NBA Line Projections – 27 Feb 21

To all of those following along and leveraging this data, I want to apologize for the recent bad run and for any money you lost on my behalf. I could not understand why the new model was performing so poorly when it was running well in trials, so I dove into the code all evening and found a critical error I made in the code. When I switched over to the new model that added Pace of play stats in, I forgot I commented out the block of code that uses the Training data I should be using for an experimental weighted system that did not perform as well in trials. So the entire time I was using the wrong data to train the model than the data that showed good performance (The data I thought I was using). It’s only been a few days since making the switch, but its been a rough few days and I want to apologize, I feel like an idiot for making this oversight.

As always, keep an eye on injuries, back to backs, COVID Issues, and players returning from injury. All stats pulled are TEAM stats from the Last 8 games (L8) and don’t account for individual players. So be aware of new absences and players returning from an absence that could skew the L8 Team stats.

I will regularly post updates throughout the day and annotate with a “*”.

Advertisements

Records

Started tracking Moneyline projections as well. Started tracking Moneyline picks 24 Feb.

Top Performing Models

Based on Average Win % between ATS and O/U.

  1. Neural Net – 53%
  2. Random Forest – 52%
  3. Ada Boost – 50%
  4. kNN – 50%
  5. Basic Model – 50%
  6. SVM – 48%
  7. Linear Regression – 46%
  8. SGD – 46%
Advertisements

Team Trends

Quick look at how teams have performed over the last 8 games as compared to their season stats. Variance stats posted on Sat & Sun reports.

This chart shows teams that are playing better or worse recently compared to there season as whole (in regards to points). The stats shown are each teams point per game (PPG) both season and last 8 games. The “Off Trend” shows if a team has averaged more points in the last 8 games than their season average, which typically indicates an increase in offense improvement recently. The next 2 columns show the “Opp” PPG or opponents points per game. That is how many points each team is giving up on average. A positive number in the “Def Trend” column indicates a team is allowing less points per game and playing better defensively lately. The overall trend just adds the Off and Def trend to give an overall Point Per Game difference compared to the season average. Positive numbers mean the team is playing better recently. The trend numbers are color coded to make it easier to read, Green is positive and indicates improvement compared to season average. Red is negative and indicates a team playing worse compared to season average.

Team Consistency Update

What this chart is showing is each teams Variance, Standard Deviation (Std Dev), Ave Score for the season, average score over the last 8 games, and a Trending value. Variance and Std Dev are calculated from the last 15 games (need a larger sample size than 8 games). The list is ordered from lowest team variance to most team variance. The variance is how wide spread the data is, a team that scores between 100 and 102 points every night will have very low variance, whereas a team that scores anywhere from 75 to 145 on a given night will have very high variance. The Std Dev is the square root of the variance and is a good measure for how consistent a team is… NOT how good or bad a team is, but how consistent they are. Std Dev shows the amount a teams score typically deviates from the average on a given night. The 76ers score (typically) will deviate about 6.11 points from there average on a given night (Last 15), where the Wizards score (typically) will deviate 20.95 points from their average on a given night (Last 15). Obviously the lower the Std Dev the easier it is for my models to project the score and provide higher probabilities. The Trending value is an indication of if a team is scoring more points over the last 8 game than there season average. If they are averaging more points over the last 8 than their season average, the trending value will be positive, if the are averaging less than their season average, the trending value will be negative.

Player and Team News

Teams on a back to back: Pacers (Bos to NY), Thunder (Home), Jazz (Mia to Orl).

Cavs/76ers: Tobias Harris is out for todays game for the 76ers, his absence will not be reflected in the stats. Drummond has now missed 6 straight games as the Cavs seek to trade him, so his absence should be mostly reflected in the stats. The Cavs where on a 0-12 ATS streak before rattling off 2 straight wins. Cavs won the first matchup between these two, 118-94. The Cavs are 2-8 straight up and ATS, and 4-5-1 to the over in the last 10 games. The 76ers are 6-4 straight up, 5-5 ATS, and 4-6 to the over in the last 10 games.

T-Wolves/Wizards: There aren’t any injuries that aren’t reflected in the stats. The Wolves are 1-9 straight up, 3-7 ATS, and 6-4 to the over in the last 10 games. The Wizards are 7-3 straight up and ATS, and 5-5 to the over in the last 10 games. Wizards won the first meeting between these 2 teams 130-109, KAT did not play in that game.

