Proposed method to track performance of our starters

Well I was hoping this attempt at manipulating statistics would hopefully brings it closer to the "eye-tests" so to speak - but it's works in progress at the moment and so will still need to rely on other members confirming if this particular analytics could be useful.....

 

Hooray for our 1st win this season - I don't have time to do this in detail but our starters PAC-score (compared at NBA level) are

J. Lin 9.6 (Decent)
J. Harden 19.4 (Wow!)
C. Parsons 7.8 (Slightly below average)
M. Morris 3.4 (Bad, I mean really really bad)
O. Asik 9.5 (Decent)

I will have to tweak the calculation method to address
(i) a need to cap each factor to a max (currently using 4.5 as a ceiling for each sub-factor - this is to address potential division by zero error i.e. for turnovers and some other figures)
(ii) a need to refine scoring FG%, 3P% and FT% - as their baselines are different so it seems to be much harder to score for those with higher baselines
(iii) a need to distinguish between 0/0, 0/1, 0/2, 0/3, 0/10 etc. they can't be all worth 0 right? Also to distinguish between 1/1, 2/2, 3/3, 10/10 they can't be worth the same score right? Will think about it - ideas are welcomed.

So will also need those of you who have watched the game if the current PAC-Score for our guys match the eye-test.

As a throw-in based on the same method PAC-Score the best players for pistons were Monroe and Stuckey - did that pass the eye-test as well?

thanks a lot!

 

Okay folks, after running the figures through the other teams - I am not happy with this methodology so am likely to try another approach - but will keep track only of our guys for interest sake just to see how it goes for a few more games.

Currently am thinking about making use of Hollinger's PER (with a twist) i.e. comparing position ranking value of starters (based on PER) and see if it makes sense in predicting our next game outcome.

Our current starters (as of today):
PG: Lin PER score 15.07 ranked number 22 amongst PGs
SG: Harden PER Score 37.09 ranked number 2 amongst SGs
SF: Parsons PER Score 11.21 ranked number 22 amongst SFs
PF: Morris PER Score 6.04 ranked number 34 amongst PFs (seriously he ought to be replaced by Smith 12.93 ranked 21 amongst PFs).
C: Asik PER Score 11.79 ranked number 19 amongst Cs

Too bad Atlanta hasn't played a game yet (so no current PER for its players at the moment) so we can't try to predict the outcome of the game. But if we used 2011 figures....
PG Harris 16.08 ranked 28 amongst PG last year
SG Teague 15.83 which would have ranked around 15 for SG last year
SF: Korver 13.59 which would have ranked around 35 for SF last year
PF: Smith 21.14 which would have ranked around 8 for PF last year
C: Horford 19.02 which would have ranked around 14 for C last year.

So Rockets total rank score adds up to 101 versus Hawks 100 - darn it if McHale wants to win this - he'd better drop Morris and start at least Smith instead - but it's gonna be pretty close one this..... or Morris finally improves...

of course the standard disclaimer applies for all kinds of predictions ;-) - past results does not guarantee future performance!

 

Okay based on PAC-Score our starters performances were:

JLin = 10.4 (great)
JHarden = 10.6 (great)
CParsons = 8.9 (decent)
P.Patterson = 8.3 (decent)
O. Asik = 7.1 (below average) - that's probably due to him not being able to score any points.

Looking ahead to the Game between HOU and POR

Our starters combined PER scores and combined Ranking Scores (in Brackets) are:
JLin 18.83 (27)
JHarden 38.42 (1)
CParsons 10.63 (31)
PPatterson 13.17 (24)
OAsik 5.15 (34)
Total 86.2 (117)

Portland's team
DLillard 27.42 (6)
WMatthews 17.6 (18)
NBatum 12.71 (24)
LAldrige 18.62 (9)
JJHickson 17.58 (5)
Total 93.93 (62)

It does not look good for our Rockets (but I am still keeping hope that Lin-Harden-Asik will continue to do well) - best wishes to our team!

 

damn, didn't want to be right based on those predictions... sigh.

Anyways, first loss - it was nearly a win should have been a foul on Harden at that last second but the refs decided it would be more exciting to have OT (our team really looked tired out there today).

