Div III bracket as of 1/20 power ranking

[pOkLE]

I know I start to try to picture where teams will fall in the playoff bracket about this time of the season. Figured others would enjoy seeing potential brackets, too. I'll try to post one after next PR release as well.

NOTE: I did not consider any teams playing less than 15, winning district, and bumping one of the top 32.

Attachment Deleted

[raiderfan]

Nice. Currently, though, District 9 doesn't have a team in the Top 32, but the district winner (likely Country Day) will make the playoffs. So, they would bump out No. 32 West Fel.


I know you gave the disclaimer that you had not considered that, but just wanted to point it out. I don't know if there is any other case where a likely district winner is outside the Top 32.

[playwide]

If the seeding system allows Team A to lose 2 games to a divisional opponent (Team B) which then finishes above Team A in District Standings, but the system then seeds Team A higher than Team B....you have a broken & flawed system.

[pOkLE]

Jan 22, 2015 17:20:08 GMT -6 playwide said:

If the seeding system allows Team A to lose 2 games to a divisional opponent (Team B) which then finishes above Team A in District Standings, but the system then seeds Team A higher than Team B....you have a broken & flawed system.

I agree that it "could" be evidence of a flawed system....but should the system weigh district matches differently than the rest of the season? Theoretically Team B could lose twice to its district rival (team A) 0-1 and then turn around and run the table (beating all the other top-10 teams in the state) while Team A loses many games to same top-10 opponents.

Seems like PR is evaluating whole season, not just district.

Now, of course, it's no where near perfect when there's no real consitency in scheduling from one team to the next....and you get same points for beating a 10-0-0 team as you do when you beat a 8-20-4 team. I'm just not sure your scenario is the smoking gun.

This is a specific scario, I assume?

[playwide]

I do not think anyone would want to downplay the fact that a ranking system has no consideration for Head to Head competition, which should be the biggest factor for teams that DO play one another.

The only reason you need a Power Ranking System is to attempt to create a level evaluation metric for Teams that do not all play each other, but if within that PR system, there is no consideration for Head to Head, it is flawed and cannot be trusted to blindly seed the playoffs. Period.

[pOkLE]

Jan 23, 2015 13:28:49 GMT -6 playwide said:

I do not think anyone would want to downplay the fact that a ranking system has no consideration for Head to Head competition, which should be the biggest factor for teams that DO play one another.

The only reason you need a Power Ranking System is to attempt to create a level evaluation metric for Teams that do not all play each other, but if within that PR system, there is no consideration for Head to Head, it is flawed and cannot be trusted to blindly seed the playoffs. Period.

I see your point. But I don't think it's accurately stated. It may not be the "right" consideration but it isn't "no consideration". Say both Team A and Team B are 10-0 at time of game at end of season: Team A wins and gets 15 points; Team B loses and gets 5.5 points. Isn't that recognizing the winner of that head-to-head match?

I'm not smart enough or have enough knowledge to advocate which system is best to use....but I don't think head-to-head would get us very far as the paramount consideration -- A beats B, B beats C, C beats A, D beats C, B beats D. Who's on top?

I realize it's easy to be devil's advocate. I guess I need more specifics to better understand. It seems to me the flaw is disparity between schedules as opposed to not giving due weight to head-to-head matches. But I am by no means a math person.

[raiderfan]

I just don't think there will ever be a system that makes everyone happy. If playwide were happy with just the right weighting, in his mind, for head-to-head results, there would be someone else unhappy, thinking it was weighting head-to-head too much.

Personally, I think H2H is somewhat important, but shouldn't be overemphasized, because anything can happen in a given match, unless one team is just way superior to another. If you've been around soccer an time at all, you've probably seen a number of games where the "better" team didn't win. It's true in all sports, but especially so in a sport like soccer, in which there isn't usually a lot of scoring. Sometimes a team that is not necessarily as good as their opponent can sneak a goal, while the "superior" team either just misses on their best scoring chances or is denied by a great play by the keeper.

