Counting the Baskets

SportVu Data to Debut

I think Andrew Lynch got everything pretty much right in this post at Hardwood Paroxysm about the release of *some* SportVu data by the NBA.  So, I am going to out source this post to him:

 

It’s assuredly an exciting development, but one would be wise to temper expectations, especially in the early going. The data simply won’t be that robust come Friday, and though the sample size will grow faster than a magic beanstalk as the season progresses, even an entire season’s worth of games isn’t enough to draw hard conclusions.

And that’s the kicker with this technology — it isn’t about answers.

When the data set becomes more thorough, the information will undoubtedly be more useful. But it’s never going to be a magic bullet that will point you to some absolute truth about the NBA. Those who seek answers with SportVU are taking the wrong approach. A system of this nature and magnitude is about asking the right questions. It’s about finding the right perspectives and the right context.

It’s about taking a look at the fallacy of assists and saying that we can do better. It’s about combining ideas like expected points per possession with a tool that can effectively track every pass that happens on the court. From there, with enough data, the brightest minds in our community will calculate the expected value of each of those passes. 

There's a good deal more there, definitely read it all.

For example, one of the new stats expected to debut tomorrow is "rebounds per rebound oppotunity," defined as something like the percent of rebounds a player pulls down of chances within 3.5 feet of them.  That is certainly interesting to know, the question is whether it is more infomative  than our current stats simply counting rebounds or estimating rebound percentages.  After all, it takes skill to get into position to get to the ball, and guys that get to the ball and secure it are already tracked in the current stats.  

It will take considerable analysis before any of the new numbers mean anything, if they ever do.  And that's a good thing. 

Bayesian Expected Win Scores

Last in my quixotic Bayesian rookie expectations series (see here for methodology), unless I use the first ten games for another update...  hmmmmm....

I calculated the Win Score per 48 minutes (old version) for all of the Rooks with significant preseason minutes.  I used Win Score, primarily because it's the easiest to calculate and doesn't differieniate offensive and defensive rebounds, which I didn't have broken out.  See here for the calculation.  It's also a reasonable enough measure of basketball production.

These aren't position adjusted, which favors front court players, so don't read too much into the score of say Len vs Caldwell-Pope.  But the direction column is the interesting bit.  

Because some preseason stats are more stable, and therefore, more reliable the updated Win Scores may not move as much as the preseason score would seem to indicate. Or even in the right direction, Victor Oladipo struggled in the areas that are more variable, like turnovers and eFG%, and therefore make less of an impact on expectations going forward, while improving in the areas that should be less variable, like assists and rebounds. So his Updated Expected Win Score is higher than either his Prior Expected Win Score or Preseason.  Though he also struggled with shooting last night.

Big positive movers are Michael Carter-Williams, Steven Adams, Tony Snell, Tony Mitchell and the Celtics rookies Phil Pressey and  Vitor Faverani, who experienced the biggest rise.

Big negative movers were Andre Roberson, Anthony Bennett, Peyton Siva, and, just barely Kelly Olynyk, who needs to figure how defend without fouling.

Player	Team	Pos	Preseason	Prior Win Score	Updated Win Score	Update Diff
Alex Len	PHX	C	5.59	8.41	7.64	-0.77
Andre Roberson	OKC	SG	0.92	10.77	6.86	-3.91
Anthony Bennett	CLE	PF	1.39	9.68	6.55	-3.13
Archie Goodwin	PHX	SG	-1.75	-1.39	-2.01	-0.62
Ben McLemore	SAC	SG	8.27	6.12	6.67	0.56
Cody Zeller	CHA	PF	8.01	9.7	9.53	-0.17
Erik Murphy	CHI	PF	1.61	6.96	6.01	-0.95
Jamaal Franklin	MEM	SG	8.49	7.13	6.92	-0.21
Kelly Olynyk	BOS	PF	8.14	10.01	8.85	-1.16
Kentavious Caldwell-Pope	DET	SG	4.45	6.82	6.47	-0.35
Mason Plumlee	BRK	PF	11.91	10.41	11.37	0.96
Michael Carter-Williams	PHI	PG	7.46	2.74	4.57	1.83
Peyton Siva	DET	PG	0.47	3.3	1.99	-1.32
Solomon Hill	IND	SF	0.59	4.34	3.67	-0.66
Steven Adams	OKC	C	14.5	9.06	10.54	1.48
Tim Hardaway Jr.	NYK	SG	3.35	1.06	1.57	0.51
Tony Mitchell	DET	PF	13.72	6.28	8.41	2.14
Tony Snell	CHI	SG	4.98	0.8	2.95	2.16
Victor Oladipo	ORL	PG	9.8	12.69	13.15	0.46
Vitor Faverani	BOS	C	13.11	6.65	9.1	2.45
Phil Pressey	BOS	PG	7.03	1.64	3.76	2.12