Okay PAC-Scores indicated:
JLin 9.1 (decent)
JHarden 12 (still wow and with a sprained foot to boot)
OAsik 5.5 (below average - a concern since it seems to be trending downwards hope the short break will help him recover somewhat)
CParsons 8.2 (decent)
PPatterson 4.8 (below average - he needs to buck up)

Our next opponent would be Denver Nuggets so based on PER (rankings) comparisons (currently Denver has another game just before they play us which might change their scores and also noticed Hollinger's site hasn't adjusted the scores at this moment of posting will probably re-post once I see new figures) but here's a comparison:

Rockets
JLin 18.83 (27 amongst PGs)
JHarden 38.42 (1 amongs SGs)
CParsons 10.63 (30 amongst SFs)
PPat 13.17 (24 amongst PFs)
OAsik 5.15 (34 amongst Cs)
Total 86.2 (116)

Nuggets
TLawson 16.42 (34)
AIguodala 12.83 (36)
DGallinari 25.68 (4)
KFaried 6.46 (45)
KKoufos 5.29 (32)
Total 66.68 (151)

So far, Houston should be good for a win against Denver.... best wishes to our team!

 

There have been changes to the PER scores and it does impact the predictions - it has made the situation more of a touch and go for the Rockets, as the Nuggets have improved themselves over the games....

Currently it stands as

Rockets
JLin 15.56 (27)
JHarden 26 (1)
CParsons 10.74 (38)
PPat 11.24 (45)
OAsik 8.74 (37)
Totals 72.28 (148)

Nuggets
TLawson 14.49 (32)
AIguodala 14.13 (29)
DGallinari 14.3 (20)
KFaried 15.77 (20)
K.Joufos 9.96 (34)
Totals 68.65 (135)

So it is likely to be a close game - I am hoping for a Rockets win - Go Rockets! (PER score wise Rockets win, rank-wise Nuggets win).

 

Sad that we lost game 4 (I blame it on poor refereeing and shooting) but did our starters really play that badly? Not really too badly if we take a look at their PAC scores....

Game 4 PAC scores
JLin - 10.5 (great)
JHarden - 9.8 (decent)
C.Parsons - 6.1 (dropped to below average)
P.Pat - 8.6 (decent)
O.Asik - 4.9 (dropping although he scored well and have a double-double)

I still think there is still room for improvement for Asik - Parsons looks to be in a slump so I hope he can shoot himself out of it. Didn't really have a chance to watch the game - but I don't think J Lin fared that badly.

Just find it extremely strange for the CFers who keep asking for change of coach, and commenting about how certain players sucked and that we should play the rookies. Guys (and gals) - Houston is currently a fresh new team - let's give them some time to settle down and grow from there - if you snip and change too often there is no stable base for our players to build upon to make themselves better.

But that brings us back to a good point - so when do we start to panic and worry about our players form and performance? My personal preference is to follow the 3-strikes you're out principle - i.e. Let's give our "senior" players (they are still rookies too really just their are older than our actual rooks) some opportunity to give it a shot. I do think however, that adjustments need to be done if we do lose 3 in a row...... so our next game against the Grizzlies could be telling if things are falling into place somewhat.

PER score predictions for now (may change later as PER score is adjusted)

Houston
JLin 15.31 (31)
JHarden 22.51 (1)
CParsons 9.62 (48)
PPatterson 12.93 (35)
OAsik 9.74 (36)
Totals 70.11 (151)

Memphis
MConley 15.61 (29)
TAllen 11.43 (45)
RGay 16.23 (10)
ZRandolph 16.2 (21)
MGasol 18.98 (6)
Totals 78.45 (111)

Sad to say we may still be losing this (sigh) - but I am hoping that our players surprise me - Go Rockets!

 

Nice ideas B and hoopstar, but I note that the method I am using is already flawed so not likely to expand it to include bench performance (but it certainly worth considering) also liked the idea of analyzing per sub-factor. Currently I am taking a straight ratio factor - this is not ideal - I should be seeking some form of normalization as well as adjust weight ages for different positions i.e pgs need emphasis on assists, centers rightly should have high rebounds etc.

I don't like the overlaps within fg% with 3p% (part of 3p is within fg so there is double counting there) and the normal nba stats don't cover other useful analytics e.g. Help Defence

I was also using comparison to averages but the averages for the different subfactors are not the same (so a need to normalize) otherwise there will be a natural inherent weightage bias.

I will likely go back to the drawing boards after 10 games and see what options I have or ideas to tweak or change it..... Thanks for the suggestions!