Last week I posted picks for a bowl “confidence pool”. Those simply require ranking the bowls 1-40 in how confident you are in who the outright winner will be. You bet/get 40 points for your most confident, 39 for next, etc. Our usual rankings are fine for that and it adds some flavor to the holiday games, hence the post.
That’s different from betting on sides. Which is the normal gig, I understand. But bowls are not normal. Our model is calibrated for the regular season. In fact we set aside bowls when using history to build our models. Using them for bowls is a little like siting-in a gun at 100 yards and then shooting at something 400 yards away. In high winds. You have to make some adjustments or you’re just wasting time.
So here’s a quick run through adjustments I’d recommend. Again, Rufus is the pro, but this is my best take. Warning: Lots of uncertainty + market moves means there’s not a lot to see here. I’m clarifying a bit, and digging for *something*.
Market lines for the bowls are much more compressed than ours. That is not how it is in the regular season, and is consistent with their recognizing how much uncertainty there is with bowls. We need to adjust for that. The simplest way to do that is “shrinking” our lines so that the standard deviation in our lines matches the standard deviation in the market’s. This year that means dividing ours by 1.6, a huge adjustment. That should tell you something about how uncertain the bowl season is.
Next is an adjustment we make all the time – blending with the market line. We do this with college (and not NFL) again because of uncertainty – more teams, fewer games, more player turnover, all make college much more uncertain (and possibly more fun). Recognizing that we incorporate other information when making a final line. The best available info comes from the market so we blend with that before declaring any edges. In short, we regress our lines toward the market. How much is an open question. We’ve used various numbers over the years. The more unique value we add to the line the less we should regress. Note: It’s *really* hard to add value to the market, so there’s no shame in using them heavily. The key is to add *something*. This year we’d recommend ~50/50.
That gives you an adjusted line and any edges that come with it. Your picks then just depend on your threshold. Historically we’ve used a 2-point edge as a threshold for an “other pick”, and something substantially higher (e.g. 3 points) for a Big pick. You’re going to be straining pretty hard to find any picks like that here. Especially given recent market moves.
So where are our biggest edges given these adjusted lines? Based on what we posted last week, and the lines available when we posted them, this was the situation. Originally it was 6 plays and a few more reaches if you were desperate. That’s softened but there’s still a bit of material for you to work with here.
Jan2: WVa (+1) vs. ASU (MP: -2.5). Line now -1.
Dec 28: Pitt (+3.5) vs. Navy (MP: +0.85). Line now +3.
Dec 30: Aub (-2.5) vs. Memphis (MP: -4.7). Line still 2.5. Call this the Priors Bowl.
Dec 24: SDState (+1.5) vs. Cincinnati (MP: -0.45). Line now -1.5. Geez.
Dec 31: FSU (-7) vs. Houston (MP: -8.95). Line still -7. Our model doesn’t know Herman from Levine.
Dec 26: Marshall (-4) vs. Connecticut (MP: -5.93). Line now -4.5.
Dec 23: Georgia Southern (+7.5) vs. Bowling Green (MP: +5.68). Line still +7.5.
Jan 1: Florida (+4.5) vs. Michigan (MP: +2.76). Line still +4.5.
Jan 2: KState (+11) vs. Arkansas (MP: +9.44). Line now +13. Finally!