Updating my Updated Expectations

(Qué Esperar cuando estás Esperando El Hombre Indestructible)

I have been working with some rookie preseason number, both how the correlate to regular season numbers and how to incorporate them into my expectations for their rookie campaign.

Go see my piece here at Hickory-High to get the methodology.  Long story short, different box score measures were affected more than others, ie stats showing consistency like rebounds effected my expectations than inconsistent measures like scoring efficiency.

In any case, I just wanted to focus on the three Celtics rooks after my expectations were updated.

First the Celtics rookies' per 48 minutes.  The 'Prior' columns are the expected performance based on their college or Euro numbers, and the Updated columns are the Updated expectations adding in their preseason performance.  Vitor Faverani is the big mover here, in part that could be because my methodology was developed for NCAA players.  Faverani is coming out of the second pro league in the world, so his numbers may not fall as much as my model initially estimated.  Pressey also out performed his expectations, particularly rebounding and assists.

Olynyk was close to expectations, but personal fouls look like an issue and to a lesser extent rebounding dipped below expectations.

Player	Prior PF	Updated PF	Prior Reb	Updated Reb	Prior Asts	Updated Asts	Prior Stl	Updated Steals	Prior Blk	Updated Blocks	Prior eFG%	Updated eFG%
Kelly Olynyk	5.63	6.82	10.98	10.14	2.9	3.39	0.95	1.03	1.42	1.35	56.3%	56.11%
Vitor Faverani	7.03	6.41	10.18	11.85	1.9	1.15	0.65	0.66	1.32	2.59	48.1%	49.19%
Phil Pressey	4.33	5.35	3.88	5.34	8.4	9.31	2.05	2.14	-	0.22	34%	34.52%

 

Then, since everyone loves one number metrics, I calculated the simplest, Win Score (old) for each rookie using their expected production and the updated expectations after preseason, assuming they shoot with the same volume per minute in the regular season that they did in the preseason.

Player	Team	Pos	Preseason	Prior Win Score	Updated Win Score	Update Diff
Kelly Olynyk	BOS	PF	8.14	10.01	8.85	-1.16
Vitor Faverani	BOS	C	13.11	6.65	9.1	2.45
Phil Pressey	BOS	PG	7.03	1.64	3.76	2.12

 

Data from RealGM for preseason and DraftExpress for college and Euro

Rebounds Revisited with Shot Type

A couple of weeks ago I wrote a piece featuring MarShon Brooks and the value of getting shots near the rim. One of the effects I highlighted, in addition to the greater expected effective field goal percentage, was the increased likelihood of getting offensive rebounds. Based on a Kirk Goldsberry's Court Vision Analytics visualization I spit balled the following table on expected offensive rebounds and estimated change in shot value compared to an estimated average of 32% offensive rebounds.

Shot Location	HoopData FG%	Expected Offiensive Rebounds	Rebound Differential	Expected Marginal Shot Value
Rim	64.60%	40%	8%	70.22%
Short (3-9)	39.70%	35%	3%	40.93%
Mid (10-15)	41.80%	31%	-1%	41.39%
Long Two (16-23)	38.30%	28%	-4%	36.83%
Three	35.93%	28%	-4%	52.51%

 

I decided to follow up a bit looking at some data from NBA Wowy. Wowy doesn't have rebounds by shots location exactly but it does have them by prior shot type, which overlaps with shot location, using Mid two point jumpers, Lay Ups, Hooks, Tips, Free Throws and Three Pointers. Below, with a lot of copying and pasting, I put together Wowy's offensive and defensive rebounds last year for each team last year for those shot types.

