As always keep an eye on injuries and COVID Issues. All stats pull from the Last 10 games (L10), so be aware of teams stats that don’t reflect recent injuries, recent returns, and recent trades.
Large slate of games today, check back for updates periodically throughout the day as I get more info in and make updates to anything I may have overlooked.
Table of contents(Sorry, Just found out this is not clickable unless I buy the business plan. Working on a work around).
Summary & Records
Yesterday I finished re-formatting my Machine Learning models to output to an easy to read format in time for the games, here is how everything did. Reminder anything with less probability than 55% is considered a toss up and not counted towards the record.
Player & Team Updates
John Wall is out for the third straight game, not going to be fully reflected in the Rockets stats.
Evan Fournier returned from injury for the Magic last game, but the Magic offense continues to struggle, they have failed to score 100 points in 4 of the last 5 games.
Darius Garland returned to the lineup last game as the Cavs took down the Nets, his impact wont be accurately reflected in the Cavs stats.
Seth Curry is back in the lineup for the 76ers. 76ers just beat Boston the other day and Tatum is still out for the Celtics. The Celtics have struggled to score consistently with Kemba back and Tatum off the floor scoring 75 and 109 points in the last 2 games.
Jimmy Butler is still out for the Heat, his impact will be somewhat reflected in the stats now that he has missed the last 5 games.
Horford is out for OKC for the fourth straight game. His loss will be somewhat reflected in the stats.
Knicks on the second night of a back to back, San Francisco last night, Sacramento tonight. Sacramento is 8th in Pace and Knicks are dead last, Knicks may fade late in the game if Sacramento decides to keep a fast pace.
KAT is still out for the Timberwolves.
Sharp money yesterday went 1-2 (not counting the Laker/Bucks game since money came in on both sides).
ATS – Early Sharp money appears to be on the Kings (-3.5). Most of the games don’t have data available yet.
O/U – Early Sharp money appears to be on the Nets/Cavs Under (228) and Mavs/Spurs Under (224). Most of the games don’t have data available yet.
[Posted early today].
ATS – Some Sharp action has come in on the Rockets (+3, Small), Raptors (-2, Small), Cavaliers (+8), Hawks (-5), and Clippers (Small).
O/U – Sharp action on Bull/Hornets Over (228), Rockets/Pistons Over (205 up to 215), Celtics/76ers Over (205 up to 222), more Nets/Cavs Under (226), and Hawks/T-wolves Over (223).
ATS – Some small sharp action came in late on the Nuggets, but as with most days sharp action happened early in the day.
O/U – There doesn’t appear to be any significant sharp action on totals since the mid-day line movements.
How it Works
This is my own basic spin on this and the first model I ever made. The Basic Model takes a teams 2pt shots made vs. opponents 2pt shots allowed and averages them to get the 2pt shots made for that particular team. It does the same thing for 3pt shots and free throws (FT) and weights each projected number of shots made for that team. It then repeats for the opponent and uses those projected scores to produce line projections.
The Basic model has been much better at predicting lines, but lately has been on a cold spell.
If you don’t understand what you are looking at, please read my post about betting tips. The percentages represent value for that particular number. Good value alone doesn’t mean you should bet it (To quantify, >30% is great value, 20-30% is really good, 10-20% is decent, <10 is ok value, blanks are negative value).
How it Works
[For the Nerds] This linear regression model also leverages a ridge regression regularization which helps it perform significantly better (R2 = 0.34).
The Linear regression model was re-optimized and modified on 21 Jan 21, in hopes to make it more accurate at projecting lines (44%). The O/U projections were pretty solid on winning at 57%.
The linear regression model seems to have much more exaggerated lines than other models (i.e. it will project a -20 line if the teams are a mismatch, even though its hard for an NAB team to cover a spread that large).
If you don’t understand what you are looking at, please read my post about betting tips. The percentages represent value for that particular bet type, number, and price. Good value alone doesn’t mean you should bet it (To quantify, >30% is great value, 20-30% is really good, 10-20% is decent, <10 is ok value, blanks are negative value).
Machine Learning Algorithms
Welcome! please take the Poll below to help provide the best content!
Random Forest (R.F.)
[Nerd Talk] Best performance is capped at about 50 trees, anything more than that and the processing time/power required for a very minor improvement increases exponentially.
Stochastic Gradient Descent (SDG)
[Nerd Talk] This model utilizes the Perceptron algorithm and I do apply ridge regression regularization to this model as well.
k-Nearest Neighbor (kNN)
The lazy learner (That’s what my grad school prof called it). Nearest Neighbor has performed best with 30 Neighbors.
Neural Network (NN)
[Nerd Talk] The activation function for this 100 layer NN is a linear function. The solver is an SDG.
Support Vector Machine (SVM)
[Nerd Talk] I originally had the Kernel set to linear for the past year, but started playing with it recently and found a drastic improvement using a Sigmoid kernel.
Adaptive Boosting (Ada Boost)
I did not learn boosting in Grad School, but I know this is some form of boosted tree algorithm. But I had the code and the data I already had didn’t need to be re-formatted so we will see how this 100 tree estimator performs together.