Einstein Round Robin Suggestions

I’m interested in seeing the community’s suggestions on possible improvements for the round robin tournament we currently have for Einstein.

Some current issues that have been pointed out are:

  • the last match(es) can have no effect on the actual seedings of the teams - additionally when a team goes 0-2 or 0-3 they don’t have a lot left to play for
  • the tiebreaker points can feel almost arbitrary, when there are many similarly strong alliances (such as Detroit this year), climb points can be an unfulfilling way to rank teams

Some positives:

  • if the tiebreakers were to seed every alliance accurately, you do have the “best” two alliances in the finals (ie. the “second best” doesnt get eliminated in quarters by the best - the finals should be the most competitive matches)
  • every Einstein alliance gets to experience at least 5 matches - the old model had some alliances being eliminated in two
  • the strategy/scouting for Einstein teams is more interesting and varied - you are forced to develop multiple strategies for differing matchups

One suggestion I’ve already seen is to have the tiebreaker be head to head matchups. For example, for Detroit this year, all the teams who had 3 wins would be sorted by the number of wins they had against the other 3 win teams - leaving Daly and Tesla as the two 2-1 teams. They would then be sorted by head to head again - as Tesla beat Daly, Tesla would seed first and Daly would seed second. A potential problem is when you have three teams, such as Houston 2017, that all went 1-1 - you would then have to find another tiebreaker.

Another suggestion I have is a mix of the old division vs. division and round robin. You start by having each division play another (ie. Archimedes v Carson, Curie v Daly , Darwin v Tesla) in a best of 3 series. Let’s say the three winners are Archimedes, Curie and Darwin. They then play a best of three series against the other two alliances (ie. Archimedes v Curie, Archimedes v Darwin, Curie v Darwin). If an alliance loses both matchups, they are eliminated and you have your finalists, with the 2-0 team becoming the red alliance. If all the alliances go 1-1, you do the same bracket (ie. Archimedes v Curie, Archimedes v Darwin, Curie v Darwin) except in a best of 1, until one alliance loses both of their matches. While this would be very exciting - every match means something and every time you have a three-way tie the suspense builds, it would have the potential to be quite long.

Does anyone else have any suggestions to improve the format?

Option 1: Pool Play

Split the line down the middle of the convention center. The three left fields are in pool 1, and the three right fields are in pool 2. Play, I don’t know? 6 matches? And the lowest rank team gets eliminated. 2nd rank team plays in blue bumpers against the 1st ranked team in the other pool in a pair of best of 3 series on Ford Field to determine the finalists, then a best of 5 determines the winner.

Pros: Gets you a semifinal round of 4 teams without having much higher total match count, can also cancel out the need for late matches if a team ends up winless after match 4, if my brain is working right.

Cons: Super confusing. We have 3 “Divisions” that break up into 2 subdivisions each, but the pools are different and actually split up a division’s worth of subdivisions??? It doesn’t make any sense (just like the division/subdivision naming convention!)

Option 2: Actual Wild Cards

Run an 8-alliance bracket for Einstein. Rank the top 8 alliances, including 2 finalist alliances based first on division finals win total, then tiebreak on average match score, then use the standard ranking tiebreakers for the given game.

Pros: Gets you more teams on Einstein, returns to a standard bracket, helps promote strong alliances that missed out by being on strong divisions since FIRST refuses to properly weight the divisions even though a tool has been made for such a purpose.

Cons: Maybe people don’t like the idea of seemingly random finalists being selected to move on to Einstein, especially if the point of Einstein is to determine the best alliance at the event, not the best 8. Have a lot more of an unpredictable match count to wrangle with managing.

Option 3: NFL Playoff Bracket

Use metrics to seed the alliances 1-6. Use finals win percentage, average match score, then game-specific tiebreakers to determine the rankings. 3/6 and 4/5 have a best of 3 round, then re-seed for a semifinal best of 3, then play a final best of 5.

1 Like

Top 4 alliances in Round Robin advance. 1 plays 4, 2 plays 3. Boom. Done.

i think all possible solutions would make the Einstein process much longer.
However, I believe that the round robin should be kept. Instead, take the top 4 alliances after the round robin, and do the traditional BO3 Semi finals and finals, where 1 faces 4 and 2 faces 3. I think this would limit the power of the tiebreaking components.

I think this is a solid option, and it does cope with the Detroit nightmare scenario well (where four teams finished 3-2, and the gap between making finals and 4 was two climbs). And I have a much harder time feeling bad if 5 is squeaked out rather than 3.

Things I’d want to see, even if a semifinal isn’t going to happen:

  1. Head-to-head matchups should be the first tiebreaker. When it’s a two-way tie, it’s obvious. When it’s more than that, it should be the record against those teams in the tie.

Start with four alliances (Daly, Carson, Tesla, Curie) that all went 3-2. Compare their records against just those other three alliances:

Daly: 2-1 (beat Curie, Carson)
Carson: 1-2 (beat Tesla)
Tesla: 2-1 (beat Daly, Curie)
Curie: 1-2 (beat Carson)

That would make the rankings:

Tesla (2-1 and beat Daly)
Daly (2-1)
Curie (1-2 and beat Carson)
Carson (1-2)

Considering Tesla outscored Carson on every stat except the all-important climb, I don’t think these results are out of whack. The tough part is communicating multi-team ties in an easy format.

