Silos should get a limited amount of teams into the playoffs

[Marvin Everett]

So last year SAC 7 seed beats GSC 2 seed and SAC three seed blows out GSC 6 seed. Then GSC 1 seed blows out SAC 3 seed.

[Boohaha]

You had mentioned winning the region earlier, which is why I ran those numbers. I also wasn't saying get rid of the 5-7, it's just the best categorization of your statement about lower seeds winning the region.

I'm sorry, I was unclear, I know you weren't arguing getting rid of 5-7 that was someone else. I meant eith my earlier statement yo include both those lower seeds that won the region AND those who made it to the championship. My point being that the current process is flawed enough that they should go to 8 teams, as the 7th and 8th teams are likely to be 3rd teams from strong conferences.

[Rational Observer]

I'm sorry, I was unclear, I know you weren't arguing getting rid of 5-7 that was someone else. I meant eith my earlier statement yo include both those lower seeds that won the region AND those who made it to the championship. My point being that the current process is flawed enough that they should go to 8 teams, as the 7th and 8th teams are likely to be 3rd teams from strong conferences.

I didn't argue for getting rid of the six and seven seeds; I said arguing over who should be the sixth or seventh seed was an argument against having them.

[Brandon]

I took the results from 2004-2018. This the average versus the seed sorted by best average over seed.

To explain:

#2. Jaguars
#7. Okra

- If the Jaguars beat the Okra, both receive 0. The seed was correct.

- If the Okra beat the Jaguars, they would receive +1. They would also receive +1 for every game they win that they weren't suppose to according to their seed.

- The #5, #6, and #7 seeds could receive a maximum of +3 points by winning the region.

- If the Okra beat the Jaguars, they would receive -2. That would be -1 for the game against the Okra that they lost plus -1 for the game that they failed to reach and win. They would not receive negative points for the next round, because they would have been expected to lose to the #1 seed.

- If there were no byes, the #1 seed would receive -3 for a first round loss. However, the most negative points they could receive is -2.

- If the #1 seed wins the region, they receive 0 points because that is the expectation.
x

Conf.	Appear	Points	v. Seed	# Over	%
GMAC	2	2	+1.0000	0	0.0%
IND	5	3	+0.6000	1	5.0%
MIAA	35	14	+0.4000	3	8.5%
PSAC	46	7	+0.1521	13	28.2%
GSC	41	6	+0.1463	13	31.7%
GLIAC	36	5	+0.1388	12	33.3%
NCC	11	1	+0.0909	4	36.3%
LSC	34	1	+0.0294	8	23.5%
SIAC	15	-1	-0.0666	4	26.6%
GAC	11	-1	-0.0909	4	36.3%
NSIC	28	-4	-0.1428	9	32.1%
NE10	20	-3	-0.1500	6	30.0%
SAC	23	-5	-0.2173	6	26.0%
MEC	15	-4	-0.2431	6	40.0%
RMAC	22	-5	-0.2272	9	40.9%
CIAA	17	-8	-0.4705	6	35.2%
GNAC	6	-3	-0.5000	2	33.3%
GLVC	6	-4	-0.6666	3	50.0%

*OK I think it's complete*

[Inkblot]

[Brandon]

I'll scan the sheet that I wrote it all on and send it to you.

[Redwing]

I guess it's easy to be critical, but a lot of these numbers are showing that the criteria are not the best metric to determine seeding(along with who is in/out.) An occasional upset should be expected... Heck even 25% upsets. But around 40%? Seems like a dartboard could be nearly as effective. I wonder if sharing the data (without including a region/conf) with NCAA would enhance them into making changes as to how teams are selected. Of course if the goal isn't trying to get the best teams into a playoff isn't a goal, then bets are off.

[Brandon]

I guess it's easy to be critical, but a lot of these numbers are showing that the criteria are not the best metric to determine seeding(along with who is in/out.) An occasional upset should be expected... Heck even 25% upsets. But around 40%? Seems like a dartboard could be nearly as effective. I wonder if sharing the data (without including a region/conf) with NCAA would enhance them into making changes as to how teams are selected. Of course if the goal isn't trying to get the best teams into a playoff isn't a goal, then bets are off.

I interpreted it as indicating that teams from the certain conferences are seeded too low far more often than they should be. They should be given the benefit of the doubt more often than they are.

Since 2004, the MIAA has only had three teams "upset" in the playoffs, and twice it was against another MIAA team.

2005 - #2 Washburn v. #6 Northwest Missouri
2011 - #4 Missouri Western v. #5 Northwest Missouri
2017 - #1 Fort Hays v. #4 Ferris State

Since 2004, the MIAA has played to seed 20 times. Since 2004, an MIAA team has "upset" another team 18 times. Some of those were obviously other MIAA teams as well.

Those the best among conferences with a reasonable amount of games.

[Redwing]

I interpreted it as indicating that teams from the certain conferences are seeded too low far more often than they should be. They should be given the benefit of the doubt more often than they are.

Since 2004, the MIAA has only had three teams "upset" in the playoffs, and twice it was against another MIAA team.

