Showing posts with label Statistics. Show all posts
Showing posts with label Statistics. Show all posts

Saturday, April 4, 2015

Basketball Isn't Easy - The 2014 NBA Finals

1 Spurs vs 2 Heat
Predicted Rockets over Pacers in 6 (That looks great now..). Heat in 5 (I needed a high variance pick in  the WoW contest, otherwise I likely would have said Heat in 6.). and the actual result was Spurs in a very convincing 5.

The NBA Finals
First of all, let's give it up for the parity in the NBA.  Two seasons ago likely the two best teams in the NBA went head to head and last season, again the two best teams likely went head to head.  /golf clap   Moving along now...

The Spurs scored fewer than 100 points only six times in the entire playoffs and three of those games were against the Mavericks.



Total Points Total Points Per Game Average


Expected Result Actual Result ER AR

Miami Heat 468.7 458 93.74 91.6

San Antonio Spurs 513.7 527 102.74 105.4

Difference pro Spurs 45 69 9 13.8







If Wade and Chalmers had their playing time cut in half  and given to more competent teammates.



Previous point total including Wade and Chalmers
69.05


New point total
86.93 (I am uncertain if I did this part entirely correctly or not)

Net gain of approximately 18 points for Heat
17.87








Pure Heat Expected Total 720.36



Pure Spurs Expected Total 744.70



Differential 24.00




It would be kind of interesting to see which players performed at greater and lesser levels as opposed to last season's finals, but I am fairly happy with the amount of content that I have for this year's finals, and it would likely take a fair while to peruse through the differences between the two series.   It would also be interesting to take a look as to how I would attempt to beat the Heat or the Spurs if I was an opposing coach.  However, due to time constraints, that will not be happening at the moment either.

There are still a fair amount of improvements and tweaks that can be made to the data sets and statistics that I put gather and put forth, but, overall, I was happy with the level of improvement that I made from last season's NBA Finals.


SAN ANTONIO SPURSTotal Expected PointsExpected points per shotActual pointsActual points per shot'Overachiever'Offensive UsageQuality Shot %



















Belinelli19.731.2320.001.251.01.05.88

Bonner 1.301.302.002.001.54.001.00

Diaw33.491.1225.00.83.75.08.77

Duncan51.601.1754.001.231.05.12.84

Ginobli54.081.1558.001.231.07.13.77

Green41.071.2143.001.261.05.10.85

Leonard60.041.2069.001.381.15.14.86

Parker72.141.0278.001.101.08.20.62

Mills39.921.1151.001.421.28.10.72

Splitter26.011.3027.001.351.04.061.00

Random Spurs7.011.176.001.00.86.02.83




























Team Totals406.391.14433.001.221.071.00.78










One quick note, Bonner only ended up with .01 total usage, so he shouldn't have even been charted as an individual player (His data sets should have been put in the Random Spurs data sets.) since he would provide very unreliable data.


I find it extremely impressive that every Spurs' player happens to have an expected pps of over 1.  And then, of course, their actual pps is even more impressive with the average random shot by a Spurs' player resulting in an average point result of 1.22.  Every main Spurs' player overachieved, except for Diaw.  On the whole, the Spurs' offense overachieved by about 7%.



Miami Heat
Total Expected Points
Expected points per shotActual pointsActual points per shot'Overachiever'Offensive 
Usage
Quality Shot %

Allen46.861.2341.001.08.87.11.89

Anderson8.301.196.00.86.72.02.86

Battier2.041.02.00.00.00.01.50

Bosh61.851.2158.001.14.94.15.88

Chalmers23.351.1115.00.71.64.06.76

Cole21.211.1215.00.79.71.06.74

James108.011.24110.001.261.02.26.92

Lewis38.631.2939.001.301.01.09.97

Wade65.891.0654.00.87.82.18.69

Jones5.741.1511.002.201.92.01.80

RR19.611.1512.00.71.61.05.76



















Team Totals401.491.18361.001.06.901.00.84










Miami, on the other hand, underachieved by around 10% and scored .12 fewer points per shot than they should have over the course of the series (which I would think is a fairly uncommon level of under-achievement over the course of a series).  For those with serious playing minutes, the Heat only had Lebron and Lewis hit approximately equal to what they should have hit, every other player underachieved the levels they should have performed at.  The Heat won the quality shot battle again this year, 84% to 78%.