SIDE	MID	3PT	LAY	HOOK	TIP	FTA
OFF	25.5%	24.5%	36.8%	21.8%	41.8%	11.9%
DEF	74.5%	75.5%	63.2%	78.2%	58.2%	88.1%

 

The total percentage of offensive rebounds was 27.4% from the field, much lower than Goldsberry's 32% estimate, but matching the numbers from Basketball Reference.  The percent of rebounds shot at the rim, using Lay Ups and Tips, comes out lower than my 'eye ball test'  at 37.3% but the differential to the average is larger nearly a full ten percent.

It's interesting that Hook shots are so much lower in offensive rebounding percentage.  It's tempting to attribute that to noise, given the relative rarity of hook shots these days at only three percent of rebound opportunities, but even with that there were over 3,000 rebounds of hook shot last year.

A hook shot is almost by definition one that has a defender between the offensive player and the basket (otherwise he would take a lay up or dunk) and nearly exclusively by big men who often get taken out of position by the hook motion.  My shot location cluster analysis unsurprisingly found a connection between shooting from nearer the basket and getting offensive rebounds.  And unlike driving lay up attempts there is much less likelihood to force help rotations from the defense, which puts the offense in much better positions to grab the rebound.

Those, of course, are factors teams will able to better quantify now that all of the teams have SportVu.

 

Below the fold offensive rebounds broken down by team:

Create Your Own Rookie Preseason Visualization Adventure

Preseason is, finally, drawing to a close.  So I think a fair question is what, if anything, we might have learned from it.  An 82games study a while ago indicated a 40% correlation between preseason win percentage and regular season.  (Yikes, for us Celtics fans, though they didn't do a point differential, which should be a slightly better predictor and would have the Celtics at -3.57 point per game projected at 31 wins rather than 11 wins. Thanks Knicks bench!).  The current batch of advanced stats can already do better than that, however.

For most players who have been in the league a few years, especially on the same team, adding a couple of exhibition games of data shouldn't change our assessment of their abilities much at all.

There are some exceptions where there is a significant discontinuity and preseason gives us significant new information, like Derrick Rose coming off of injury or, to a lesser extent, a player on a new team with a new role.  But, I think rookies' performance is probably the most informative and interesting preseason data since this is the first time we have seen them play against NBA competition.

With that in mind I used data from RealGM on rookies for the last three years in both preseason and regular season to create the Tableau visualization below.  I kept things pretty simple looking at simple linear box score metrics Alt Wins and Win Score (new). Also Rebounds, Assists, field goal percentage and turnovers. The highlights and my inferences:

Once you get to 50 minutes most measures are starting to give you statistically valid information, with this set of data 70 to 80 minutes in the preseason seemed to be the 'sweet spot' where the correlations were high(ish) without dropping too many players.  

Win Score correlated better than Alt Win score as far as correlation between pre-season and regular season, with a Goodness of Fit score of .39 at my favored minute filters.  I suspect the higher correlation is mostly because of the heavier weight on rebounds since I didn't apply a position adjustment.

Looking at the stats scaled to per 48 minutes, rebounds and blocks are the most predictive stats, followed by assists,  But all are somewhat positionally influenced (as you can see on that tab).  For rebounds the R2 drops from .75 to .47 when only looking at Bigs.  For assists it dropped from .78 for all positions to .68 only including Guards and .12 for Point Guards.

Field goal percentage is much more variable given the amount of minutes/shots the rookies got and the variable nature of shooting.  Even free throws had little predictive power.

Steals, turn overs and personal fouls were all in middle ground.  In bad news for Kelly Olynyk fans the fouls committed R2 was .58, indicating some shorter outings for the big guy.

All the correlations look pretty similar to what I have seen others report from college to pros, with exception of shooting because of the small preseason sample.  The interesting questions will be what happens when the two data sets contradict each other.

The next step will be to apply this information to this year's batch of rookies once the preseason is done.