So, head to head shouldn't be ignored, but winning head to head doesn't mean you should automatically be ranked above the team you beat. Overall results are more important, in my opinion.

[playwide]

Now that the various mindsets have been expressed. The big reveal:

Lusher beats Haynes twice in District Play...3-1 at Haynes, 5-2 at Lusher.
Lusher finishes 2nd to Newman in District ahead of Haynes.
Coaches have Lusher in the top 10 because they know they aren't a #18 team.

Power Rankings say Haynes at #14, Lusher at #18.

Your Honor, the Prosecution rests.

[raiderfan]

One problem with your analysis, playwide, is that the 2nd game against Haynes was not included in the latest PR.

[playwide]

Look forward to seeing what significant impact the second head to head victory has that the first head to head didn't recognize.
I thought the Jan 20 result would be in as of Jan. 20 report. My bad.

The jury is still out....

[p_malinich]

Jan 23, 2015 16:04:55 GMT -6 playwide said:

Look forward to seeing what significant impact the second head to head victory has that the first head to head didn't recognize.
I thought the Jan 20 result would be in as of Jan. 20 report. My bad.

The jury is still out....

Yes, PR's come out Tuesday a.m. for some reason.  I'd like them to be released Monday a.m. (thru Sunday).

I don't know that you'll be totally satisfied with the number you're waiting to see.  I did a quick calculation as if the 2nd win over Haynes were the only game played that night.  If that were the case, Lusher would get close to passing them, but not quite.  Lusher would increase to 7.83 while Haynes would fall to 7.85.  Now the reality is that other games are being played and will affect it.  

I totally understand why H2H feels like it should be considered, but when comparing 50+ teams, there are simply too many variables that are also important.  It's not really any different than 3 or 4 years ago when Lusher finished ahead of Haynes in district and had beaten them twice.  But Haynes pulled off an upset (over Newman I think) heading into the last week of the season.  They had a better marketing department than we did and the coaches seeded them ahead of us (despite our 2 victories over them).

In a tie situation, I totally agree H2H should be tiebreaker, but when looking at a season there's just too many chances of A beating B who beats C who in turn beat A.  

Now for the interesting part, if they finished at 7.83 and 7.85, it would be for the 16th & 17th spots and Lusher would open on the road at Haynes.   Hmmmm....

[playwide]

Thanks for the projections Paul. All I need to hear is that it's possible for Haynes to be seeded higher than a team that beat them twice, finished higher in the District, and also is higher ranked in the Coach Poll.....I listened long enough in my Logic Class in college to be able to evaluate that situation.

Sometimes there isn't a math formula for common sense....

They do have a common opponent this week. Lusher lost to Newman 2-1 and Haynes plays Newman on Saturday.

[p_malinich]

Jan 23, 2015 17:35:19 GMT -6 playwide said:

Thanks for the projections Paul. All I need to hear is that it's possible for Haynes to be seeded higher than a team that beat them twice, finished higher in the District, and also is higher ranked in the Coach Poll.....I listened long enough in my Logic Class in college to be able to evaluate that situation.

Sometimes there isn't a math formula for common sense....

They do have a common opponent this week. Lusher lost to Newman 2-1 and Haynes plays Newman on Saturday.

I thought Lusher was going to get the draw with Newman the other night.  Tough back-to-back games.

So play through this with me.  If Haynes should beat Newman, they would still be in 3rd in District, but would've beat one of the top 3 DIII teams in the state.  How should that affect seedings?  Should Newman drop below Lusher because they lost to a team Lusher swept (even though Newman swept Lusher)?  Should Haynes move up since they split with the top dog or stay below Lusher because they lost both to them?  

I'm not looking for answers per se.  In this isolated situation, I think district standings make sense, but once you start to look statewide, you need something that tries to compare performance.  Since Lusher's only losses in DIII were to Newman (others were DI & DII), it's hard to continue the comparison.

So let's go to St. Louis & Pope John Paul.   They tied 1-1 when they played head-to-head.  Are you suggesting that they should be at same position because of that?  St. Louis is #1 in PR and #4 in Coaches Poll.  Pope is #22 in PR and #14 in Coaches.  I don't think you are, but I'm just trying to figure out where that line is for H2H as a criterion.  And if I can build that into a model, I'm always happy to do so.  

