Quote Originally Posted by HilliardBuck:
I know Brad Powers tweeted that Bill Connelly's returning production #'s for Stanford and SMU seemed off. Duck, are you hearing/seeing that others are off too? I have used Connelly's SP+ projection (for now), dumping a SP+derived power rating into the schedules for the P4 conferences to come up with RSW projections. The two that are really far off comparing those projection to FanDuel's and Bet365's RSW numbers are: UCF and Texas Tech. The adapted SP+ RSW totals are far less than the odds. Odds below are as of today: >>TTec is: 7.5, Ov -172, Und +138 at FanD, and it's 8.5, Ov -105, Und -125 at Bet365. The adapted SP+ RSW projection I have for TTech is 6.58 wins >>UCF is: 7.5, Ov -144, Und +118 at FanD, and it's 8.5, Ov +145, Und -180 at Bet365. The adapted SP+ RSW projection I have for UCF is 5.10 wins There is a decent size disconnect for much of the Big 12 when comparing SP+ projections to RSW available at the books right now. Colorado (no surprise) I show at 1.32 wins less than their O/U's at FanD & B365. On the flip side Arizona, Iowa St & Okl State each project a win total that is 1+ full game greater than the odds (when you factor in the juice). Iowa St has a ton of production coming back. Also, note that Oregon scrapped their game with Hawaii and is back to having a 12-game regular season schedule like the rest of the P4 conf teams.
I'm not sure that I'd rely too heavily on Connelly's SP + numbers at this
point as I'm pretty sure that they likely do not factor in all of the Transfer
Portal moves that have happened to date.
Back before Connelly went to ESPN , I did a study on his Opening SP +
numbers and just how accurate they really were and posted it here on the
forum. That was back in 2016 or so. At that time , as I recall , he was about
51 % accurate ( within + /- 7 points of the actual result. ). On 46 % of his
projections, he was off by 8 - 39 points on his projected plays from
the actual result. On 3 % of his plays , he was off by 40 + points on his
projected plays from the actual result. But most importantly , he projected
the wrong team to win approx. 30 % of all of the projected games.
Granted, my research dates back to 2016, prior to the Transfer Portal,
but that should make his work much easier to define than it is now.
Connelly's method of calculation of his SP + data has to be much more
difficult and potentially much more likely to have accuracy issues due
to the Transfer Portal and I would caution all who use his data to be
careful and check it to a second source to validate potential RSW's.
Helping Kama'aina to beat their " Local " since 1994.