On the visualization minutes on both preseason and regular season are adjustable, as are positions.  Tab over to the various stats.

 

Learn About Tableau

Green Means Go Shoot There

 

Applying Scoring Versatility Index to the Celtics Roster

I wrote an article over at Hickory High detailing a new scoring metric I created called Scoring Versatility Index. The idea of the index is to track how efficient a player is at scoring at different points across the court and at getting to the foul line. Being efficient in different spots should make a player harder to guard on defense and give the offense more flexibility.

In that article I pulled out some of the most versatile scorers lead by Kevin Durant and LeBron James, naturally. Here I am going to apply it to the Celtics roster.

The index uses the shot location data from HoopData, each player is scored by their efficiency relative to the average player's efficiency from that zone plus their ability to get to the line and make free throws per field goal attempt. A score of over one indicated that the player is above average in efficiency at that location.

 

Player	Total FGA	At Rim Ratio	Short Ratio (3-9)	Mid Ratio (10-15)	Long Two Ratio (16-23)	Three Ratio	FTM/FGA Ratio	Versatility Index
Avery Bradley	478	0.83	0.95	0.88	1.26	1.14	0.41	0.91
Brandon Bass	589	1.02	0.99	1.16	1.35	0.00	1.15	0.94
Courtney Lee	530	1.06	0.80	0.96	1.32	1.32	0.62	1.01
Gerald Wallace	458	0.89	1.07	0.00	0.82	1.00	1.39	0.86
Jared Sullinger	225	1.11	0.60	1.62	0.97	0.71	1.10	1.02
Jeff Green	806	1.07	0.86	1.08	1.05	1.35	1.40	1.13
Jordan Crawford	725	0.97	1.46	1.33	1.13	1.18	0.94	1.17
Keith Bogans	284	1.21	1.71	0.00	1.35	1.20	0.20	0.79
Kris Humphries	310	0.92	0.97	0.83	1.00	0.00	1.65	0.89
MarShon Brooks	335	1.10	1.30	1.16	0.91	0.95	1.08	1.08
Rajon Rondo	465	1.00	1.17	1.00	1.40	0.85	0.68	1.02
Grand Total	494	1.01	1.11	0.95	1.14	0.91	0.96	1.00

 

Jordan Crawford and Jeff Green scored as the highest on the Index.  Green was above average nearly everywhere and Crawford was at least near average (though this once again demonstrates the importance of shot selection).  On the other hand Bogans complete lack of a mid-range game gave him the lowest VI score.  And we can only hope the shooting coach Humphries worked with over the summer pays off.

At this point, I consider the Scoring Versatility Index to be a version 1.0 and may make some changes based on the feedback I have already gotten for a version 2.0.  Over the weekend developing the Index one of the versions I looked at was an eFG Versatility that used the NBA average Effective Field goal percentage as the basis for evaluation of each player's efficiency in each zone.  It did not make as much sense for free throw attempts, so there I stuck with using the average player's efficiency as the divisor.

I thought looking at eFG Versatility for the Celtics roster would be pretty interesting, both as a tool to evaluate the roster as well as to demonstrate the importance of shot selection.

I added color coding, which should be pretty self explanatory, green means go ahead and shoot. As I always like to write, observant reader may notice a pattern in the column towards the right.

	Total FGA	At Rim Ratio	Short Ratio (3-9)	Mid Ratio (10-15)	Long Two Ratio (16-23)	Three Ratio	FTM/FGA Ratio	eFG Versatility Index
Avery Bradley	478	1.03	0.67	0.67	0.86	0.97	0.41	0.77
Brandon Bass	589	1.28	0.70	0.89	0.92	0.00	1.15	0.82
Courtney Lee	530	1.32	0.56	0.73	0.90	1.12	0.62	0.88
Gerald Wallace	458	1.11	0.75	0.00	0.56	0.85	1.39	0.78
Jared Sullinger	225	1.38	0.42	1.23	0.66	0.60	1.10	0.90
Jeff Green	806	1.33	0.61	0.82	0.72	1.15	1.40	1.01
Jordan Crawford	725	1.20	1.03	1.02	0.77	1.01	0.94	0.99
Keith Bogans	284	1.51	1.20	0.00	0.92	1.02	0.20	0.68
Kris Humphries	310	1.15	0.68	0.63	0.68	0.00	1.65	0.80
MarShon Brooks	335	1.38	0.91	0.89	0.62	0.81	1.08	0.95
Rajon Rondo	465	1.25	0.83	0.76	0.96	0.72	0.68	0.87
Grand Total	494	1.26	0.78	0.72	0.78	0.77	0.96	0.87

 

Looking at the Three's column, spacing from Green and Lee will be key, as well as Bradley regaining his three point form.

