Heatmap Intro


*the net is at the top. the endline is closest to this text

In the land of cones and zones…and subzones, it’s easy to forget that these are merely representations of locations on the court. Equipped with that logic – and sparked by our friends at VolleyMetrics – I converted the zones to xy coordinates.

Yes, it’s annoying that the x-axis is above the heatmap, but I wanted to throw this up here anyway. It’s the hitting efficiency of attackers in the Big Ten in 2016 based on where the set came from – and it’s what you’d expect. Perfect passes lead to good things, moving your setter forward isn’t terrible, but moving your setter backwards is horrible. Below the hitting efficiencies in each grid are the frequencies.

One thing I think that gets overlooked is being content with medium passes. It goes back to the FBSO posts and why getting aced hurts. Yeah you’re hitting .100 instead of .300, but at least you aren’t losing the point 100% of the time like when you get aced or fail to dig a ball. That’s why keeping digs on your side is huge. As long as they don’t result in a set from the endline, you’ll be hitting positively – rather than defending against a perfect pass after you overpass.


Here’s the same thing for the Pac 12 from 2016. I’ve set the cutoff for both graphs between green and red at 0.150. But again, the frequencies are below the efficiencies so I would definitely not devise an offensive system that runs through zone 1D. Just thought I’d add it for those who were curious.

Anyway, looking forward to messing around with heatmap-type stuff to answer a variety of questions. The inherently great thing about heatmaps is that they convey ideas incredibly quickly. You glance at it and understand that you’d want to pass into the green rather than the red. Stuff like this for serving targets, setter tendencies, attacker tendencies, etc  can all be addressed in this fashion. Heatmaps aren’t a novel concept for scouting by any means – both VM and Oppia have employed them (I believe) – but I got curious if I could build something similar and it looks like it’s prety easy to do with either R + ggplot2 or Tableau. If you have suggestions for ways to employ this, I’m all ears.


Top Passers in the Big Ten


Here’s the same type of viz I made earlier for servers.

Just because I was curious who the best passers were, relative to the league average for FBSO value provided by each reception quality. What you may immediately notice is that Nebraska has 3 of the top 4 passers in Albrecht, JWO, and Rolfzen. So for those who unabashedly declare Kelly Hunter the best setter in the Big Ten, just realize she’s dealing with great raw materials to work with.

Another insight that isn’t super insightful is that getting aced strongly correlates with your overall effectiveness as a passer, but anywhere from 1 to 5% reception error can still drop you in the upper echelon of passers in the Big Ten. Reception error has the strongest relationship to overall effectiveness out of all of the pass qualities which isn’t necessarily surprising. This makes sense since relative to the average FBSO eff, getting aced is the farthest from the mean. Even if you were expected to hit zero in FBSO, a perfect pass only raises you .300 points higher than that – but getting aced drops you more than 3x this distance, all the way to -1.00. And this is where coaches lose valuable insight. On a 3 point passing scale, the distance from 3 to 2 and 2 to 1 and 1 to 0 appears equal, but in reality a 3 and 2 are very very similar in terms of the value they provide your offense. On the other side, 1 and 0 point passes seem similar, but even on a 1 point pass you’re still getting a swing and hitting above .000 whereas for a 0 pass, you lost 100% of the point every time. This is why we use FBSO eff and breakdown passers in this fashion.

The distinction between passers who exhibit similar effectiveness yet different reception error is of course the breakdown of each quality. Below is this breakdown for the top 5 passers in the Big Ten in 2016.


Because JWO passes so many balls perfectly, she can get away with having a slightly higher error rate. Albrecht passes the second lowest of the 5 perfectly…but is at 38% R+. Much like serving, each player has their unique footprint and the underlying factor is the value each type of pass brings to the table.

Screen Shot 2017-03-11 at 10.24.02 PM

Working from the perfect pass at the top, to the reception error at the bottom, these are the FBSO efficiencies for the Big Ten in 2016 on average.

The idea that is often overlooked by coaches is the insignificant difference between good and great when it comes to serve receive. The league as a whole is hitting .258 on passes with only 2 options. R! boils down to setting the ball from the 10-15 foot line to the Go or the Red. So while coaches may freak out that they’re losing their quick hitter option, it’s really not a huge deal when you glance at the numbers. Of course these relationships between reception qualities are unique to each team and the setter in your arsenal, but this trend is not uncommon.

This is why players who may not pass perfectly, but always get your team a swing are invaluable. These are the consistently medium players – just a really hard dead medium passer. They don’t pass nails, but they don’t get aced. That’s a big deal – and it’s currently underutilized because it’s undervalued because it’s misunderstood.

For those who are interested, here are how the teams overall shake out in the passing ranks from this season: Team – expected FBSO eff (rec. error%)

  • Nebraska – 0.211 (3.56%)
  • Wisconsin – 0.194 (3.41%)
  • Maryland – 0.188 (4.63%)
  • Penn State – 0.186 (4.64%)
  • Michigan – 0.177 (4.68%)
  • Minnesota – 0.170 (5.19%)
  • Michigan State – 0.157 (5.35%)
  • Indiana – 0.156 (5.86%)
  • Ohio State – 0.154 (6.26%)
  • Iowa – 0.150 (5.87%)
  • Illinois – 0.143 (6.50%)
  • Purdue – 0.136 (6.59%)
  • Northwestern – 0.112 (7.38%)
  • Rutgers – 0.072 (10.1%)

Again these are looking at each team’s passes in the context of what they are worth to the league as a whole, not just how each team hit in FBSO… (bravo, Maryland)

Top Servers in the Big Ten


Coming back from the SSAC in Boston this past week, I’ve been putting more thought in player evaluation against the market they’re situated in. Much like how baseball uses WAR (wins above replacement) to compare players’ values against that of an average MLB player. That stat has of course evolved over the years with different measurements for position players and pitchers, but the underlying principle has remained constant.

