for RMD version: https://rpubs.com/chadgordon09/step8

1. No reason to bury the lede – **all attacking situations are not created equal**.

2. A middle attacking on a Front 1 after a perfect pass is expected to score at a better rate than an outside hitter receiving a high ball over her shoulder from 30 feet away. I think we can agree on this. One of these is objectively easier to score on.

3. Because these situations are not equally challenging, **our expectations should be dynamic** to account for the context in which attacks occur.

4. How we arrive at our expectations however, is open for debate. Above, I’ve include 3 potential ways to do it – but there are certainly others. I’ve also done these calculations in the aggregate as you can see from the (*Counts*) in the graph – but doing these for your specific team and/or opponent is likely the way to go.

4.1 Input Expected Value numbers are created by looking at a given situation (let’s say OOS attack + 2 blockers) and averaging the number of times a team in that situation won the rally vs. lost the rally.

4.2 This is very similar to Expected Goals in soccer or Win Probability in baseball – we’re using historical data *given situation X to say that we have seen result Y on average*. Teams with an OOS attack + 2 blockers historically win the rally at 0.091 **( A won rallies – B lost rallies) / C all rallies.** That’s all we’re doing. Nothing crazy.

4.3 For those in the NCAA, you’ll understand that hitting an OOS ball against Stanford is not the same as hitting an OOS ball against Northwest Missouri State Community College…East. Because of that, doing this for your specific level – team – opponent – etc…is probably the ideal setup. But for our argument, we’ll just talk conceptually.

5. So here’s how you might use this type of analysis. Above is the 2019 National Championship. In this case, we use the 3rd example of how to create “input situations” and “input expected value” – looking at whether the attack was In/OOS combined with the type of block the attacker faced.

5.1 The **Input Expected Value** is in the 4th column and is constant for each attacker in a given situation (again, you might prefer to customize these input eVs so they align with the specific team and opponent in question).

5.2 The player’s actual hitting efficiency in each situation is color coded on the right. You’ll notice that OOS vs. 2 blockers, is where Kathryn Plummer eats, sleeps, and collects her championship rings.

6. This type of thinking can be **applied to all the skills**. Once we start talking about the game in terms of probabilities, efficiencies, and terms which directly relate to points, Expected Value type metrics can be used universally.

7. For example – we could look at blocking as simply the inverse of attacking. If the degree of difficulty for an attacker has to do with whether the set is In-System or OOS and the number of blockers faced – then could we flip that and say that blocking difficulty changes depending on In/OOS conditions and the number of blockers alongside you? The same could be said for digging. We’ll look at these other skills down the road…

8. Adding input expectations is key to getting us where we really want to go: looking at ** Value Added**. Again, not the bury the lede – we’re simply going to look at

**Result minus Expectation**. Output minus Input as our means of assigning value added on a given contact. Did you take your team from a good situation to a great one? Or did you take a good situation and make it worse?

Step 9.

## 2 Comments