Below are the rookies using the eFG Versatility Index, with Kelly Olynyk and Phil Pressey's data coming via Hoopmath and Vitor Faverani's coming from Draft Express.

	Total FGA	At Rim Ratio		Three Ratio	FTM/FGA Ratio	eFG Versalitiy Index
Kelly Olynyk	340	1.47	1.04	0.90	2.05	1.37
Phil Pressey	376	0.90	0.76	0.96	1.05	0.92
	Total FGA	Two Point Ratio		Three Ratio	FTM/FGA Ratio	eFG Versalitiy Index
Vitor Faverani	150	1.19		0.57	1.79	0.97

 

Given the lower level of competition one has to take these numbers with some salt.  But, Faverani's 19% from three makes me wonder about Steven's plan to turn him into a 'Stretch Center,' given that he has only hit over 30% once in his Spanish career.  Also, even though Olynyk can hit from range, his ability to score near the basket is something to exploit.

The Celtics are going into this year without a light out scorer, something that wont change once Rondo comes back. That leads us right back to analytic fundamentals, open shots and shot selection.

Where Shooting Fouls Happen (And to Whom)

Last post I gave a spit balled estimate of the expected offensive rebound percentage by shot location, based on some courtvisionanalytics research.  A commenter mentioned the differential in getting to the line by shot location as another important piece of expected offensive value for each shot.

Unfortunately, it doesn't appear that shooting fouls are tracked that way in the play by play data.  The NBA feeds simply list shooting foul with the fouler and foulee.  I did, however, find a previous study at 82games that employed trackers to supplement their data that, among other things tracked where a shot was being attempted when the foul occurred.

 

Below is the relevant portion of the 82games.com end of possession table:

Zones	FTA/FGA	Shooting Foul	Foul
Corner 3's	0.04	0%	2%
Wing 3's	0.08	1%	9%
Straight-up 3's	0.1	1%	13%
Baseline 2's	0.13	3%	13%
Wing 2's	0.11	2%	11%
Straight-up 2's	0.2	2%	8%
High Paint	0.45	15%	22%
Low Paint	0.46	19%	23%

 

They found that 19% percent of plays ending in the 'Low Paint" led to shooting fouls compared to 1% for three point shots, and .46 free throws for every shot taken.  Apparently everyone knows 'Never Foul a Jump Shooter.'

Again, free throws are a very effective way to score.  In fact, every player that can muster an over 50% free throw percentage, or most players not named Andre Drummond, is beating the NBA average points per possession.

Before I found the 82games study, I had decided to run a few regression analyses on free throw attempt to field goal attempt ratio.  I ran a series of step wise regressions analysis using data from Hoop Data. I analyzed the data with all players and by positions,  using percent of shots from each location as well as percent of two point shots assisted, which I have used as my shot creator indicator in the past.

Unsurprisingly, percent of shots at the rim came out as a statistically significant predictor in every configuration. (Three Point shots have a strong enough negative correlation to shots at the rim to be confounded in the analysis, showing up only if percent shots at the rim was eliminated).

Percent of two point shots assisted came in as an independent predictor with a negative correlation to getting to the the line overall, for Guards and for Wings, but not Bigs.  Driving with the ball gets you the line more than cutting or spotting up.

For the overall analysis the two variables Percent of Shots at the Rim and Percent of Two Point Shots Assisted were enough to give me a .416 R2, or Goodness of Fit, for all players with more than two hundred total minutes who were not named Jason Collins.  (Collins absolute refusal to launch the ball towards the basket made him a such an outlier  I had to pull his numbers out of the analysis).