[passion]

I agree that Lusher should be ranked ahead of Haynes, Team A with two H2H wins against Team Bwith an equal to or better body of work should net Team A a higher seeding. And just so happens that Runnels High, a team that Haynes decisively beat 4-0, is ranked ahead of Haynes. So I think that the Power Ranking has a couple of things wrong right now, hopefully the system corrects itself.

[playwide]

I am not suggesting that H2H is the only factor to be considered, I'm not even suggesting I have the answer....since I don't have to be the one to fix it, I have the luxury of pointing out it's flawed without having to figure out the solution. I'll let people that enjoy that kind of thing, do their thing.

But, clearly we've outlined a real scenario (no if's) where a team will be under-valued by math formulas vs looking at results with a human eye and applying reason.

It looks like power rankings will accurately evaluate the TOP and the Bottom, but will not hit the mark for the middle of the pack. This will create some interesting first round games. But then the giant killer teams will ultimately prevail and the 4 teams we could all pick right now will be in the Semi's and everyone will miss the bigger point, but say, "Look, the PRS works.. the top teams were correctly positioned", meanwhile in the middle some of those teams will be seeded lower than they should, then end up playing tougher competition in the first round, then probably meeting a giant killer in the second round, it may be interesting to the neutral observer, but it will not reflect accurate seedings top to bottom.

That's all. I hope I'm not coming across as combative in any way. But I do believe there is a crucial component of the system that is being under-valued.

Goodnight, I have FA Cup to watch in the morning. I wonder how Man Utd's power ranking will look after today's Draw with Cambridge....

[p_malinich]

I don't think you're being combative at all.  There's definitely a gap there, but like i say, it was there before with the Coaches poll.  Any team on any day.  I think the Cowboys went 1-15 years ago & their one win was against my 14-2 'Skins.  Those district games are intense, but may or may not reflect the portfolio of the whole season.  In the Lusher/Haynes case I agree Lusher should be higher, but I'm not sure how to work that in.  Like I say, a few years ago, Haynes finished behind us in district, but jumped us (by several spots) in seeding.

While Haynes is playing Newman today, Lusher goes up against Pope John Paul and needs to take care of business.  If Lusher wins today and Haynes loses, then all would look fine on Monday's PR.  These won't be the exact numbers since it doesn't factor all previous opponents (just a few games), but based on this week's results only for just Lusher & Haynes, Lusher would be about 7.92 in 16th and Haynes would be about 7.77 in 18th.  

The bigger concern in my mind for both of them is next week where they play opponents with very few wins.  Haynes faces Country Day (2.5 wins) and Thomas Jefferson (2.0 wins) - winning both would still have a slightly negative impact on their PR.  Lusher has NOMMA (0.0 wins) and Country Day (2.5 wins) - again even winning will hurt the PR.  Ties or losses in any of those games would be very problematic.

[stinkycity]

Lusher and Pope John Paul should be a strong match today.

[passion]

Haynes also plays Pope John Paul II in addition to Country Day and Thomas Jefferson.

[playwide]

Lusher wins 2-1 vs PJP.
Newman wins 2-1 vs Haynes.

[p_malinich]

Jan 24, 2015 15:43:50 GMT -6 playwide said:

Lusher wins 2-1 vs PJP.
Newman wins 2-1 vs Haynes.

Jan 24, 2015 11:05:53 GMT -6 p_malinich said:

While Haynes is playing Newman today, Lusher goes up against Pope John Paul and needs to take care of business.  If Lusher wins today and Haynes loses, then all would look fine on Monday's PR.  These won't be the exact numbers since it doesn't factor all previous opponents (just a few games), but based on this week's results only for just Lusher & Haynes, Lusher would be about 7.92 in 16th and Haynes would be about 7.77 in 18th.  

And things are as they should be.... Winning the next 4 is key to hosting a first game.