SAN ANTONIO SPURS.Defensive RD# UsageTotal Usage

Greater than 1 means a better 
defender than average.






Belinelli.91.03.04
Bonner1.13.02.01
Diaw.97.11.10
Duncan1.05.20.16
Ginobli.82.09.11
Green1.24.12.11
Leonard1.18.20.17
Parker.98.09.14
Mills1.04.05.07
Splitter1.12.10.08
Random Spurs.78.01.01












Team Totals1.061.001.00

Green and Leonard were quite impressive in their levels of defensive quality, had a high usage rate, and helped to defend almost a third of the Heat's shot attempts.  One of the aspects of basketball that I have not find a way to properly integrate is good passing/assists, as such, Diaw has a fairly bad rating for the series, but his passing was, at times, fairly integral to the Spurs' offense.

Miami HeatDefensive RatingDefensive UsageTotal Usage
Allen.91.10.11
Anderson1.17.09.05
Battier.97.04.02
Bosh1.16.14.15
Chalmers.95.10.08
Cole.86.07.06
James.99.12.19
Lewis1.06.09.09
Wade.82.14.16
Jones.78.01.01
RH.93.10.07








Team Totals.981.001.00

I was a little surprised to see Allen struggle so much defensively, I thought that he was generally closer to an average defender.  It was also a bit surprising to see that over the course of the series Lebron was a slightly below average defender.  If it had not been for Bosh, Anderson, and Lewis putting up decent defensive ratings and their combined defensive usage of .32, then Miami would have had a completely porous defense.. Instead the Heat only had a mostly porous defense.


SAN ANTONIO SPURS(Actual Ortg and Drtg combined with modulated 'prorated' effects of each)(Expected Ortg and Actual Drtg combined with modulated 'prorated' effects of each)(MVP vote) [And prorated]

Total Expected Value (TEV)Total Actual Value (TAV)Impact on SeriesDouble Impact
Belinelli1.121.13.04.03
Bonner1.151.24.01.01
Diaw1.04.91.09.06
Duncan1.101.12.18.16
Ginobli1.021.07.12.10
Green1.221.25.13.13
Leonard1.191.27.21.21
Parker1.001.06.15.13
Mills1.091.30.10.10
Splitter1.191.20.09.09
Random Spurs

.00.00















Team Totals1.101.141.121.00

San Antonio outperformed their expected play by approximately 4%.  Having an entire team that has a TEV equal to or above 1.00 is quite impressive.  And then most of the Spurs' player's TAV numbers are somewhat ridiculous.  Green, Leonard, Mills, and Splitter had a TAV in between 1.20 to 1.30, having a TAV at 1.10 is slightly above average, but having one above 1.20 is impressive.

The way that I selected my MVP award this year (Though the proper selection is kind of obvious both here and in the real world.) is =(CX136*CR136*CX136)/1.3...  Or at least that is what the formula looks like (For Belinelli anyways)..  No, actually though, those numbers simply just represent total usage multiplied by actual value^2 which is then all prorated, so in this case that simply means that formula represents (Total Usage*Actual Value^2)/1.3, or in base form, (TU*AV^2)/U (where U stands for an unknown number).

Leonard carries the MVP vote away comfortably, and it would have been a landslide victory for him had it not been for Duncan.  Green and Mils could have been more serious MVP contenders had their usage rates been higher.





