Analytics Boxscore ideas

[hoopstudies]

One of the things I've done for a while is build a more analytic boxscore. Basketball-reference has a nice analytical page for a game, as does Cleaning the Glass. I like each of them, but you have to scroll a lot and they almost have too much to be called a "boxscore". I really want one box where the players' names are listed just once.

Mine has evolved over the years and this is how it now looks.



At the top is the final score, as well as a win probability version of the line score - showing the home team's average chance to win by quarter. You can see how San Antonio came back very well in this game. The far right number at the top indicates that 100% of the game was in "Normal" time. There no garbage time here.

I've extended the shot information to have layups, as well as the number of shots that were assisted.

Assists include number of assists for layups, which tend to be higher value.

Pos -- that is the average position the player played in the game. So Paul George went between the 2 and the 3 a lot, it appears.

Miss Orb -- that is the number of missed shots a player had that were rebounded by his team (as team or individual rebounds). It tends to be an indicator of how much help they draw, though definitely imperfect.

FB Pts is incredibly not in traditional boxscores and it is definitely important. The NBA does it officially but there are variations on what is transition vs not. I use their official number.

DNet Pts - is how many points a player adds through their defense. I've done a lot of play-level division of credit and that goes into this number. Danny Green had a good defensive game here.

The right section has individual offensive efficiency, defensive efficiency, possession usage, and the total net points a player added to the game. The old estimates of Basketball on Paper have been replaced by much more accurate estimates that add up (as long as the official data is correct, which it isn't sometimes).

The Four Factors are underneath each team's players - not an ideal placement because it doesn't stand out, but it's useful.

I put this out there to request feedback. Given the constraints of space, which I would like to stick to, are there things you would get rid of and replace with other things? I've always felt like it doesn't have enough defense, which is why I have the DNet Pts field, but it doesn't explain a lot of why.

[hoopstudies]

One other thing not shown here. Under the GS column, I will put an "X" if a player played 50+% of their minutes in garbage time.

[Crow]

I agree with ideal of trying to keep it all on one player line, but it may be a major limitation. You've added a lot of good stuff. Other possibilities: charges drawn, presumed counterpart points (different production view than net def pts), blocks against. Maybe rebound chances.

One thing I've never seen is fast break points against. Total opponent FA pts while on court and / or specifically from the player's presumed counterpart.

Maybe the Vegas line for game to Net SRS plus HCA or some form of expected win% split.

Injured - left game? Blocks recovered? Player level efg% and / or TS%? # stints? Techs? True shot attempts (to try to dissuade the "35 points on 22 fgas" misconstruction)? Day game, back to back markers? Screen assists, touches, passes? Points given up on their specific turnovers? Game pace, time of offensive possession. "Creation %"? "Finishing %"? % of time against "starters" (counterpart or defined as against 4 or more starters?) Expected net counterpart points (to compare against the actual)?
4th quarter or cluchtime points?

Maybe put position to left or right of name. And maybe put name in middle and all of offensive on left, all of defense or right and shorten avg. distance from name to columns. Given that you've added columns and may add more this is an important consideration. I sometimes have to follow from name to column several times to be sure I read it right.

If you want it to be the world's greatest box score by far, it might end up being two lines per player; though maybe you could try putting them one under the other, have a top and bottom legend and alternate slight left and right text alignments for the top and bottom lines (and maybe main level and exponent levels)? Might be worth trying and perhaps with different colors or fonts to further distinguish? Perhaps after you finalize the content you could check with a graphics specialist to make it even clearer, prettier.

[Mike G]

Pretty cool, Dean. It is much better when the whole boxscore, both teams, fits on the screen.
Questions and suggestions:
- Do you know the % of a player's minutes that were Normal (not garbage) time? If so, how about just putting that number in a column, rather than * or x
- The first 5 players listed could be separated by a horizontal line, indicating starters -- and making it easier to visually connect names and numbers.
- Parentheses add some characters to a line. And your assisted FG number is separated from the FG, by the FGA in between. You could alleviate both issues by writing Westbrook's Total, instead of 5-22 (1), as 1/5-22
- - - In fact, unassisted FG might be the more useful stat; in which case, that line could be 4/5-22

Miss Orb sounds like a lady who looks into a crystal ball.
And there's inconsistency in abbreviation formats -- REB vs Ast, for example -- but that's probably not the important stuff.

[Crow]

Using exponents could save significant width for sets of data like fgm-fga if used for lots of sets on same player line. Some of the legend labels could be slight shorter or smaller font and the space between columns could be chopped to fit more. Use of a notably narrow font would help even further (if not taken too far).

[hoopstudies]

Good ideas. Some are easy and I can probably do pretty quickly.

On positions - notice the order of this. I arrange the starters from PG to C, then the subs the same general way. I've never liked the way players are organized in a typical box. This gives implicit position information.

Defensive stats are so often misleading. That's why I ended up choosing my individual defensive metric, which divides credit. Defensive Transition Points Given Up - well, who gave that up? It's not just one player. It's the responsibility of a lot of people. I used to put in the boxscore the player's matchup points scored, but that was misleading, especially when I had my individual defensive metric.

Thanks. I will plan to make some changes.

[nbacouchside]

The right section has individual offensive efficiency, defensive efficiency, possession usage, and the total net points a player added to the game. The old estimates of Basketball on Paper have been replaced by much more accurate estimates that add up (as long as the official data is correct, which it isn't sometimes).

These new estimates are secret I'm assuming? If not, please share!