The other outliers were pretty interesting too.  Using the overall analysis here are the biggest under performers and over performers at drawing shooting fouls.

Name	Std. Residual	FTFG	Predicted Value
Kosta Koufos	-2.252	20.5%	42.91%
Cole Aldrich	-2.101	25%	45.94%
Marquis Teague	-2.06	14.2%	34.69%
Name	Std. Residual	FTFG	Predicted Value
Kevin Durant	2.885	52.3%	23.59%
James Harden	2.98	59.2%	29.54%
Jeff Adrien	3.2	59.5%	27.64%
Tyler Hansbrough	3.865	73.7%	35.20%
Ryan Hollins	4.112	77.2%	36.22%
Corey Maggette	4.383	63.2%	19.49%
Dwight Howard	4.704	88.7%	41.81%

 

For some reason I think Koufos will draw more fouls in the Grind house than in Denver, just seems like it will be expected.  Durant and Harden certainly get the calls, as did Corey Maggette.  I suspect opponents just couldn't resist taking a swipe at Hansbrough.  Then there's Hack a Howard leading everybody.  Feels like Houston games are going to be long.

Take it to the Hoop, a Free MarShon Brooks Post

Welcome to 2013-2014's first official Free MarShon Brooks post.

To get this out of the way, in terms of efficiency so far Brooks has been middling in two years with Brooklyn. His player metrics overall show him better than replacement, but not quite average in terms of production. But, with the exception of Rondo, and to a lesser extent Lee I have just described the entire Celtics guard rotation. Brooks is also one of the Celtics guards in the stage of his career where we can reasonably expect to see some improvement in those numbers.

In addition, Brooks does have some identifiable skills that make him intriguing. In my previous cluster analysis Brooks scored out as a Slashing Guard and a Shot Creator. Brooks can get to the rim and score effectively.

	Attempts at Rim per 48 minutes	Pct Assisted at Rim	Field Goal Pct at Rim
Avery Bradley-BOS-2013	4.57	46.40	51.50
Courtney Lee-BOS-2013	4.17	53.60	65.90
Jordan Crawford-BOS-2013	4.34	37.90	58.00
Keith Bogans-BKN-2013	0.68	46.70	75.00
Marshon Brooks-BKN-2013	5.89	36.70	68.70
Rajon Rondo-BOS-2013	6.14	15.90	62.10
Average	4.30	39.53	63.53

Even though I am working on research trying to estimate a value for stretch in the offense, and the NBA is gaga over three point shots, there's a huge value to getting to the rim. (Of course, getting to the rim and burying threes are complimentary. Players drive to kick the ball out and space floor to open lanes to the basket).

But, overall there is more value to getting to the basket than the three ball, at least on any individual shot attempt. Shots at the basket have a higher effective field goal percentage, generate more free throws and are more likely to lead to offensive rebounds than shots from further away. 

Here's a great visualization of the rebound effect from Kirk Goldsberry using an analysis with SportsVu video tracking.

via courtvisionanalytics.com

 

I squinted really hard at the image, since I haven't found any data representation on-line.  Then I combined the Hoopdata shot location data and the Kirk Goldsberry rebounding effect to spitball the value of a shot at each range with the rebounding effect included. Since shots near the rim are 8% more like to be rebounded by the offense than the average of 32%, I added the differential to the expected effective field goal percentage to get what I am calling the Expected Offense Value.

Shot Location	HoopData FG%	Expected Offensive Rebounds	Rebound Differential	Expected Offense Value
Rim	64.60%	40%	8%	70.22%
Short (3-9)	39.70%	35%	3%	40.93%
Mid (10-15)	41.80%	31%	-1%	41.39%
Long Two (16-23)	38.30%	28%	-4%	36.83%
Three	35.93%	28%	-4%	52.51%

 

I could go even further, because as every coach preaches and the data shows, making your shots helps the defense. According to numbers from Evan Zamir last year, effective field goal percentage by the opposing team after a made shot was about 3.5% lower than after a defensive rebound (47.1% vs 50.6%), because the defense had time to get back.