Pacers/Knicks: Mitchell Robinson will miss the 7th straight game for the Knicks, his absence will be mostly reflected in the stats. Elfrid Payton is doubtful and it appears he will miss his second straight game for the Knicks, his absence will not be reflected in the stats. Pacers are 4-1 on the second night of a back to back. The KNicks won the first meeting between these two teams 106-102. The Pacers are 3-7 straight up and ATS, and 5-5 to the over in the last 10 games. The Knicks are 6-4 straight up, 7-3 ATS, and 4-6 to the over in the last 10 games.

Jazz/Magic: James Ennis is out for the Magic today and has missed 2 of the last 8 games. His absence will not be reflected in the stats. Fournier has played in 5 of the last 8, his impact should be mostly reflected in the stats now. The Jazz are 4-1 on the second night of a back to back. The Jazz are 8-2 straight up and ATS, and 5-4-1 to the over in the last 10 games. The Magic are 4-6 straight up, 5-5- ATS, and 2-8 to the over in the last 10 games.

Pelicans/Spurs: The Spurs returned from a 7 day absence and lost to the Thunder last game. They are still impacted by COVID issues and have 5 players out today. Derozan is considered a game time decision for today, he has only missed 1 of the last 8 games so if he doesn’t play his impact will not be reflected in the stats. Aldridge is also questionable, but he has only played in 1 of the last 8, if he can’t go his absence will be reflected in the stats. Pelicans have been an over machine recently and the Trend chart above shows they are scoring and giving up 9 PPG more than there season average over the last 8 games. The Pelicans are 4-6 straight up and ATS, and 10-0 to the over in the last 10 games. The Spurs are 6-4 straight up, 5-5 ATS, and 5-5 to the over in the last 10 games. Pelicans won the first meeting, 98-95.

*Derozan will play today.

Nuggets/Thunder: Al Horford returns after taking a rest day yesterday, he has missed 3 of the last 8, his impact will be mostly reflected in the stats. Millsap has missed 6 straight for the Nuggets and is out again tonight. His absence will be mostly reflected in the stats. Nuggets are 2-4 straight up and 1-5 ATS in the games Millsap has missed. The Thunder are 2-4 on the second night of a back to back. The Nuggets are 5-5 straight up, 3-7 ATS, and 5-5 to the over in the last 10 games. The Thunder are 4-6 straight up, 6-4 ATS, and 3-7 to the over in the last 10 games. The Nuggets lead the season series 2-0.

Mavericks/Nets: Porzingis has missed 3 straight and is listed as questionable for today. If he cant go, his absence will not be reflected in the stats. Durant has missed 7 of the last 8 and remains out. His absence will be reflected in the stats. Mavericks are 7-3 straight up, 4-6 ATS, and 7-3 to the over in the last 10 games. The Nets are 8-2 straight up and ATS, and 5-5 to the over for the last 10 games.

Advertisements

Sharp Report

Early Money

Not a lot of betting data is available this morning.

ATS – Some Shar money has come in on the Knicks (+1) & Magic (+11, small).

O/U – Some Sharp money has come in on the Wolves/Wizards Over (235) & Nuggets/Thunder Under ( 225.5).

Mid Day

ATS – Some Sharp money has come in on the Nets (+0.5).

O/U – Sharp money has come in on the Jazz/Magic over (219).

Late Afternoon

No noticeable sharp action late.

Advertisements

Summary of Projections

ATS Projections

O/U Projections

Advertisements

Model Projections

1. Neural Network (NN)

Record

Record: 50% ATS, 55% O/U.

ATS
O/U

New Model Projections

How to read the projections: The model type is at the very top. Matchups are denoted by the alternating white/green pairings, away teams are on top, home teams on bottom. The model analyzes the matchup and projects the home and away teams score (Proj Score). The difference between the home teams score and the away teams score give us what the model says the line should be (Proj Line). The “Line” column is what the Vegas line is at the time I run the model. The “Line Diff” is the difference in the projected line and the Vegas line. A positive line diff means the projected line is in the away teams favor compared to the Vegas line, negative means the projected line is in the home teams favor. The “Cover Prob” uses a normal distribution and the teams variance to project each teams probability to cover the listed Vegas line. Same thing for totals, “Proj Total” is the sum of the projected scores, “Line Total” is the listed Vegas total for the game, the difference between the two, and the probability to go over or stay under denoted by “O” and “U”.

