NBA - Relative Point Spread Rankings (1/25)

Question

Pop open the champagne!  We&#x27;re now one month into the season and this means that teams have accumulated enough data for my models to be statistically relevant.  Typically it takes about 30 days for this to happen.  Before this point, data sets are somewhat reliable, but still may contain statistical anomalies that haven&#x27;t had a chance to normalize.  At this point, every team has played at least 15 games - some as many as 19 - and I can be confident that my models are accurate enough to share with you all.  Below I will list my relative point spread rankings for each team in order of best to worst.  This number is the result of a complex formula that gives different weights to the metrics that most accurately predict the outcome of a basketball game (the big 4 are eFG%, TOR, ORR, FTR among other things).  eFG% is the most important - it has 50% relative weight in my model.  This works much like power rankings in the NFL.  For any given match-up, take the differential between the two teams and that gives you the point spread on a neutral site.  This number will then have to be adjusted depending on various non-statistical factors such as home court, revenge, injuries, fatigue, look-ahead, and any other intangible angle you might come up with.  A good baseline for this number is 3.5 points for the home team (but can range from 0-7).01.  Chicago Bulls  &#x2B;5.802.  Philadelphia 76ers  &#x2B;4.903.  Orlando Magic  &#x2B;4.304.  Miami Heat  &#x2B;4.105.  Atlanta Hawks  &#x2B;3.706.  Denver Nuggets  &#x2B;3.707.  Oklahoma City Thunder  &#x2B;3.408.  Los Angeles Lakers  &#x2B;1.709.  Utah Jazz  &#x2B;1.410.  Los Angeles Clippers  &#x2B;1.311.  Dallas Mavericks  &#x2B;1.212.  Portland Trailblazers  &#x2B;1.113.  Boston Celtics  &#x2B;1.114.  San Antonio Spurs  &#x2B;0.815.  Houston Rockets  &#x2B;0.516.  Indiana Pacers  &#x2B;0.517.  Memphis Grizzlies  &#x2B;0.418.  Minnesota Timberwolves  &#x2B;0.419.  Cleveland Cavaliers  -0.820.  New York Knicks  -0.921.  New Orleans Hornets  -1.022.  Phoenix Suns  -1.623.  Milwaukee Bucks  -2.324.  Golden State Warriors  -2.625.  Toronto Raptors  -3.226.  Sacramento Kings  -5.327.  New Jersey Nets  -5.428.  Detroit Pistons  -5.629.  Charlotte Bobcats  -5.830.  Washington Wizards  -5.8

si1ly · Answer

Pop open the champagne!  We&#x27;re now one month into the season and this means that teams have accumulated enough data for my models to be statistically relevant.  Typically it takes about 30 days for this to happen.  Before this point, data sets are somewhat reliable, but still may contain statistical anomalies that haven&#x27;t had a chance to normalize.  At this point, every team has played at least 15 games - some as many as 19 - and I can be confident that my models are accurate enough to share with you all.  Below I will list my relative point spread rankings for each team in order of best to worst.  This number is the result of a complex formula that gives different weights to the metrics that most accurately predict the outcome of a basketball game (the big 4 are eFG%, TOR, ORR, FTR among other things).  eFG% is the most important - it has 50% relative weight in my model.  This works much like power rankings in the NFL.  For any given match-up, take the differential between the two teams and that gives you the point spread on a neutral site.  This number will then have to be adjusted depending on various non-statistical factors such as home court, revenge, injuries, fatigue, look-ahead, and any other intangible angle you might come up with.  A good baseline for this number is 3.5 points for the home team (but can range from 0-7).01.  Chicago Bulls  &#x2B;5.802.  Philadelphia 76ers  &#x2B;4.903.  Orlando Magic  &#x2B;4.304.  Miami Heat  &#x2B;4.105.  Atlanta Hawks  &#x2B;3.706.  Denver Nuggets  &#x2B;3.707.  Oklahoma City Thunder  &#x2B;3.408.  Los Angeles Lakers  &#x2B;1.709.  Utah Jazz  &#x2B;1.410.  Los Angeles Clippers  &#x2B;1.311.  Dallas Mavericks  &#x2B;1.212.  Portland Trailblazers  &#x2B;1.113.  Boston Celtics  &#x2B;1.114.  San Antonio Spurs  &#x2B;0.815.  Houston Rockets  &#x2B;0.516.  Indiana Pacers  &#x2B;0.517.  Memphis Grizzlies  &#x2B;0.418.  Minnesota Timberwolves  &#x2B;0.419.  Cleveland Cavaliers  -0.820.  New York Knicks  -0.921.  New Orleans Hornets  -1.022.  Phoenix Suns  -1.623.  Milwaukee Bucks  -2.324.  Golden State Warriors  -2.625.  Toronto Raptors  -3.226.  Sacramento Kings  -5.327.  New Jersey Nets  -5.428.  Detroit Pistons  -5.629.  Charlotte Bobcats  -5.830.  Washington Wizards  -5.8

