Like in the soccer world cup where teams are divided in tiers so there is an equal number of teams with similar strength per division.
You could randomly arrange teams 8 by 8 in order of some “strenght” metric like EPA or an official FRC ranking etc.
Say it with me
universal points model
45 Likes
Hmm, I guess as a stop gap regional teams could earn “district” points at their event(s). Each nationals-bound team could have their average “district” points per event calculated (removing the 3X multiplier & multi division playoff points from actual district championships), and then just snake draft those teams into all the divisions? Have a separate draft for rookies maybe?
I dunno, its not a very good system. (would top teams be manipulating event outcomes to “choose” their division? probably) But at least it remove some of the noise, is game-independent, and could be consistant from one year to the next.
1 Like
I don’t think they’d be able to choose their division. However, multi-regional teams could certainly qualify at their first event and then tank at the second which would drastically decrease their average points and lead to them being ranked lower than their ability. You might run into issues with the no game-throwing rules, but nothing prevents them from showing up at these events with a poorly built mecanum drivetrain bot with no mechanisms, very limited repair tools, and 2 batteries. On the other hand, I don’t really see these teams doing this. I’d expect them to still show up and compete in good faith in which case it probably wouldn’t be a major issue.
3 Likes
No as there is no metric works well every year.
Its why the first rule of picking is don’t pick via OPR or EPA or any other derived stat. You need to watch the teams.
Even worse teams if FIRST came up with a known formula, teams would take it as a challenge to reverse engineer it to put themselves in the division they wanted.
But what you can do is take a page from District Championships choose some metrics and use them to make tests to see if the divisions are actually balanced. If they aren’t reroll.
(Of course one needs to trust and verify that they ran the tests or else you’ll end up with DTEv1)
6 Likes
Last night we were discussing how cool it would be if there was a “World Cup Draw” level of excitement for the release of divisions. A LIVE event that all teams could watch to build excitement.
To save time of 620 teams getting placed I would pre group teams into units of 10, and then draw those units into each division. Would be great fun!
- Search World Cup draw on YouTube if you haven’t seen how fun it is
7 Likes
I’d watch the full 620 draw
3 Likes
Just as they divide rookies equally between divisions, I think it would make sense to equally divide other teams by how they qualified.
This way each division would have an equal number of regional alliance captains, regional first picks, second picks, impact award winners, points based district teams, etc.
I don’t think that’s as good a method as using district points like they do for district champs divisions, but it’s something that I think they could do using the existing data and methods they have.
14 Likes
There’s a small number of high OPR teams in Arch. But the divisions are pretty balanced overall. Well, Hopper seems to have more low OPR teams.
3 Likes
The closest thing they ever did like this was in 2016:
2 Likes
The band in that chart is wider in a practical sense than it appears to be…
If viewed from the perspective of the top 24 teams in the divisions by EPAExpected Points Added (admittedly only a proxy for potential performance), where most of the action is going to be for Einstein qualification, the divisions have a remarkable amount of imbalance.
Here is a table showing where a theoretical team with several levels of EPAExpected Points Added would be ranked within each division.
Rank in Division
| Division | 65 EPAExpected Points Added | 60 EPAExpected Points Added | 55 EPAExpected Points Added |
|---|---|---|---|
| Archimedes | 6 | 10 | 18 |
| Milstein | 3 | 12 | 18 |
| Galileo | 4 | 7 | 16 |
| Hopper | 2 | 7 | 14 |
| Curie | 2 | 6 | 13 |
| Johnson | 3 | 5 | 13 |
| Newton | 2 | 5 | 14 |
| Daly | 1 | 4 | 15 |
The ludicrously top-heavy Archimedes would see a 65 EPAExpected Points Added team in 6th position, outranked by 5 other teams, yet this same team would be holding the 1 rank in Daly.
Milstein has a big upper-middle bulge with a 60 EPAExpected Points Added team only holding the 12th rank while the same team would be ranked 4 in Daly. Wow!
Things get slightly more even at the 55 EPAExpected Points Added level, but even then, a team that would be at the 18th best EPAExpected Points Added rank in Archimedes or Milstein would get a substantial boost to 13th in Curie or Johnson.
Of course, division rankings, division championships, and the Einstein championship are not won on EPAExpected Points Added or OPR or Elo rankings. However, those rankings are still meaningful and show a pretty big disparity in division balance.
3 Likes
EPA is still a new statistic, so I would want to see how it holds up in different games before using it to do something like balance divisions. For example, in a game like Steamworks where points were scored nonlinearly, the system seems to become far less accurate, which was also a weak point for OPR.
“District” points are probably a much better option, but I don’t know how one would handle the unbalanced distribution of points across different regions. For example, a team that won their division at the Michigan DCMP would get points from both their division play and FIMstein. A team that won the Indiana DCMP would get fewer points because their event doesn’t have any divisions and therefore no additional opportunity for points. A Regional team wouldn’t have any DCMP points at all. You could just take the first two events (and double points for those with one event) but then it undervalues any teams that improved significantly between their qualifying events and DCMP. Are there any good ways to handle this? Maybe it’s worth not worrying about and just accepting it as being better than the completely random system.
3 Likes
We actually had plans to do something like this with FIRST this year, but a lot happened in the last few days in regards to getting organized and it just didn’t work for both parties. I hope we can consider this next year.
We will have a champs preview and prediction show on Saturday at 7pm eastern. I’ll have a separate announcement.
1 Like
I agree that an imperfect but less random method could still be worthwhile. Since points can’t be directly compared between districts, maybe they can do something like snake draft teams one district at a time into divisions, and then draft the at-large teams ranked by their, uhh… “regional points” is I guess what we’d call them?
FIRST could randomly pick the order in which districts/at-large gets drafted. That would also have the benefit of evenly sprinkling teams from each district evenly across all the divisions too, getting as much inter-district play as possible. Is there a big drawback I’m overlooking?
1 Like
Something weird also happens with their random. Out of the top 6 teams at the Bayou regional (my team included) 4 of us ended up together in Newton. Which could be just a coincidence but is strange since Bayou was the last event of all of us. And also the 2 remaining endend up together in Johnson.
This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.