Generally speaking, the projections below are based on past performance of teams, players, injuries and matchups. The problem is that the algorithm used requires 3 full weeks of data in order to run. Therefore, in the past, I wouldn’t post at all until Week 4. However, several years ago I found that, with team and player adjustments (and a few other tweaks), I could use last season’s data, and – at least – approximate reliable projections for the first three weeks.
Bear in mind that the model’s end product is NOT to produce picks against the spread (which vary according to the spreads), nor game totals, but straight-up margins of victory. Totals, however, as well as ATS picks can be inferred, from the projected scores; but these are not the algorithm’s designed purpose.
Each week (except Week 1) I’ll be posting, usually, 4 times: Initial projections on Tuesdays (based on the prior week’s performance); Thursdays, with injuries included in the projections, and the final projections for Thursday’s game; Saturdays, with final projections for Sundays’ games; and Monday’s final projections for MNF.
For record-keeping purposes, ATS wins and losses are based on the final posted Straight Up projections and the closing lines for each game. This means that, although “picks” are specified at each posting, final picks for each game could flip, depending on the model’s collection of closing lines. The ATS record is a fair measure of reliability, but the key stat is always the SU projection.
My “door” is always open for questions, criticisms and bug-spotting.
GLTA
Bear in mind that the model’s end product is NOT to produce picks against the spread (which vary according to the spreads), nor game totals, but straight-up margins of victory. Totals, however, as well as ATS picks can be inferred, from the projected scores; but these are not the algorithm’s designed purpose.
Each week (except Week 1) I’ll be posting, usually, 4 times: Initial projections on Tuesdays (based on the prior week’s performance); Thursdays, with injuries included in the projections, and the final projections for Thursday’s game; Saturdays, with final projections for Sundays’ games; and Monday’s final projections for MNF.
For record-keeping purposes, ATS wins and losses are based on the final posted Straight Up projections and the closing lines for each game. This means that, although “picks” are specified at each posting, final picks for each game could flip, depending on the model’s collection of closing lines. The ATS record is a fair measure of reliability, but the key stat is always the SU projection.
My “door” is always open for questions, criticisms and bug-spotting.
GLTA