si1ly · Answer

Orlando is one of the bigger surprises on this list.  Their 3rd best ranking in the NBA is the result of high positive differentials in eFG%, ORR and FTR with a low negative differential in TOR.  Basically, they&#x27;re shooting well, playing good defense, getting to the free throw line more often than their opponents and dominating the offensive glass.Another surprise is Memphis as only the 17th best in the NBA.  Their eFG% and ORR differentials are basically a wash.  Their only strong positive differential is in TOR but this is neutralized by a very poor FTR.Boston&#x27;s positive relative point spread ranking is purely a result of good defense.  While they&#x27;ve been horrible offensively for most of the year (probably due to the number of injuries they&#x27;ve sustained), they&#x27;re still one of the better defensive teams in the league holding opponents to an eFG% of only 45.85%.

canadiantruth · Answer

Great work silly!You are an asset to this forum.

NoFumbles · Answer

You raised this shit to another level bro!  RESPECT!

si1ly · Answer

For anyone that is still unsure how to use these relative rankings, here's an example from today's card (1/25):

Denver +3.7

Sacramento -5.3

Take the differential (3.7 - -5.3) and you find Denver to be 9 points better than Sacramento on a neutral site. Both teams are well rested for this game and there are no major injuries or other intangible angles to speak of so I would use a standard adjustment of 3.5 points for the home team. 9 - 3.5 give you the true point spread on this game: Denver -5.5. This seems to be in-line with what Vegas put out for this game at Denver -6.

si1ly · Answer

Sportsbooks use this same type of statistical analysis to derive the point spreads that they put out with one key difference.  They give much more weight to the intangible angles - namely PUBLIC PERCEPTION.  I&#x27;m convinced that the best way to profit in the NBA is to find value between the true point spread and the public-perception adjusted line that Vegas puts out.

CoverLane · Answer

Thank you for the kenpom like stats. These will be very useful. I agree with your statement about public perception; a prime example would be the raping of the public in last nights cavs vs. Heat

si1ly · Answer

Absolutely.  The true point spread on that game was Miami -8.5.  At -13 they were way overvalued.  And golly, the Heat only won by 7.

RBears22 · Answer

si1ly heavy insite much appreciated

rlawson · Answer

This all looks very interesting. To test my understanding of what you're doing here, let's look at everybody's game of the night: the Wizards hosting the Bobcats.

Both sitting at (-5.8) on your scale would give us a relatively even match-up on a neutral floor. Because Washington is at home, we give them 3.5 points in their favor, putting the line (according to your system) at Wizards (-3.5).

Vegas set this line at (-7), so it appears that we have good value in Charlotte (+7). Is this correct? If not, please show me where I went wrong. Thanks for all of the work you put into this.

Davey · Answer

I love modeling. I try to do it and learn as much as I can....

Based on your rankings and taking into account public perception and oddsmaker objectives what do you think of these opening lines today:

Was -5.5
Hou -5
Phi -10.5
Gs -3

Thanks and nice work as always

ChaseWill · Answer

[Quote: Originally Posted by rlawson]  This all looks very interesting. To test my understanding of what you&#x27;re doing here, let&#x27;s look at everybody&#x27;s game of the night: the Wizards hosting the Bobcats.&#xD;&#xA;Both sitting at (-5.8) on your scale would give us a relatively even match-up on a neutral floor. Because Washington is at home, we give them 3.5 points in their favor, putting the line (according to your system) at Wizards (-3.5).&#xD;&#xA;Vegas set this line at (-7), so it appears that we have good value in Charlotte (&#x2B;7). Is this correct? If not, please show me where I went wrong. Thanks for all of the work you put into this.  [/Quote]A quick note, you need to take into mind injuries. D.J Augustine is out. Also, b2b games.. etc etc..

si1ly · Answer

[Quote: Originally Posted by rlawson]  This all looks very interesting. To test my understanding of what you&#x27;re doing here, let&#x27;s look at everybody&#x27;s game of the night: the Wizards hosting the Bobcats.&#xD;&#xA;Both sitting at (-5.8) on your scale would give us a relatively even match-up on a neutral floor. Because Washington is at home, we give them 3.5 points in their favor, putting the line (according to your system) at Wizards (-3.5).&#xD;&#xA;Vegas set this line at (-7), so it appears that we have good value in Charlotte (&#x2B;7). Is this correct? If not, please show me where I went wrong. Thanks for all of the work you put into this.  [/Quote]100% correct.  Although I would give Washington 4.5 points for being at home because Charlotte is playing on a back-to-back and also 4th game in 5 nights.  Washington also scores 5 ppg more at home than they do on the road.  Thus a true line of Washington -4.5.  Your process is stop on though.But before you back Charlotte because of the pure value you have to ask yourself why Vegas set the line so high?  Think back to the Philadelphia @ Miami game last week.  Philly shouldn&#x27;t have been &#x2B;7 in that game.. it should have been closer to Philly &#x2B;3.5.  However, Vegas set a shockingly high line.  Perhaps this had something to do with public perception.  But maybe Vegas knew that Miami would blow Philly out of the water and that&#x27;s why they set the line so high - to entice people to jump on Philly&#x27;s value.  Tonight&#x27;s Charlotte at Washington game has that same feeling to me.  I think Vegas set the line as high as they did to encourage people to back the underdog all the while knowing that Washington will dominate.  This is the difficult part about handicapping the NBA.  Sometimes value isn&#x27;t everything.. you&#x27;ve got to know how to READ BETWEEN THE LINES.  I lean Washington in this game for that reason, but because of the premium I&#x27;d have to pay to back them, it&#x27;s an easy pass.