2005 - #2 Washburn v. #6 Northwest Missouri
2011 - #4 Missouri Western v. #5 Northwest Missouri
2017 - #1 Fort Hays v. #4 Ferris State

Since 2004, the MIAA has played to seed 20 times. Since 2004, an MIAA team has "upset" another team 18 times. Some of those were obviously other MIAA teams as well.

Those the best among conferences with a reasonable amount of games.

I figured that was your point. Still, I'll continue to be in that camp that the metrics aren't resulting in the best seedings. I won't go through the math, since it would be a huge undertaking, but I wonder how different the results would be if the Comm. used only w/l and SOS and used no other criteria.... including h2h. We would gripe the same.. or more, but I'd bet that the upset number wouldn't be much different. My point is that it doesn't appear as if much data analysus was used to come up with criteria that would indeed result in having the higher seeds winning at a much higher frequency. Maybe my expectations of favorites winning is too high, but Massey seems to do better with predictions so there appears to be a potentially better way.


Again, I'm not sure the NCAA (whoever that actually may be) is that concerned with getting the best team into the PO's with the best seeding order. At times I think it is structured to give all teams a fare chance with having as little bias as possible. They try to favor avoiding a perceived bias over trying to get the teams seeded as best as one can. For me, I prefer it the other way... get the best teams even if it looks like there may be some bias with how it is done.

[Brandon]

I figured that was your point. Still, I'll continue to be in that camp that the metrics aren't resulting in the best seedings. I won't go through the math, since it would be a huge undertaking, but I wonder how different the results would be if the Comm. used only w/l and SOS and used no other criteria.... including h2h. We would gripe the same.. or more, but I'd bet that the upset number wouldn't be much different. My point is that it doesn't appear as if much data analysus was used to come up with criteria that would indeed result in having the higher seeds winning at a much higher frequency. Maybe my expectations of favorites winning is too high, but Massey seems to do better with predictions so there appears to be a potentially better way.


Again, I'm not sure the NCAA (whoever that actually may be) is that concerned with getting the best team into the PO's with the best seeding order. At times I think it is structured to give all teams a fare chance with having as little bias as possible. They try to favor avoiding a perceived bias over trying to get the teams seeded as best as one can. For me, I prefer it the other way... get the best teams even if it looks like there may be some bias with how it is done.

I really didn't have a point. Just looking at where the numbers took us.

The numbers appear to back your claim that the metrics aren't achieving the best selections. Over time, one conference was .4 better that its average seed and only once was upset by a team not from the same conference. To me, that's an indicator that teams from that conference have consistently been seeded lower than they should have been.

For a lot of that time period, w/l and sos were the only criteria uses sans h-2-h.

[boyblue]

I interpreted it as indicating that teams from the certain conferences are seeded too low far more often than they should be. They should be given the benefit of the doubt more often than they are.

Since 2004, the MIAA has only had three teams "upset" in the playoffs, and twice it was against another MIAA team.

2005 - #2 Washburn v. #6 Northwest Missouri
2011 - #4 Missouri Western v. #5 Northwest Missouri
2017 - #1 Fort Hays v. #4 Ferris State

Since 2004, the MIAA has played to seed 20 times. Since 2004, an MIAA team has "upset" another team 18 times. Some of those were obviously other MIAA teams as well.

Those the best among conferences with a reasonable amount of games.

Team seeding seems to be more about wins and losses, an 11-0 team or 10-1 team will always be ranked above a two loss team no matter the quality of wins and losses. SOS seems to be used more for separating teams with similar records. On occasion you might see a 8-2 team from a stronger conference ranked above a 9-1 team from a weaker conference but that's about far as they will go.

It would seem that the criteria used is meant only to separate the wheat (Playoff worthy) from chaff (Non qualifiers), it's your resume. As we know a resume is no guarantee of performance. No matter the pecking order, the playoffs does a pretty good job of sorting out the mess. My one big take away is that it seems that home playoff games are a direct reward of winning more games.

[CatFan88]

2005 PO season might be skewing some of Brandon's MIAA results a smidge.

[Brandon]

2005 PO season might be skewing some of Brandon's MIAA results a smidge.

NW's run was a +3.
Pitt was a +2.
Washburn was a -1.

[Wallst]

If you want to limit teams from conferences that have silo scheduling we should also look at limiting teams from conferences who never win a game in the playoffs and are keeping the #4 team from the GLIAC out. Make it a Power 6 and then a group of 10. 4 teams each from the power 6 conferences and 4 teams in from the group of 10. Just like division 1 does except with more teams. At first it would cause much anger and would be very unfair. Within 5 years there would be significant conference realignment ( which Jim Johnson of the MIAA and now Pitt would screw up) and you would have 6 megaconferences and 4-5 group of 5 type conferences who would be assured of getting their conference champ in.

[WVIAC-F-EVER]

conference size has an issue - if the northern sun went to two divisions, played a non conference and a championship game - they could 4 to 5 teams into the play off like the psac has done in SR1

Best thing a conference can do, get upto 14 teams, go 2 divisions, 1 non conference and a title game that last week...

on the strength of a conference - the 2nd place team in the MEC made the play off two years in a row, hosted an IUP team and gotnit destroyed by IUP both times...

problem with this discussion - sr3 and sr4 teams are so spread out

good discussion