Up until now, the Limper has acted like a blind man throwing darts at a moving dartboard. The results so far are right down there with the first couple of seasons when I hatched this model, and its hard but not impossible to explain.
All statistical projections are based on prior performance numbers (PPNs), and the numbers I use are completely reliable. Also, PPNs are honest to goodness fair indicators of what will happen in the future. These are demonstrable facts. When I run these numbers through the first 6 weeks, the SU and ATS results are through the roof, which, of course, are false indicators of reliability. However, side by side comparisons of projected stats, game by game, of say yards gained rushing, or pass completion percentage, vs specific matchups - are well within the margin of error which means the models calculations are valid. Valid, but when it comes to being indicators of future events sadly wrong. At least, wrong thus far.
What has happened this season is that teams and players have been over and over either exceeding their PPNs or failing to meet to meet those PPNs by wide margins from one week to the next. As a rule, this doesnt happen as routinely as it has this season and, unless this season ends up as an extreme outlier, this should settle down.
Every year since I began posting my model’s projections, I have tried to develop a reliable Over / Under projection model. Never succeeded. But, like they say: if you don’t succeed the first time, try, try again – and again, and again, and again, and again, and again, and again, and again…
I ran it for weeks 5 and 6, and it went 8-2, so based on data thru week six, this is the 2018 version – FWIW.
(The Delta, as indicated, is the difference between the model’s projected totals and the current O/U line. There is no correspondence with the model’s projected spread totals, and the totals used to determine the O/U picks. Posted are the top 5 games, ranked according the delta.)