hits 0.200 on First Ball,
this server lowers that to 0.100, which is pretty impressive...
You'll also notice that their ranking, relative to opponent Passer Ratings, are sometimes similar, sometimes way way different.
This is yet another reason why Passer Ratings are dumb and can blur the picture when evaluating actual value.
You can see more about their stupidity here: