2024-2025 Triangle Primer

Back to the basics

Other Articles In This Series
Part 2

---

The Triangle series of posts was the backbone of the first year of SmarterVolley. I reference those articles often, but I think, after 4 years of writing SmarterVolley, that it’s worth dropping an updated primer on what I mean by The Triangle and how I use this framework. Plus, I’ll include some Sideout Differential and MTP #s from the past season.

What Makes A Stat Useful?

What’s the most important volleyball stat?

This is a question I’ve been asked a bunch throughout my public volleyball coaching career. There’s short answers to that question¹ but I think it’s worth thinking about that question really means. The trite answer is that the score is the most important stat. If you score 25 points before the other team does, you win! Slightly less facile is Sideout %. If you Sideout better than the opponent, you win!

If you run correlations, you’ll find this is the case. The correlation between Sideout % and Win % is very high² and the correlation between Sideout Differential and Win % is essentially 1.³ But this is where we run into a general challenge of statistics:

As a statistic becomes more predictive, it becomes broad to the point of uselessness.

Saying, “get better at Sideout,” has some utility, but it’s also only the start of the conversation. Get better at Sideout… how? Pass better? Set better? Attack better? In-System or Out-of-System? Kill more balls or make fewer errors? And saying, “increase your Sideout Differential,” essentially boils down to, “be better at volleyball.”

Predictive, but not useful.

The sweet spot of usefulness to me are statistics that:

1. Correlate to winning, but have a manageable scope.

2. Can be zoomed into for more specifics.

3. Can be put into a framework with other, relatively uncorrelated, statistics.

Enter The Triangle

The concept of The Triangle is my attempt to solve this. I divide the game up into three related but distinct areas of the game:

Terminal Serving
These are aces or service errors. In other words, the serve terminates the play and no “volleyball” (in terms of setting, spiking, blocking, digging, etc) happens. Framing in terms of specific statistics, there are:

Our Serve Aces
Our Serve Errors
Opponent Serve Aces
Opponent Serve Errors

First Ball
These are plays that don’t extend beyond the first attack out of serve receive. So there’s a serve, there’s a pass, and there’s an attack. This attack terminates the rally, so there’s no transition play after that. The sub-stats involved in First Ball are:

Our First Ball Kills
Opponent First Ball Stops (they either blocked us or we hit out)
Opponent First Ball Kills
Our First Ball Stops (we either blocked the opponent or they hit out)

Transition
These are plays that DO involve a second attack in the rally. So the First Ball Attack by one team is dug (or perhaps they shanked the pass and had to send a freeball) and the serving team is going to get a swing. (And perhaps the rally will continue beyond that.) Resulting sub-stats are:

Our Transition Kills
Opponent Transition Stops (they either blocked us or we hit out)
Opponent Transition Kills
Our Transition Stops (we either blocked the opponent or they hit out)

Match Triangles

For an individual match, I tend to express Triangle numbers in raw totals. For example, in the Penn State v Louisville NCAA Tournament match, the Triangle numbers were:

+12 Penn State
-3 Terminal Serves
+1 First Ball
+14 Transition

So this shows that Penn State won by 12 total points⁴ as well as the differential in each phase of the game. Louisville outscored them by 3 in the Terminal Serving phase, PSU won the First Ball phase by 1, and then they won the Transition phase by 14.

You can see that using the raw points is handy for individual matches, or a small grouping of matches, because our human minds easily process the raw numbers and can see where the advantages and deficits are.

Season Triangles

When you start analyzing a whole bunch of matches, the raw numbers start to become a little meaningless. I have an intuitive feel for being +5 in a given match. It gets fuzzier to feel what being +21 over 11 matches means. Plus, when I start doing extended analysis (of the sort I’m about to be rolling out over the next few months), I’ll be comparing teams who played different numbers of matches. Therefore, I like to translate into %s. For example, here’s that PSU team’s seasonal Triangle:

55% (Penn State)
51% Terminal Serves
54% First Ball
56% Transition

We see here that PSU won 55% of total points throughout the season. Breaking it into phases, we see they had a slight edge in Terminal Serves and a big edge in First Ball and Transition.

The %s are a little harder to process intuitively, but I tend to think of them as:

51-52% = Slight edge
53-54% = Solid advantage
55%+ = Dominant

Differentials

To put those differentials in context, let’s look at some from the NCAA season.

As you can imagine, it’s a pretty tight plot. Note that a 10% differential corresponds to winning 55% of points.⁵ The trendline says that you’ll win about 80% of your matches, or go about 24-6. Congrats to Towson with a 10.2% Sideout Differential and 24-6 record. You win the award for most statistically predictable team in the country.

Penn State is actually the big outlier up there at a 95% Win % with a 10% Sideout Differential. On the far right side of the graph you see Pitt and Creighton with 20.1% and 19.1% differentials which predict that they should never lose and they each lost a couple of matches. (Linear trendlines don’t do so well at the edges.)

Statistically I like to find trends or standards that correspond to this 80% Win%. When you are doing broad analysis, 80% chance to win puts you in the ballpark of competing for a championship at your level. And then I think going from “in the ballpark” to “winning a championship” becomes much more about individual nuances.

Using The Triangle

Over the next few months, I’ll unpack all the pieces of The Triangle along with examples and data from this NCAA season. But the broad strokes are going to be: we want to use this data as a diagnostic tool. Where are we strong? Where are we weak? And then, we’re going to use that to plan training and be more efficient with our time.

In particular, I like to look for areas that we can flip from a slight positive to a slight negative. If I’m at 48% or 49% in one of those phases, I’m really honing in on that and figure out how we can go from giving up 2 points per match in a phase to gaining 2 points per match in that phase. It’s hard to take strengths from great to greater. And weaknesses often require changes in systems or personnel to significantly upgrade. Often the battle is won in the middle ground.

1. It’s either Serve-In %, MTP, or FBK, depending on your level of play.

2. 0.72, in this recent NCAAW season.

3. 0.9, to be exact. And it literally is 1.0 in a given 25-point set.

4. They actually won by 13, but the VM scout file is missing an assigned “Won Point” somewhere… which sometimes happens when there is a ball handling error or out-of-rotation or something.

5. If you won 55% then your opponents won 45% and 55 - 45 = 10.