si1ly · Answer

These are the spreads that my model spits out.Washington -4.5 (see analysis above)Houston -6.5 (I like HOU ML - I parlayed them with DEN ML at even money)Philadelphia -14.5 (Philly opens about 3 points too low when the spread is double digits)Golden State PK (Portland on a back-to-back-to-back so I&#x27;m not touching this one)

si1ly · Answer

Correction on post #14:  "Your process is SPOT on..."

si1ly · Answer

Keep in mind that this is just a starting point.  It&#x27;s a good way to narrow down the card but to really handicap the games you have to dig a lot deeper.  Looking at player match-ups and where teams like to shoot and how well the opponent defends that area are crucial additional steps that these rankings don&#x27;t account for.  Going back to the Denver @ Sacramento example.  Sacramento is one of the worst teams at defending the rim.  Denver is one of the best teams at scoring inside of 9 feet.  I can see this game playing out much like the Sacramento @ Memphis game a few days ago (Memphis also excels in the paint).

HDTP · Answer

Nice work, bro. I work on models as well and have a similar system.Your system appears to be more accurate.

si1ly · Answer

I&#x27;ll have my plays posted in a separate thread but for the record I&#x27;m already on:Atlanta &#x2B;4.5 (-110)   -  True line on this game is ATL &#x2B;0.5Denver ML  &amp;  Houston ML Parlay (&#x2B;100)Strong lean on Denver -5.5 (-110) as well because of the above information.  I also like New Orleans &#x2B;13, Philadelphia -11 an Cleveland &#x2B;3.

Davey · Answer

Thanks. The reason I posted those lines in particular is because I was curious if you had any opinion similar to your Washington one - I believe the line is trying to draw Charlotte action.

My opinion on the others was that Hou and phi opened low and gs opened strong for the same reasons. Any other intangible thoughts?

Davey · Answer

I do a linear regression in excel on the scores and closing betting lines to get a sense what the line should be. I get very similar numbers to yours but still experimenting. I think least squares puts too much weight on blowout scores and I am looking at ways to minimize that. Also thinking of complementing it with a wins-losses regression which I can then convert into a pointspread as well to get 3 indicators.

si1ly · Answer

It sounds to me like you&#x27;re doing a good job converting the intangible angles like public perception into tangible values that you can apply to the spread.  This is a very difficult thing to do, but I think you&#x27;re making the right moves.  My numbers above have absolutely nothing to do with points per game or offensive or defensive efficiency.  I try my best to leave them out of the equation because like you said, it gives too much weight to the teams that blow out their lesser competition.Good stuff.

si1ly · Answer

Philly may have opened a little low, but once you get into double digit spreads only the very public teams get extra weight.  I don&#x27;t think the Philly line is trying to entice the public into betting them - I think they&#x27;re just that good.  Same thing with Houston.  They may be on a nice winning streak but they&#x27;re still the Rockets and they have no real superstars to speak of.  Their spreads will tend to have some value associated with them because they fly a bit under the radar.  You need a superstar or a big market city to inflate lines and Houston has neither.  I think you&#x27;re right on Golden State though.  Looks to me like the oddsmakers want you to back Portland catching 3.5 points.  Out of the 4 games you listed, I think Washington and Portland are traps and Houston and Philadelphia have real value.

d0pebo1 · Answer

everyone do yourself a favor and get on this model QUICK!Great work si1ly

GerManiac · Answer

Wow buddy very well done and a great thing to get more insights of your capping process. I like it and it is very much appreciated

One thing I've always been busy ... is there really a home team advantage for each team ? If so, is it always 3,5 points ? For that Cavs @ Wiz example you gave a good reason why it sits at 3,5 points almost 4,5 like you said. Do you always go that way on capping the games ? What about the Blazers, is there the biggest home court advantage ? Can you help me with that ?
I hope you understood my point ... my english/gramma is upgradeable

sorry mate.
All the best on your picks...

I guess I take some too

NBA - Relative Point Spread Rankings (1/25)

Hot Forum Topics