Celtics Wins Regression Analysis

[ManchesterCelticsFan]

Boston Celtics Regression Analysis    (Pun Intended)   
Season           W   L   Season
2007-2008   66   16   1
2008-2009   62   20   2
2009-2010   50   32   3
2010-2011   44.6   37.4   4
2012-2013   38.8   43.2   5

Using the last 3 regular seasons as a sample (albeit a very small sample), I calculated via Excel an Exponential Decline of the Boston Celtics for the next 2 seasons with KG, assuming no major changes to the roster. Of course, there are a lot of variables involved, injuries, effort, equation not an exact correlation with data, or whatever so this isn’t an exact science. An R-Squared Value of 0.9085 shows that the projected formula of
 
y = 77.781e^(-0.139x)
R² = 0.9085

Where,
x = # of Seasons with Big 3
y = # of Extrapolated Wins Produced
e = 2.718281828459045
ln(e) = 1

has roughly 91% correlation with the data.

45 and 39 wins the next 2 seasons with the Big 3 (assuming RA/PP resigned and KG kept with comparable supporting cast) definitely seems reasonable to me based on the last 3 seasons. Then again, the C's went 0.500 or 27-27 since their 23-5 start so maybe this regression analysis is giving them too much credit.

What is the regression analysis on winning a Championship in 2008, making it to the last game of the second round last year, then this year? Let’s hope it’s not a first round ouster. lol. I predict the series to go 7 and it can go either way. If the C’s face the Cavs, even at full effort and "three Rondo's out there", I’m still not sure they can beat them.

[PosImpos]

That looks about right to me.  45 wins for next year, unless the roster is gutted (or somehow super upgraded), seems like a reasonably optimistic estimate for next season.

If there are big chemistry or injury issues, though, or if the team gets off to a poor start, we could be talking about a 35-40 win team. 

Or we could be talking about a 55 win team - just look at the Suns as an example of a team that can put it all together with only a few small additions (and some key subtractions).

Two seasons from now won't matter because there will be a lockout.  After that it's really too hard to say because that's when the Big 3 will begin to retire and the roster will likely be very, very different.

[More Banners]

Barring some significant change in the team, I'd say 45 and 39 are right on for the next two seasons.

Next year will be a tough call for Danny.  The team needs to be more athletic (lateral quickness and hops on D, but offensively too to RUN BABY!).  I suppose it's possible to add two players TO THE ACTUAL PLAYING ROTATION that could really change the character of the team that way; perhaps a PF and a SF might be enough to do it, with KG at the 5 and Ray or Paul pulling up or trailing for 3.  They're smart and unselfish enough to get near halfcourt, take an outlet, and relay it up the court to someone younger/faster.

I think this definitely could be a reality, though, but would mean the likely trading of Perk and the use of the MLE to fill those two spots.  If that were to happen, I think we might be a 50 win team next year, but a stronger and deeper one than we are now.

[Jon]

Well, couldn't one argue also that with Rondo's improvement and if everyone stays healthy (which is certainly possible) that the team wins 55+ games?  I mean they would've likely won 55-60 had they been healthy all year this year. 

That doesn't mean that they're necessarily a contender, but I do think we should keep that in mind. 

[PosImpos]

Well, couldn't one argue also that with Rondo's improvement and if everyone stays healthy (which is certainly possible) that the team wins 55+ games?  I mean they would've likely won 55-60 had they been healthy all year this year. 

That doesn't mean that they're necessarily a contender, but I do think we should keep that in mind. 

Well, the number of losses wasn't just because of injuries; it was a lack of focus and the falling off of the play of the Big 3.  The former might improve next year, but the latter doesn't seem likely to.

I wouldn't be at all surprised if Perk got traded for a vet + a pick or a prospect and an expiring contract, something like that.  I agree that aiming to rebuild the team as a contender-lite with younger, quicker, more athletic role players could definitely be the way to go.

Two players I'd definitely be interested in are Kelenna Azubuike and Tyrus Thomas.  My dream is that the Celtics could manage a trade for Anthony Randolph, who I think is in for a breakout year.

[snively]

It's too hard to predict right now.  We don't know if Ray will re-sign, who we'll get for the MLE (if we even spend it) or which of our free agents we'\ll re-sign.

[Edgar]

I will go with this graphic for this years playoffs

[Fafnir]

This is supposed to mean what? Basketball team records aren't an exponential process....

[Q_FBE]

What concerns me is that Danny will probably have his hands tied and be forced to make deals that weaken his club long term. Does Garnett still have any trade value? Pierce has an expiring deal but would he bring back young talent to play with Perkins and Rondo.

Marquis Daniels is out of here along with a host of other players leaving us short handed.

Next year is going to suck no matter what.

[guava_wrench]

Statistically speaking, I don't think this analysis makes any sense. You can do that for players because there is a typical career pattern we can see over a player's career. How can such an analysis be justified for a team?

[action781]

This is supposed to mean what? Basketball team records aren't an exponential process....

You're right!  They are linear!!!  Don't you guys remember how this played out the last time I projected wins by function modeling with small sample sizes?  I looked at our records during the 2006-07 season and the 2007-08 season:

2006-07:  24-58
2007-08:  66-16

to accurately predict with a 1.000 r^2 value  - perfect correlation! - on my linear regression model (y = 42x - 18) that in 2008-09 that our record would be 108 wins and -26 losses.

[PosImpos]

This is supposed to mean what? Basketball team records aren't an exponential process....

You're right!  They are linear!!!  Don't you guys remember how this played out the last time I projected wins by function modeling with small sample sizes?  I looked at our records during the 2006-07 season and the 2007-08 season:

2006-07:  24-58
2007-08:  66-16

to accurately predict with a 1.000 r^2 value  - perfect correlation! - on my linear regression model (y = 42x - 18) that in 2008-09 that our record would be 108 wins and -26 losses.

Well, you're deliberately ignoring the elephant in the room: the Celtics changed drastically in the summer of 07.

It seems pretty unlikely that the Celtics will be very different next year except, perhaps, for some new faces on the bench.