Updated Simulator, follow the link here or scroll to the bottom***
I try not to feel compelled to bet the NBA throughout the season, so I followed the games as little as I possibly could. I’ve mentioned before, my unofficial NBA record is something like 0-501-1. But I couldn’t resist formulating a dynamic spreadsheet / simulator in order to grab some future value, for it worked out well for me in March Madness.
Having now developed some optimizing framework to see how teams interact under the different rules of expected value and probability, it was eminently practical to apply similar methods to the NBA. Therefore I’ve compiled an excel project of sorts that assesses the playoff bracket based on two conditions of two relationships. Resulting in four different ways of calculating team performance as a whole.
The linear relationship based on average line of which was explained to capacity here, in regards to the MLB. Additionally the more widely known pythagorean method using efficiency numbers explained here.
Combining these factors creates four different win probability scenarios for each team in each conference seeded 1-8, with the appropriate log 5 formula applied to test the odds of winning a seven game series.
I created a WIP playoff simulation as well for flavor, including HCA, the NBA finals 7 game series format, and a random normalizer integrated with team winning percentage under the four conditions (I used a standard deviation of .035, based on my judgment coupled with the very readable and worthy research paper here). I commandeered some of the code from the XLSSports baseball season simulator and adding myriad complex yet crucial IF statements to sufficiently satisfy my desired resolve.
Much of how the spreadsheet works is explained inside the spreadsheet itself. If you are just wondering about the various expected winning percentages to compare and look for future positions, here are links to the four different measures (true calculations minus the random normal inverse distribution of winning percentage):
Average line linear winning percentage.
Efficiency ratings linear winning percentage.
Average line pythagorean winning percentage.
Efficiency ratings pythagorean winning percentage.
Here is an example of one of the tables:
The most prominent and promising feature of the entire spreadsheet is this playoff scenarios sheet. The teams are placed in a drop down box so you can see the series prices for each potential matchup under the four different methods of analysis. As of yet you can’t do an user friendly scenario analysis of a potential finals matchup, but by simply entering the relevant data in the correct boxes and switching the location of games five and six, you can see what the prices would be. If you want more direct odds without randomness, remove the “norminv” formula from the equation in home and road win probability cells on the left.
Even with the integration of randomness for the most part the teams with the greatest probability, basically teams with HCA, chalkify the simulator. I might add some conditional components, such as:
If TeamA wins / simulations is 25% less than their win probability for that game, then something of consequential randomness happened to occur. In such case I would declare the underdog the winner
merely for entertainment.
Surveying the information and what I was able to gather is Orlando is every bit as good as Cleveland. +350 is currently their eastern conference odds per BODOG. In the Western Conference, Denver appears to have value versus the market. As well possibly long shots in Portland and Atlanta.
I’ll revisit the data before the playoffs start and mark down the various scenario odds for the teams of interest, and compare to what the books show.
Spreadsheets (enable macros):
NBA Playoff Scenarios (If you are only interested in the true odd playoff scenarios without the simulator)
NBA Playoff Scenario and Simulator
Enter the number of iterations in the blue box on the sheet “BRACKET SIM.” 1000 sims means each game is played 1000x. The first round can take up to 2-4 minutes. Time decreases as tournament progresses.
Related posts:
Recent Comments