The hot topic this week? Strength of schedule. How is it computed? How much is it factored by the selection committee?
KPI math can solve this problem. And it all goes back to the 2013-14 Wichita State men’s basketball team. First though, let’s look at more traditional strength of schedule (SOS) data using KPI.
KPI Strength of Schedule is computed by the average KPI of each opponent on a team’s schedule to date, including FCS (plus an adjustment for home/away/neutral). Remaining SOS (KPI SOS of future opponents) and Future SOS (KPI SOS of both past and future opponents) can also be computed.
KPI SOS numbers for the Top 25 teams in the KPI Rankings:
|KPI||School||Conf||W-L||KPI||KPI SOS||Remain SOS||Future SOS||Conf SOS|
|10||Ohio State||Big Ten||8-1||.285||(.019)||61||.084||58||.007||59||.085||44|
|12||Boise State||Mtn West||7-2||.245||.031||28||.014||73||.027||50||.012||65|
|16||Michigan State||Big Ten||7-2||.209||(.015)||56||.100||52||.014||58||.082||45|
|21||Kansas State||Big 12||7-2||.187||.013||41||.148||41||.047||36||.054||52|
The strength of schedule argument is tough. The sample size (12-13 games) is so small that specific SOS numbers are difficult to be representative of how good a team’s schedule actually is. In many cases, teams don’t control a large portion of their schedule. Also, teams can’t always predict who will be good and who won’t when they sign non-conference game contracts.
Penalizing a team because their opponents couldn’t win other games seems short-sighted. In college basketball last season, Wichita State went undefeated against a schedule that was significantly inferior to other teams in the conversation for a No. 1 seed. Yet they still got that top seed (third overall) from the men’s basketball committee.
I have a solution.
Instead of looking at SOS numbers, review KPI data vs. Top-25 opponents, vs. Top-50 opponents (cut line would be different in basketball, typically Top-50 and Top-100). This means that a team who plays fewer quality opponents has those games count MORE whereas a team who plays more quality opponents has an increased margin of error.
KPI Data against Top-25 and Top-50 opponents for the Top 25 teams in the KPI Rankings:
|KPI||School||Conf||W-L||KPI||W-L vs Top 25||Top 25 KPI||W-L vs Top 50||Top 50 KPI|
|10||Ohio State||Big Ten||8-1||.285||1-0||.784||2||2-1||.374||12|
|12||Boise State||Mtn West||7-2||.245||0-1||(.008)||64||4-2||.329||14|
|16||Michigan State||Big Ten||7-2||.209||1-2||.230||24||1-2||.230||27|
|21||Kansas State||Big 12||7-2||.187||1-2||.178||27||1-2||.178||34|
Notice that the top four teams in the KPI Rankings against Top 50 opponents are the same four that the committee put in their top four rankings this week. That’s not a coincidence.
Teams are not penalized for playing fewer good teams in these breakdowns, instead the fewer games against quality opponents take on greater importance. It was not Wichita State’s fault last basketball season that they played 21 games against Missouri Valley Conference opponents, none of whom made the NCAA Tournament. Instead, their few games against top competition took on greater importance.
This is the major benefit of the KPI being calculated on a PER GAME basis rather than a PER TEAM basis. The RPI can’t do this. Rather than looking simply at Top-50 wins, etc., these rankings give a more accurate representation of what that means. There’s a difference between three Top-50 wins against #6, #10, and #27 and four Top-50 wins against #29, #38, #46, and #49.
There are some downfalls to essentially eliminating all games against “bad” teams from the equation. Bad wins are eliminated, but bad losses (if they exist) would also be wiped from the resume in this case. Take Syracuse’s men’s basketball team last season – they lost to non-postseason teams Boston College and Georgia Tech at home. These outliers (see Virginia Tech at Ohio State, the second largest outlier this football season) are important but also not absolute.
Nothing is perfect, but THIS is why I wrote the KPI algorithm to be tracked by game rather than by team. The fundamental difference is what makes the math the most accurate. The interpretation is the next step.
The TCU vs. Baylor Debate: If Baylor and TCU both finish 11-1, the debate as to who finishes ahead of the other will be fascinating. Baylor won the head to head game, but lost to a West Virginia team that TCU beat. While everyone points to the SOS numbers being so different now, they will be eerily similar by the end of the season. The two teams have nine common opponents (8 Big 12 teams plus SMU) and play a 10th game head to head. The only difference between the two resumes in the end (sans home/away on opponents) will be that TCU hosted Minnesota and Samford while Baylor traveled to Buffalo and hosted Northwestern State. Is 1/12 of a SOS difference between scheduling Minnesota at home over Buffalo on the road really enough to make up for head to head?
I’ll save my opinion for later.
Comparing TCU & Baylor:
– Baylor beat TCU
– TCU has easier remaining SOS
– Teams have 9 common opponents pic.twitter.com/aq8V6F77Gt
— KPI Sports (@KPIsports) November 11, 2014
The Eliminator: The Eliminator assumes (not necessarily justly) that a Power 5 conference team with three or more losses and a Group of 5 team with one or more losses will not make the playoff. Under that assumption, 24 teams remain in play for the postseason playoff.
- ACC (4): Florida State, Georgia Tech, Clemson, Duke
- BIG 12 (3): TCU, Baylor, Kansas State
- BIG TEN (5): Ohio State, Michigan State, Nebraska, Wisconsin, Minnesota
- CONFERENCE USA (1): Marshall
- PAC-12 (4): Oregon, UCLA, Arizona State, Arizona
- SEC (6): Mississippi, Auburn, Alabama, Mississippi State, Georgia, Missouri
- INDEPENDENT (1): Notre Dame
Top 10 Week 12 Games ranked by G-Score (KPI Rankings entering Thursday’s games)
- #4 Mississippi State (9-0, 5-0 SEC) at #3 Alabama (8-1, 5-1 SEC), Saturday 3:30 p.m. ET/2:30 p.m. CT, CBS
- #2 Auburn (7-2, 4-2 SEC) at #11 Georgia (7-2, 5-2 SEC), Saturday 7:15 p.m. ET, ESPN
- #8 Florida State (9-0, 6-0 ACC) at #15 Miami-FL (6-3, 3-2 ACC), Saturday 8 p.m. ET, ABC
- #17 Clemson (7-2, 6-1 ACC) at #18 Georgia Tech (8-2, 5-2 ACC), Saturday 12 p.m. ET, ESPN
- #20 Nebraska (8-1, 4-1 Big Ten) at #25 Wisconsin (7-2, 4-1 Big Ten), Saturday 3:30 p.m. ET/2:30 p.m. CT, ABC
- #23 Missouri (7-2, 4-1 SEC) at #26 Texas A&M (7-3, 3-3 SEC), Saturday 7:30 p.m. ET/6:30 p.m. CT, SEC Network
- #10 Ohio State (8-1, 5-0 Big Ten) at #35 Minnesota (7-2, 4-1 Big Ten), Saturday 12 p.m. ET/11 a.m. CT, ABC
- #16 Michigan State (7-2, 4-1 Big Ten) at #37 Maryland (6-3, 3-2 Big Ten), Saturday 8 p.m. ET, BTN
- #14 LSU (7-3, 3-3 SEC) at #42 Arkansas (4-5, 0-5 SEC), Saturday 8 p.m. ET/7 p.m. CT, ESPN2
- #39 Virginia Tech (4-5, 1-4 ACC) at #28 Duke (8-1, 4-1 ACC), Saturday 12 p.m. ET, ESPNU