(MVP vote) [And prorated]
Miami Heat
TEVTAVImpact on SeriesDouble Impact
Allen1.081.00.11.10
Anderson1.171.11.06.06
Battier.97.83.02.01
Bosh1.191.15.17.18
Chalmers1.01.86.07.06
Cole.97.83.05.04
James1.161.18.22.24
Lewis1.171.18.11.12
Wade.96.85.14.11
Jones1.011.67.02.03
RR1.01.85.06.05










Team Totals1.081.021.021.00


Meanwhile, the Heat underperformed the level that they should have achieved by nearly 6%.  On the bright side, Cole, Chalmers, and Wade all should have performed semi-significantly better, on the downside, even with that improvement they still should have been playing very few minutes in the series.

I almost think that I should go with an MVP value of (TU*AV^3)/U because I do not like that Wade gets the 4th most MVP votes for the Heat, despite being their LVP.  This problem results because total usage has a somewhat significant margin of impact on votes that a player receives and Wade had the second highest TU rate on the Heat.

James and Bosh run away with the MVP votes, with James procuring a solid victory.  In the single IoS numbers, James claims the overall MVP of the series, albeit barely, but since the Spurs obviously won the series, a .01 differential should certainly be overlooked and the overall MVP of the series should be given to Kawhi Leonard.
Thanks for reading.

Saturday, August 3, 2013

MLB, NBA, and NFL Variance

Not the most interesting post in the world..  But, I figured I would go ahead and post it anyways.  Basically I was doing some statistics a little while back and I wanted to attempt to apply a semi-real world application for variance to the three major sports (in the USA), NFL, MLB, and the NBA.  I'm not certain if I actually did it correctly (And I didn't use very much data.)..  I don't really think the charts even provide much information, but here they are anyways.

I'm running into some time constraints, but I think I'm going to end up posting my NFL predictions a while before week 1, and then if there are any small changes I want to make, I'll just change them in a separate post.  I have a couple posts which only need to be quickly edited in order to post, so there will likely be more content in the next week or two.



Variance Average game differential from .500 Variance final result Average game differential from .500 A smaller number means  less parody
NFL - AFC 185 47 0.72 2.9

NFL - NFC 111 36 0.43 2.3

NFL - Complete 296 83 0.58 2.6 0.16












NBA – West C 2438 174 1.98 11.6

NBA – East C 2083 160 1.69 10.6

NBA – Complete 4421 334 1.84 11.1 0.14












MLB - AL 1976 142 0.87 10.1

MLB - NL 2548 162 0.98 10.1

MLB - Complete 4524 304 0.93 10.1 0.06













Main things from the first table..  The most stable leagues, in order, in terms of variance, were: first the NBA, second the MLB, and third the NFL.  Also therefore the leagues with the most parody would simply be reversed.. Hence, NFL, MLB, and then the NBA. (I could be reading variance erroneously, the order may actually be reversed.. And plus it is quite possible that I did something incorrectly..  I think I should have prorated beforehand.. But, yet I don't know if they should have really been on an even playing field or not.  Overall, I don't entirely agree with what I think variance said was true.)

Average game differential from .500 made a bit more sense.  It, combined with the smaller number column (Which was an after-hand prorated version of variance), said that the MLB has very little parody, whilst the NFL has just slightly more parody than the NBA.  That I would agree with.

Average game differential from .500 NFL NBA MLB
NFL 2.6 2.2 1,0
NBA 13.1 11.1 4.9
MLB 25.9 22.7 10.1








Variance





NFL 296 863 447
NBA 1517 4421 2290
MLB 2997 8734 4524








NFL 0.58 1.69 0.87
NBA 0.62 1.80 0.93
MLB 0.62 1.80 0.93

The bottom 2/3rds of the chart do not really add much to anything, I am keeping them just in case they could be useful to me at some much later date.  But, the top part is a bit more interesting and just lists what things would be like in terms of differential from .500 if the leagues would hypothetically swap teams. (And those teams would play exactly the same in their new sport as their previous sport..  The likelihood of that is effectively equal to zero.)
As an example, if the teams in MLB were transposed to the NFL, then the average team differential from .500 would only be a one game difference over a given season.  While if the teams in the NFL were transposed over to the MLB, then there would be an extremely large gap between teams and a lot less parody, with an average of 25.9 games separating a team from .500.