Our SOS ranking should end somewhere around 170, give or take 15 spots.
Printable View
I like Saragin, but I can do better. If we don't lose any games this year we are holding another Natty troph!! How's that!?
Sent from my iPhone using Tapatalk
I am wealthy!! I just finished eating a piece of apple pie given to me by a "former" UND fan from BisMan who watched our game with us last weekend at the Stadium, my little girl is sleeping after a fun nite together, I get to send her off to school tomorrow, and then do it over again tomorrow nite. Wait ... R you talking $$$?! Life is not so bad--plus I get to enjoy Bison FB on Saturday with friends, family and my fellow tailgaters. Not sure what else a man needs in life!! I already have guns, and Lakes is a Facebook friend--just sayin'!!
Sent from my iPhone using Tapatalk
thank you hammersmith
true, but does SoS factor into the ranking? I'll bet it does. Ndsus ranking is getting pulled down by pvamus horrendous ranking. No matter how well teams playing lower division teams do, they're not experiencing a similar effect. A perfect example is UNI. They are ranked highly despite being 0-4 (central state doesn't count). PVAMU is also winless. What's the difference in their ranking? SoS.
Sagarin’s strength of schedule is implicit in his rankings, as each rating compounds both records of opponents and records of opponents’ opponents. teams playing lower opponents get a free pass while teams playing poor division I teams get penalized (and rightly so).
SOS factors into the ratings, but I wouldn't necessarily say that NDSU's ranking is getting pulled down by Prairie View's poor SOS. The model predicts NDSU should have won by 52. NDSU won by 59. Outperforming the projection, all else equal, increases your rating. If Central State would have been in the Sagarin model, UNI's 59 point win may not have been more than the expected margin and their rating may have been harmed as a result, or as with NDSU and Prairie View, it may have been a marginal benefit. UNI likely remains highly rated for a couple of reasons. This is off the cuff w/out a deep dive into the numbers. One, they had a strong start value which will remain in the model until for another couple of weeks. Two, they lost two close games to D-IA teams on the road. Margin matters. If you lose close to a superior team it's entirely possible that your rating will increase. If you win close against an inferior team it's entirely possible that your rating will decrease. People punish UNI more severely for losing close to better teams than the computer does.
So looking at UNI's numbers in a way similar to how Hammersmith presents NDSU's Sagarin data each week shows that they played Wisconsin 10 points better than the model predicts, 1 point poorer against Iowa, 3 points poorer against YSU, and 6 points poorer against NDSU. The net is zero. Which I'd interpret as saying that the starting value Sagarin assigned them is quite close to their current rating of 65.53, though there could be other things going in with how the algorithm fits it's prediction to the actual results thus far.
As Dennis Green would say "They are who we thought the were." Sagarin's algorithm isn't "surprised" that UNI is 0-4 so it hasn't moved them off of their starting position very much. They were at 77th and 68.36 in the predict column before they played Iowa and they're at 87th and 65.53 now.