Best Case Scenario
Posted: Sat Dec 13, 2014 11:59 amSo I was messing around on RPIForecast and I decided to try to estimate what the best case scenario for the league would be heading into NYE. I went to each team's page on the RPIWizard (where you can re-evaluate projected RPIs when entering in the outcome of games not played yet) and played out the league's non-conference schedule. I gave each team with a >30% shot at winning a W, re-processed the projection to see what the projected RPI was, then matched the projection to where it would fall in the rankings.
Even with the decision to award a win to anyone who was above a 30 percent projection, there was only one game that was a sub-50% winner that I awarded a league win for: Hall at Georgia, which is something like 36% on the site but I think in reality is closer to a tossup anyway. So most of what I entered were wins that we'd at least be a slight favorite for anyway. Two teams took losses: Providence against Miami, and St. John's vs. Duke, which were both sub-30% chances.
Factoring in DePaul was harder; with their Diamond Head games, I couldn't decide if it would be more beneficial for them to beat Colorado (which was a 36% game), because they'd be projected to lose the next two, or to lose that game and then beat the two teams in the bottom half of the bracket. I chose the latter, mainly because doing so would help the league more.
Finally, the Big 12 is projected at 103-21, with a weaker SOS. We'd probable eke past them for #1 if we pulled it off.
Here's what I came up with - new projected RPI first:
11 Nova
13 Butler
15 Xavier
19 Hall
29 St. John's
30 Georgetown
70 Creighton
84 Providence
91 Marquette
122 DePaul
That's six top 8 quality seeds for the Tourney, plus three more teams who aren't out of it with the right combination of results. Could be seven. YES STEVER, could be.
Now a few addendums: our projected conference record is 101-22. That's a six game improvement leaguewide (with two extra W's from Depaul's undefined Hawaii appearance as well that aren't reflected) from the estimation provided by the site. Having such a record would only wind up improving these numbers even more, but there's no way that I know of to calculate the level of combined improvement. It'll be a real help to everyone's RPI and SOS though, above and beyond these numbers.
Also, the conference tournament isn't factored in. The 4-5 and the 3-6 winner in such a scenario would receive a nice boost without penalizing the loser, while the 1-8 and 2-7 winners would either receive a small bonus (if it's the higher seed) or a huge one (if it's a smaller seed). That's not reflected in the rankings either.
St. John's and Providence have additional upside due to the negative projection of their games vs. Duke and Miami, both winnable, though tough. For the the record, the Johnnies would move up to 22 and Providence would move to 75.
One other reason to think we could head towards seven: the projections, being 'averages', tend to homogenize the performances of the very top and very bottom, moving them closer to the mean. The bottom three teams are projected to win 18 conference games as of now. That's 12 victories vs. the teams that will populate the top 7, and six versus each other. To me that sounds like the absolute ceiling for those teams: assuming one is DePaul, who I don't think could win more than 2 versus what will be the league's top 7, that leaves whoever the 8 and 9 teams wind up being 5-9 against the top 7. That's really optimistic. Possible but optimistic. The point being, if the bottom can donate a few more wins to the top 7, the numbers will jump for a few of those teams even more who wind up with the extra win over the projected total - an extra win is worth roughly 10 RPI spots if you're on the bubble, 6-8 if you're a middling seed, and even almost a whole seed line for a top 20 team.
I just found it interesting. Let's close out strong over Christmas and really produce a pounding.
Even with the decision to award a win to anyone who was above a 30 percent projection, there was only one game that was a sub-50% winner that I awarded a league win for: Hall at Georgia, which is something like 36% on the site but I think in reality is closer to a tossup anyway. So most of what I entered were wins that we'd at least be a slight favorite for anyway. Two teams took losses: Providence against Miami, and St. John's vs. Duke, which were both sub-30% chances.
Factoring in DePaul was harder; with their Diamond Head games, I couldn't decide if it would be more beneficial for them to beat Colorado (which was a 36% game), because they'd be projected to lose the next two, or to lose that game and then beat the two teams in the bottom half of the bracket. I chose the latter, mainly because doing so would help the league more.
Finally, the Big 12 is projected at 103-21, with a weaker SOS. We'd probable eke past them for #1 if we pulled it off.
Here's what I came up with - new projected RPI first:
11 Nova
13 Butler
15 Xavier
19 Hall
29 St. John's
30 Georgetown
70 Creighton
84 Providence
91 Marquette
122 DePaul
That's six top 8 quality seeds for the Tourney, plus three more teams who aren't out of it with the right combination of results. Could be seven. YES STEVER, could be.
Now a few addendums: our projected conference record is 101-22. That's a six game improvement leaguewide (with two extra W's from Depaul's undefined Hawaii appearance as well that aren't reflected) from the estimation provided by the site. Having such a record would only wind up improving these numbers even more, but there's no way that I know of to calculate the level of combined improvement. It'll be a real help to everyone's RPI and SOS though, above and beyond these numbers.
Also, the conference tournament isn't factored in. The 4-5 and the 3-6 winner in such a scenario would receive a nice boost without penalizing the loser, while the 1-8 and 2-7 winners would either receive a small bonus (if it's the higher seed) or a huge one (if it's a smaller seed). That's not reflected in the rankings either.
St. John's and Providence have additional upside due to the negative projection of their games vs. Duke and Miami, both winnable, though tough. For the the record, the Johnnies would move up to 22 and Providence would move to 75.
One other reason to think we could head towards seven: the projections, being 'averages', tend to homogenize the performances of the very top and very bottom, moving them closer to the mean. The bottom three teams are projected to win 18 conference games as of now. That's 12 victories vs. the teams that will populate the top 7, and six versus each other. To me that sounds like the absolute ceiling for those teams: assuming one is DePaul, who I don't think could win more than 2 versus what will be the league's top 7, that leaves whoever the 8 and 9 teams wind up being 5-9 against the top 7. That's really optimistic. Possible but optimistic. The point being, if the bottom can donate a few more wins to the top 7, the numbers will jump for a few of those teams even more who wind up with the extra win over the projected total - an extra win is worth roughly 10 RPI spots if you're on the bubble, 6-8 if you're a middling seed, and even almost a whole seed line for a top 20 team.
I just found it interesting. Let's close out strong over Christmas and really produce a pounding.