This relates back to finding time for MarShon Brooks because he excels at the game's most efficient shot. Leading up to training camp some of his new team mates commented that Brooks was the best one on one player, I imagine it was because of his ability to get to the bucket.

The percentiles compare Brook's number to all other guards in the NBA over the last three years where he was in the top 80% percentile for each according to HoopData.

	Pct Shots at Rim	Percentile Shots at Rim	Attempts at the Rim per 48 min	Percentile in Att at Rim per 48 min	Field Goal Pct at Rim	Percentile Rim FG%
Marshon Brooks	35.60%	82.50%	5.89	82.60%	68.70	87.7%

 

Of course, there are reason's you haven't seen Brooks in any All Star games. By all accounts his defense needs work, though his blocks and steals numbers aren't terrible due to his long wing span. But, I don't want to know how a Crawford/Brooks back court would do on the defensive end over an extended period.

Also, Brooks' stretch game is weak. He shot 27% from three last year, which put him in the 23 percentile over the last three years, though that was down from 31% the year before. It's hard to imagine too much time with a Rondo/Brooks back court if those numbers from distance don't improve.

The good news is that there is some reason to believe that Brooks can improve in his weak areas. One is simply age and NBA experience, Brooks is still on the right side of the growth curve. The other is that he did shoot better from three the year before at 31%, and his 73.4% free throw shooting indicates a decent stroke.  And with his excellent attacking ability Brooks doesn't need to be Danny Green, he just needs make opponents respect his outside shot.

I think Brooks could pair nicely with Lee or Bradley, if Bradley's shot recovers.  And to a lesser extent Phil Pressey.  As long as he's clear why he's in the game, to take it to the basket and kick it out when he can't.

Visualized Coaching

'Cause nothin' beats a scatterplot, or six.

(Update): After playing around with it a little there are some interesting differences by team.  I am glad I kept that filter, the metrics all love how the Spurs and Hawks assigned time.  Philly had the only negative slope I could find with Win Shares, meaning Win Shares thought Doug Collins was slightly more likely to play bad players than good.

 

 

Learn About Tableau

Waiting for Rondo, Balancing the Celtics Line Up

Tossing aside my milquetoast-y luke warm-ish semi-endorsement of starting Jordan Crawford at the point, like so much warm milquetoast, Brad Stevens has, more or less, come out in favor of starting Avery Bradley at the Point. My actual case was that Bradley is ill suited to be a Point Guard and is playing out of position whenever he's stationed there.

In contrast, Stevens said a bunch of things which are true, but address completely different points, via Celtics Hub:

Stevens said Avery Bradley may indeed see the bulk of the point guard duties until Rondo finds his way back from offseason knee surgery.

 

"I don’t think there is any doubt that Avery has elite ability in a lot of ways as a point guard," Stevens said at TD Garden yesterday morning, where he was a guest at the breakfast to promote November’s Coaches vs. Cancer college basketball tripleheader. "He’s an elite defender at the position. He’s an elite athlete at the point guard position. I think he’s a guy that’s gotten better. I think he’s a guy with more confidence, and I think he’s excited about the challenge if Rajon is out."

 

No doubt Avery Bradley is an elite defender, an elite athlete, and a fierce competitor excited for the opportunity to play while Rondo is out. None of these things mean that he has a NBA point guard handle or passing ability.

In my last post I closed with this table breaking down the attributes I thought one should weigh and balance in setting up a back court rotation, which I would like to discuss a bit more here.

 

Creators	Stretch	Rim Attack	Defense
Crawford	Lee	Brooks	Bradley
Pressey	Crawford	Lee	Lee
Brooks	Bogans	Crawford	Bogans
Bradley	Bradley	Bradley	Pressey
Lee	Pressey	Pressey	Crawford
Bogans	Brooks	Bogans	Brooks

 

On Celtics media day Stevens specifically mentioned two items on the table above; defense (though I think NBA coaches get fined if they fail to prominently mention the importance of defense, it's called the D'Antoni Rule) and spacing.

 

On defense there's no question Bradley is the Celtic's best back court option, especially on ball defense. Courtney Lee is likely a fairly distant second. This is backed up by the defensive xRAPM numbers for the Celtic's guards.

