The Limper Line is a numbers-based system for projecting probable outcomes for NBA and NFL games; and, although it’s done pretty well for the NBA, it’s had a couple of horrendous years projecting for the NFL. For what it’s worth, I’ve done significant remodeling of the algorithm in the off-season as well as a full reweighting of the variables used in the calculations, and hopefully the results will be improved, but there’s no guarantee (see below).
Basically, the model’s projections are based on past team and player performance, head-to-head matchups and injuries. The model requires at least 3 weeks of data, just to run. Therefore, the first three weeks of the season are based nearly entirely on data from last season. As current data is collected each week, new numbers are melded with old, until the 4th week’s projections. The first projection of each week will be posted on Tuesdays, and will be based on the last games’ numbers, regardless of injuries. Injury adjustments will be posted on Thursdays, and final projections posted on Saturdays.
It’s important to remind followers that what the model produces are “projections” only and should be used merely as a guide for the bettor. Projected “scores” are mostly irrelevant, but the “SU Favorites” or MOVs (Margins of Victory) are the ultimate point of the calculations. I post “picks” and track results only as a measure of the model’s reliability; and, although the last posted MOVs are final, the final picks, which are for ATS records-keeping, will often change depending on the closing lines.
GLTA
(FWIW- The NBA is a far easier sport for a data-based projection model to succeed than the NFL. First, the NBA has far few players to track, and a far bigger data base to draw from, given the much greater number of games played. And, second, luck plays a far larger role in football outcomes, than in basketball. That said, statistical projections always begin to fail after the NBA All-Star break due injuries, load management, and tanking, so there are really only 9-10 weeks of the NBA season (late November until early January) which model can reliably project.)
Basically, the model’s projections are based on past team and player performance, head-to-head matchups and injuries. The model requires at least 3 weeks of data, just to run. Therefore, the first three weeks of the season are based nearly entirely on data from last season. As current data is collected each week, new numbers are melded with old, until the 4th week’s projections. The first projection of each week will be posted on Tuesdays, and will be based on the last games’ numbers, regardless of injuries. Injury adjustments will be posted on Thursdays, and final projections posted on Saturdays.
It’s important to remind followers that what the model produces are “projections” only and should be used merely as a guide for the bettor. Projected “scores” are mostly irrelevant, but the “SU Favorites” or MOVs (Margins of Victory) are the ultimate point of the calculations. I post “picks” and track results only as a measure of the model’s reliability; and, although the last posted MOVs are final, the final picks, which are for ATS records-keeping, will often change depending on the closing lines.
GLTA
(FWIW- The NBA is a far easier sport for a data-based projection model to succeed than the NFL. First, the NBA has far few players to track, and a far bigger data base to draw from, given the much greater number of games played. And, second, luck plays a far larger role in football outcomes, than in basketball. That said, statistical projections always begin to fail after the NBA All-Star break due injuries, load management, and tanking, so there are really only 9-10 weeks of the NBA season (late November until early January) which model can reliably project.)