Looking at volleyball, there are 6 (7 if you count freeball passing) discrete skills so a single skill WAR metric makes a little less sense, but the general philosophy can be applied as a way to compare performance against league expectancies.

So in the above viz, I’ve used the league average PS Eff for each of the receive qualities. Which looks like this table below:

Screen Shot 2017-03-11 at 4.12.43 PM

Service Ace on the top, working down to service error at the bottom. And yes. Service ace should absolutely be at 1.0 and I’m not sure why it isn’t, but .998 and 1.0 are pretty darn close for our purposes at the moment.

Using these numbers, we then look at the frequencies a player served and got each of these specific outcomes. Multiply frequencies by efficiencies, add them up, and divide by the number of serve attempts and voilà!!

I’ve built this viz to again look at the relationship to service error percentage (while highlighting the top servers). You’ll notice there’s a slight negative relationship between effective serving and lower service error, but it’s not definite. Especially when looking at the servers who bring the most value, there’s certainly a range of error in that group – and almost a correlation of 0 if you draw a box between SSS, Davis, Swackenberg, and Kranda.

However in a general sense, you can assess the value each server brings to the table based on what her results are worth against the league average. In this case, Kranda comes out on top as giving you the best shot to point score.

Clever folks might be wondering what her breakdown looks like in terms of percentages of each outcome. So voilà again!


^ Here are the top 4 servers’ breakdown by each of the outcomes.

What you’ll notice is that they all have a unique footprint. Kranda makes her money by serving aces (around 14%) whereas SSS lives in the consistently good realm. SSS only misses 2% of her serves. That’s a huge deal. She keeps consistent pressure and even though her sum total of ok+good+perfect passes is higher than the others, she doesn’t give up free points, which results in her being the 3rd best server in the Big Ten in 2016.

I’m just starting to look at the data from a “what’s it worth relative to the league” type of standpoint, so I’ll likely have more posts like this soon. Previously, I’ve focused more on “what’s a player worth to her team” and specifically “what’s a player worth to her team, in this context, when playing opponent X.” I think the way I’ve approached this previously has merit, especially since we don’t have a marketplace for trading players like professional teams do – but you could easily evaluate All-Americans and other interesting things by comparing players to league data.


Out of System Thought


It’s a pretty crude visual, but an interesting topic. So I was out in Boston this past weekend for the Sloan Sports Analytics Conference and got a chance to talk stats with the VolleyMetrics crew for what probably seemed like 48 straight hours to them. You should read their paper, it’s a great step forward for volleyball analytics. When I landed on Thursday, I grabbed a beer with Matt Jones, a guy I coached with in SB who has gone on to bigger and better things as an assistant for the men and women at Harvard.

One of the topics was around where you’d be better off setting a high ball from. My contention was that when I attack an out of system ball, if I can see the set and the block at the same time, this should make it easier to aim my swing. Biomechanically, this angle is also more similar to what an outside is used to seeing – rather than a ball coming over his shoulder from 15 feet off the net.

So what you see above is the court broken into zones/subzones and the hitting efficiency of attackers in the Pac 12 from the 2016 season when the set originates from the zone in question. This is only for high sets to the OH. It does not differentiate who is setting, just that these are classified as High and not Go sets to the left side. If there were fewer than 40 attempts from the specified subzone, I crossed them out.

I’ve circled subzones for which hitting eff is above .130, a completely arbitrary level. You can determine on your own what level you expect, but it offers an interesting conclusion that seems to support my hypothesis. Attackers in the Pac 12 hit better on this type of set when the set originates closer to the net/closer to the middle of the court. I’m admittedly posting this quickly rather than creating a pretty visual and running it for the Big Ten, but I hope it sparks a new thought for some readers.

What this might suggest is that rather than having your setter dig balls over to your libero in left back, you might want to dig to about 5-10 from the net and let the next set originate from the middle of the court. As you can see, zone 3 is a great spot for this set to come from and may influence how you handle easy digs by your setter.

Of course this is a little different for each team and if your go-to high ball hitter is your rightside, you’d run this looking in the other direction. But hopefully this at least makes you rethink casually shoveling setter digs to left back. If you look at Stanford over the last few years, often it would be Inky or Lutz setting out of zone 3 to the pins. Admittedly this takes time to get your middles to do this, but a bumpset by your libero or even using your BR outside to handset from zone 3 might be the way to go.

*edit* just kidding, ran it for the Big Ten – circled those above .130 again. Some subzones have a small number in the upper left corner, that is the number of attempts if they were on the lower end of things. So we might view the effs up around .290 with a grain of salt for zones 2C and 3B.


Just looking at the numbers and ignoring my circles, this viz might ask more questions than it answers. Couple thoughts just off the top of my head that are merely possibilities. Do the setters who set from tight to the net also put up sets that are too tight for most hitters to deal with – leading to those .009 and .071 effs? Are blockers in the Big Ten better overall than the Pac12? Are attackers in the Big Ten worse? Are defenders in the Big Ten better? Why are zones 8C/D and 5B better than others? Is it because teams train from 8c/d more often? Is this actionable if I’m a coach in the Big Ten? (probably not).

I need to convert my subzones to xy-coordinates so I can start running “heatmap” type visuals…

One idea is to create a setter cheatsheet showing how different set origins should be distributed to different players (perhaps even by rotation if you’re fancy). Created visually, a setter could see that “perfect” passes should go to Mary-Kate, while medium passes toward zones 4/7 should go to Ashley, while poor passes to zone 6 should go to some other Olsen sister – all in an attempt to help your setter distribute the ball more efficiently.