 

Name	Def. xRAPM
Avery Bradley	-0.25
Rajon Rondo	-0.65
Courtney Lee	-1.38
Keith Bogans	-1.73
Jordan Crawford	-2.13
MarShon Brooks	-2.3

 

In terms of spacing, Lee is the top Celtic guard, with a career 38.4% mark from three point land and borderline 'Stretch' score in terms of shot location. Bradley had a down shooting season as did Marshon Brooks, though Bradley's down year three point percentage matched Brook's up year. The prudent early estimate would be to bracket the number between the two years for both, which would put Bradley as a mid-thirties three shooter, certainly enough to make defenses concerned about him out on the line.

 

One thing to add in terms of ability to get to the rim among Celtic's guards.  I am not sure how this play's into Stevens' line up decisions, but getting to the basket generates some of the most efficient shots and can cause break downs for defense. Below I have my stats on the Celtics guards proficiency getting to the basket.  It also shows a little something about new Celtic Marshon Brook's offensive game, which revolves around his ability to get to the rim and score often and effectively.

 

	Attempts at Rim per 48 minutes	Pct Assisted at Rim	Field Goal Pct at Rim
Avery Bradley-BOS-2013	4.57	46.40	51.50
Courtney Lee-BOS-2013	4.17	53.60	65.90
Jordan Crawford-BOS-2013	4.34	37.90	58.00
Keith Bogans-BKN-2013	0.68	46.70	75.00
Marshon Brooks-BKN-2013	5.89	36.70	68.70
Rajon Rondo-BOS-2013	6.14	15.90	62.10
Average	4.30	39.53	63.53

 

Stevens has stated that he wants the Celtics to run this year, which may be a necessity to generate points.  But, as much as all teams want to run, someone still has to handle the ball in the half court, since every defense is intent on stopping you from running. Here's the issue with the Bradley/Lee back court, they're the two of least proficient guard Shot Creators the C's have. Of course, there's no rule the back court must handle the ball, as we saw last year with Paul Pierce and that guy in Miami.

 

With that in mind, we can look at the ball handling numbers for the Celtics front court.

 

Player- Team- Yr	Pct Two Point Offense Assisted	Total Field Offense Assisted	AST Pct	TOV Pct
Brandon Bass-BOS-2013	73.60%	73.60	6	10.90%
Gerald Wallace-BKN-2013	63.12%	74.25	13.3	16.90%
Jared Sullinger-BOS-2013	55.05%	55.66	6.2	10.30%
Jeff Green-BOS-2013	50.33%	60.02	9.6	12.50%
Kris Humphries-BKN-2013	72.81%	72.81	4.3	13.50%
Kelly Olynyk-Gonzaga-2013	54.59%	62.68	15.10	15.40%
Vitor Faverani-Valencia-2013			8.29	20.46%

Most of the true bigs don't look like much help handling the ball. Humphries beats No Pass Bass in a race to the bottom on assist percentage. Not to put too fine a point on it, a Humphries/Bass/Green/Lee/Bradley line up has the potential to be shut down as a breeding ground for mosquitoes and have the lowest assist to turnover percentages of any starting NBA line up.

Vitor Faverani's painful tun over percentage comes via Basketball RealGM. I had heard Faverani had turnover issues, but it looks like his time on the court is going to be largely determined by his ability to keep track of the ball.

Green scored the best last year at being able to create shots, with Sullinger (off of offensive rebounds?) and Olynyk's college numbers right behind. I actually think Wallace's Creation numbers are a bit deceptive, his assisted shot percentages were lower in 2012 and, according to NBA WoWy his percentage of shots assisted numbers plunged in the, relatively few, possessions he was on the court without Deron Williams.

In terms of passing, Olynyk, the former Point Guard, is clearly the most interesting front court player on the roster. His 15.1% assist percentage was among the highest in college ball last year for Bigs, and puts him at the top of the Boston front court players.

Ok, long story short. The best Celtic's guard tandem in terms of Defense and Spacing, is also the weakest in terms of ball handling duties. That creates a conundrum, if the coaching staff is going to start them together Lee and Bradley will need help in the half court.