Stat Clusters [Preseason Edition]
This is the first edition of what I hope can be a running series throughout the season. The idea here is to group stats logically for different player types and have a format which can give an overview of a player’s vital statistics at a glance and put them in some context.
Positional Breakdown
I have finally relented to the logical breakdown of positions into guards/wings/bigs. I don’t accept that only point guards fall outside of the wings definition, so combo guards (any player either a 1/2 or a 2/1) fall into the guards category for me. A wing can be a shooting guard, small forward or in rare cases primarily a power forward. Bigs are always power forwards or centers. For our roster (16 man, pre cut, active players only) this means:
Guards: Lillard, Price, Nolan, Karl
Wings: Batum, Wesley, Claver, Babbitt, Pavlovic, Barton, Ammo
Bigs: LaMarcus, JJ, Meyers, Freeland, Jeffries
Statistical Breakdown
I made some executive decisions on which stats would be included in the clusters for guards vs wings vs bigs. All include basic and advanced shooting percentages, PER (calculated using the simplified PER formula), USG% and TOV%. Guards get per game and per 36 minute numbers for points, assists, rebounds, steals and turnovers in that order, as well as an advanced distributing statistic in AST%. Wings have the same for points, rebounds, assists, steals, blocks and turnovers, with an advanced metric for rebounding (TRB%). Bigs’ statistical profile ignores steals and assists, but includes personal foul statistics and a deeper emphasis on rebounding, with offensive, defensive and total rebounding percentages, and shot-blocking (BLK%). The goal here is to eliminate presenting less relevant stats for such as blocks for guards or steals for bigs (although whether I’ve gotten it right in terms of what is unnecessary is something I’d love feedback on)
Layout
The way the stats are grouped in the clusters is designed to be simple and logical. PER central as the pre-eminent catch-all stat, eFG% under 3P% to signify the latters’ influence on the former, TS% under FT% for the same reason, TS% and USG% next to each other so they can be analysed in concert, per game stats and per 36 stats next to each other for easy reference with advanced stats attached to the appropriate row of raw/per minute numbers. Hopefully that logic makes for something easy to read as a quick reference (any suggestions on changes that might improve the layout would be appreciated).
Context
The way I’ve looked to add context to the stats provided in these clusters is by adding a colour coding system. If this proves worth continuing with, I’ll look to make the comparisons against leagues bests/worsts/averages but for this small preseason data set, I’ve used team averages to judge performance. The way it works in this post is as follows:
Black squares mark a team worst performance, bronze a below average one, silver above average and finally gold for being best in a category. Fairly simple. The only exception to this rule are where stats are only measured for a single group (bigs particularly with ORB%, DRB%, BLK% and PF stats but also guards with AST%), then the colours are awarded for best/worst/above average/below average within that group.
Preseason Numbers
The purpose of this post is really to see how these stat clusters look in practice and get some feedback, but it also serves as a fun with SSS post for the preseason. All these numbers should be taken with heapings of salt, because although a lot of it looks relatively close to what we might expect in the regular season, the variance possible when looking a 5, 6 and 7 game sets of numbers is quite extreme. With that said, here they are, presented without comment:
NB: Scroll to the bottom and hit the expand button to see the clusters in their full glory
The Starters
LAMARCUS ALDRIDGE
DAMIAN LILLARD
NICOLAS BATUM
WESLEY MATTHEWS
JJ HICKSON
The Backup Guards
RONNIE PRICE
NOLAN SMITH
COBY KARL
The Backup Wings
VICTOR CLAVER
LUKE BABBITT
SASHA PAVLOVIC
WILL BARTON
ADAM MORRISON
The Backup Bigs
MEYERS LEONARD
JOEL FREELAND
JARED JEFFRIES
If you made it to the point then please let me know what you think of the concept and any ideas for improving it. I plan on expanding these during the season when more comprehensive statistics become available and would quite like to look at them for prior seasons to see if any new insight could be gained. Any help would be invaluable.
Awesome visuals...you musta put a lot of work into this.
Great presentation.... I like it. Interesting to see Leonard, Babbit and Wes looking pretty good based on the visual.
Interesting idea. I wonder if this is any more useful than just a horizontal list (like typical stats), where players can be compared more easily, without looking all over the place. I like the color coding, though.
Also, while I applaud your embrace of the wing concept, I don't know that using different stats by position is of much use. If a big is a good passer, or a guard is a good rebounder, that affects the game, and I want to know about it.
The omission I regret in hindsight is bigs passing. Need at least cursory information on that like for guard rebounding. I looked at using different stats after doing countless traditional tables full of guard shotblocking and fouling stats, which are useful roughly 1% of the time. I don't want to exclude anything informative (again, big man passing) but do you need advanced numbers on guard rebounding or big man passing? If I know what they're doing per 36 that tends to suffice for me but that's the type of info I'm looking for. What I am looking to pare down is the unnecessary detail you get if you just go category by category for every player.
You're probably right about this being less useful for comparing players than sortable basic tables, so maybe the focus should just be on getting it right for an at a glance statistical overview. I think the colour coding is nice and easy to follow but the layout of the clusters was born out of me wanting stuff that related to each other, near each other. It's easy for me to read cos that's obviously how my mind works but the sense I'm getting it that it's confusing for everyone else, which sucks.
@TiMB Whether or not this has appeal probably has a lot to do with how they like information presented. This isn't my style, so I don't get a lot out of it. It could be useful if I wanted a very fast snapshot, though.
thanks for putting this together Tim, it's awesome.
Meyers per 36 minute stats aren't bad.
Very cool concept
I like the concept. Obviously it's a little bit insane to do it for every player and also for per game / percentage / per 36 / rate stats all in one, but for a comparison of 3-5 players with one or two types of stats it'd be readable and interesting.
Yeah, my thoughts were eventually to use it to compare players within each group (guard/wing/big) to each other and perhaps to similar players on other teams. This was just getting the idea out of my head and onto 'paper'. The number of stats used is pretty easy to adjust, these ones were supposed to be a basic template that gives a total overview - sorta like pulling the stuff I look at from BBR (and eventually synergy, hoopdata, 82games etc)into a group then meddling with em. Is it too busy with the number of stats here in your view?
@TiMB it could be very powerful with situational stats for example if you are willing to share your synergy powers.
@christopher_m Just saw this....gives me a good idea for around mid-season, when we have good synergy data.
@TiMB yeah I'd think you want to zero in on 2-3 categories (maybe looking at shooting stats, defensive stats, rebound stats, etc) and 2-3 players. Once a table has a certain amount of data you can't really assume people absorbed it all. Not totally sure what that amount is. But for the purposes of analysis beyond "which player is better" you don't need all the categories anyways right?