Betting Edge

If you don’t understand what you are looking at, I recommend reading my post about betting tips. The percentages show the betting Edge, which is the cover prob (from above) minus the implied probability (-110 odds implied prob is 52.4%). If a team has a 92.4% chance to win and a 52.4% implied probability (or -110 odds), the Edge is 40%. The Edge alone doesn’t mean you should blindly bet it. To quantify, >35% is great value, 20-35% is really good, 10-20% is decent, <10 is ok value, blanks are negative value. ML is moneyline.

Advertisements

2. Random Forest (R.F.)

Record

Record: 48% ATS, 57% O/U.

ATS
O/U

Model Projections

How to read the projections: The model type is at the very top. Matchups are denoted by the alternating white/green pairings, away teams are on top, home teams on bottom. The model analyzes the matchup and projects the home and away teams score (Proj Score). The difference between the home teams score and the away teams score give us what the model says the line should be (Proj Line). The “Line” column is what the Vegas line is at the time I run the model. The “Line Diff” is the difference in the projected line and the Vegas line. A positive line diff means the projected line is in the away teams favor compared to the Vegas line, negative means the projected line is in the home teams favor. The “Cover Prob” uses a normal distribution and the teams variance to project each teams probability to cover the listed Vegas line. Same thing for totals, “Proj Total” is the sum of the projected scores, “Line Total” is the listed Vegas total for the game, the difference between the two, and the probability to go over or stay under denoted by “O” and “U”.

Betting Edge

If you don’t understand what you are looking at, I recommend reading my post about betting tips. The percentages show the betting Edge, which is the cover prob (from above) minus the implied probability (-110 odds implied prob is 52.4%). If a team has a 92.4% chance to win and a 52.4% implied probability (or -110 odds), the Edge is 40%. The Edge alone doesn’t mean you should blindly bet it. To quantify, >35% is great value, 20-35% is really good, 10-20% is decent, <10 is ok value, blanks are negative value. ML is moneyline.

Advertisements

3. Adaptive Boosting (Ada Boost)

Record

Record: 50% ATS, 51% O/U.

ATS
O/U

Model Projections

How to read the projections: The model type is at the very top. Matchups are denoted by the alternating white/green pairings, away teams are on top, home teams on bottom. The model analyzes the matchup and projects the home and away teams score (Proj Score). The difference between the home teams score and the away teams score give us what the model says the line should be (Proj Line). The “Line” column is what the Vegas line is at the time I run the model. The “Line Diff” is the difference in the projected line and the Vegas line. A positive line diff means the projected line is in the away teams favor compared to the Vegas line, negative means the projected line is in the home teams favor. The “Cover Prob” uses a normal distribution and the teams variance to project each teams probability to cover the listed Vegas line. Same thing for totals, “Proj Total” is the sum of the projected scores, “Line Total” is the listed Vegas total for the game, the difference between the two, and the probability to go over or stay under denoted by “O” and “U”.

Betting Edge

If you don’t understand what you are looking at, I recommend reading my post about betting tips. The percentages show the betting Edge, which is the cover prob (from above) minus the implied probability (-110 odds implied prob is 52.4%). If a team has a 92.4% chance to win and a 52.4% implied probability (or -110 odds), the Edge is 40%. The Edge alone doesn’t mean you should blindly bet it. To quantify, >35% is great value, 20-35% is really good, 10-20% is decent, <10 is ok value, blanks are negative value. ML is moneyline.

Advertisements

3 comments

  1. Thanks for posting the Ada Boost. Have you been keeping track of the old model for the RF and NN? How are they performing vs the newer models?

    Like

    1. The old models have been better the last 2 days, but I messed up the new model and was using the wrong training data for the new model. I tracked them for a week before making the switch, the O/U was about the same but some of the ATS projections were significantly better. hopefully correcting the data puts it back on the right path. I will continue to track the old model for at least a couple more weeks to see how they do compared to each other over a longer period of time.

      Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s