Hey All:
I've been working on creating a math system to try and predict games since the season started - I originally had a post up detailing the plays - but the system has evolved a lot over the months and it's at a point that I feel like posting the plays again to track (which is why I'm creating a new thread).
Methodology: This system works off of team's offensive and defensive efficiencies as well as pace. Numbers are crunched for both away and home situation, so I have two views of a team - how they play at home and how they play away. There is no overlap between the two.
The efficiencies and pace are adjusted to weigh recent games more heavily (I have it now where the last 5 games make up roughly 50% of the calculation), because we care more about how a team is playing recently than how they played in November.
To account for strength of schedule, I have used Excel to reduce the error between my predictions and the actual results.
Each game does contain an error, regardless of how small, and these errors are averaged (using the same weight for the efficiencies) to come up with an average error. The weighted standard distribution of the error is also taken to create a distribution of guesses as to what the score for each team will be.
I'll take the estimated score based on my offensive efficiencies and then apply both the team's offensive error and standard distribution as well as the opponent's defensive error and standard distribution to come up with a team efficiency. Pace is computed the same way. I run 1,000 sims for each game 10 times and take the average of the %s from those ten sims.
What this gives us is a guess as to how team A on the road will do against team B at home based on prior statistics. Injuries are not taken into account (I don't have the computational power to get things down to the player level), so I recommend using +/- numbers for making the decision about whether a player is important enough to his team.
This is always a work in progress, as something will hit me that might seem like a proper addition down the line. Also, I might discover that the math is a bit off, so these bets should only be used to back up prior leans if you'd like to use this information at all.
Without further ado, here is tonight's only play (threshold of 58% or greater)
Miami Heat - Spread: -5.5 (64.03%)
It should be noted that the sim predicts this game to be a blow out, with Miami winning ~108 to ~ 94 (110 to 95 if based off the efficiencies alone without taking the errors into account - I'm still deciding whether I should work off of this as a baseline). They should be safe even at the current line of -6.5.
Best of Luck.