  1. I can see why FIRST might put endgame first–they want to encourage the big camera-friendly finish. I think I’d prefer total match score for most games, since it prescribes less of the gameplay and encourages clean play–fouls against you would sting, since they’d literally be helping your opponent seed higher. As a sort beyond head-to-head, it’s easy to explain. Then I’d do auto, then (this year) ownership, vault, endgame.

I struggle to see a solution that doesn’t

A) Hinge on some metric for ranking / organizing teams that never played each other

or

B) Involve some potential late irrelevant matches

or

C) Involve a ridiculously large number of matches.

Of all these issues, I would rather take B over A or C. The round robin, in my mind, is fine. I would say that there should be a different metric for tiebreakers. The first order sort behind wins should be net point differential, and if the 2nd and 3rd teams wind up with the same record, I would have a grudge match to decide the 2nd spot on Einstein. From there, I would prefer a best of 5 to decide the champion, bu that’s just my 2 cents.

I think FIRST’s choice to use the endgame would make more sense if the endgame were a true endgame rather than a self-declared imitation of one.

In competitive strategy games (for example, Chess) the endgame is an implicit point where the game doesn’t develop any further, rather moving directly towards attempts to win. If FIRST was able to make a game that had a proper endgame (that fit in well with the “setup” that takes place during the primary game period) the endgame would make a lot more sense, in my opinion, to use as a first-order tiebreaker. 2016 is an example of a game with a more sensible endgame task (given you had to weaken the tower in order to get endgame credit and the RP).

That said, I think FIRST’s version of an endgame is better for ease of understanding and visual appeal. It gives a losing alliance a way to catch up at the end of the match and helps prevent a clear meta-strategy from completely dominating every other strategy.

Wipe all ranking points for eliminations. Start the ranking scores over for alliances during division eliminations but only count matches that an alliance wins towards ranking score. This ensures every team on Einstein has 6 matches calculating the ranking score. The 6 alliances are then ranked based on their ranking score. If there is a tie in ranking score than the same tiebreakers are used. The top two teams are given a bye to the semi-finals.

Sticking with the half baked theme of Einstein since 2champs came along, I think we should go to a Swiss triple elimination tournament.

Firstly, all teams chosen for eliminations are now considered to be on Einstein, or you could just create another division for it. Either way, all 42 alliances are now in their own tournament. Similar to round robin, there is no bracket. Alliances will be randomly sorted to play against each other. After this first round is complete, a new set of rounds, based on the Win/Loss of each alliance. (Example: Newton 3 won their first match, and Hopper 6 also won their first match. For the 2nd round, they will be sorted into those who all won their first match, and will be matched up against each other.) The first teams to lose 3 matches total get eliminated. After maybe 4-6 sets, we take the top remaining 6-12 teams (whatever you desire) and send them to Einstein for another round robin set, and then finals. Considerably more matches which means more fun for the viewers, and more matches in general for teams to give them their money’s worth!

Have 4 divisions again. It was always better. And go back to one championship while you’re at it.

Could not agree more.

What about not playing matches in the final round that aren’t necessary to determine who goes onwards?

Obviously there are going to be a ton of issues with that, but it should be relatively simple* between Round 4 and Round 5 to determine who’s definitely out, who’s definitely in, and if there’s a case where both alliances are in OR out, simply skip the match entirely. If there’s a seeding issue for the finals, take the head-head (if there is one), overall win-loss, then coin flip–or even play each alliance once on each side, then coin-flip any tiebreakers.

I think I’d prefer the bracket, but that requires either byes or more divisions that there may or may not be room to hold.

Oh, yeah: and 4Champs. 4 divisions per Champs, Champs are North, South, East, West, and FoC returns as a round-robin event for winners of each 1/4Champs followed by a bracket event. (Like that’ll go over well with HQ.)

*Comparison point: cold fusion.

Why be reasonable when you can be unreasonable?

Those scouters have families :frowning:

I’ve thought about this a fair bit, because I agree that climb points is an unfulfilling tiebreaker. The best tiebreaker I could come up with is median winning margin. There would still be some weird dynamics where teams could be guaranteed a spot in the finals regardless of their last match. However, I think it does the best job of finding the best of the tied teams.

While it’s not a sentiment I’m disagreeing on principle, the spaces in which the Postseason expos operate don’t give you enough seating if you were to put 100+ teams each across 4 fields. In Detroit, the seating for the divisions is pretty tight and you can’t go any higher or wider

We could try actually utilizing that dome that we rented for 5 days, instead of playing only 2 matches there.

With a hundred team and seven matches for each team? uhm…no…

Don’t have 100 teams per division…Get rid of waitlist teams and perhaps some district spots. I’m hesitant to say toss the wildcard as well cause I love it so much, but maybe even that.

The Einstein Round Robin is the one thing about the Championship split that I actually like. From a spectator standpoint it’s more entertaining to get to see every alliance face each other and as a team on Einstein it’s interesting to strategize for 5 different alliances.

I just think it could be improved with some minor tweaks. Einstein matches are always going to be some of the best matches for the entire year so I’m all for increasing the number of matches.

One idea I had was to do the current format with 6 alliances playing a total of 15 matches and then advancing the top 4 alliances to another round robin series where 4 alliances play 6 matches. The advantage to this over advancing 4 alliances to the traditional best-of-3 bracket is that you’re playing 6 matches instead of 8-12. I don’t necessarily think this is the best solution my point is just that there are a lot of options for FIRST to take and I’ve yet to read a suggestion that’s less meritocratic than the current format.