How often should the #1 seed win

If an FRC game was perfectly designed, how often should the first seed win? There are some statistics from previous games here:

The notion of a perfectly designed game is left the poll voters interpretation. Should the best robot always be seeded #1 in a perfectly designed game? Are upsets exciting and should they be a part of a perfectly designed game?

If an FRC game was perfectly designed, how often should the first seed win?

  • 0-10%
  • 10-20%
  • 20-30%
  • 30-40%
  • 40-50%
  • 50-60%
  • 60-70%
  • 70-80%
  • 80-90%
  • 90-100%

0 voters

I mean here the thing about this question even in a ideal game, it depends so much on what teams are attending the event, what are there robots capable of, what there financing is like, how they are taught, etc.

For example if the every event was like IRI (Indiana Robotics Invitational), then of course it possible for any of the alliances to topple the 1st seed (if they have the right strategy that is). But at at certain district or regional, some events can just be clean sweep by the first alliance since they usually don’t have competition if they just have the best bot in general and no one can oppose them.

4 Likes

Indiana events in general are hit or miss with the 1st seed. I think it was 2017 where not a single 1st seed won in Indiana

1 Like

In an ideal game, all robots are equal so there would be a 1/8 chance.
Correct me if I’m wrong here.

6 Likes

This is very wrong the whole point is that they are rewarded for doing well in quals.

2 Likes

Sounds like communism to me.

23 Likes

CHS is very similar with number 1 seeds having about a 30-40 percent win rate for the past 4 years.

Depends on the number of qualification matches and if extra RPs are given for tasks other than wins.

The best alliance should always win. Regardless of what their rank is. If 1 is the best and they play as such they deserve it.

2 Likes

I put 70-80%. It really depends how deep the event is, and how long qualifications runs.

1 Like

I think that’s where we get the 1st seed 50/50 from. Super deep field with 12 quals matches

I put 50-60 because in general the 1st seed should be able to get through quarters and semis pretty easily. In finals they will generally face 2 or 3 which will be a much more even match which makes it possible for them to lose but it really depends on the region/competition. Here in cali where you generally have 2 powerhouses teaming up and destroying everybody else but in other regions that’s probably different.

I mean one game that meets your definition of an “ideal game” is an 8-sided dice roll.

The benefit of having a game where the odds of winning are better for the #1 seed is that the game rewards good teams. It incentivizes teams like 254 and 118 to build good robots, and teams just a tier below them to try to improve, etc. If everyone had an equal chance of winning, then 254 could achieve the same results while producing 6662-quality robots. Everyone having a chance to win is a trade-off.

I think game design can only do so much in balancing the win rate without sacrificing more important things.

1 Like

He did say perfectly designed.

1 Like

As this is a competition, not a random event, teams should come to event with unequal capabilities, meaning that the top seeded team should definitely have a much greater than 1/8 chance of winning the event, especially as they get the first choice of an alliance partner. I would expect the #1 alliance to win more events than any other alliance; if others were higher across events, it would be an indication that the qualification game is not a sufficiently good indicator of elimination performance*. On the other hand, the game should have enough variation in strategies and synergies that the #1 alliance winning should be far from a certainty. I think that a well designed game would likely have between 1/3 and 2/3 of events won by the #1 alliance. (I voted for 50-60%.)

* For example, if the bonus rank points were too dissimilar from eliminations scoring.

3 Likes

I feel like the 1st alliance should have a higher chance of winning, and the upper alliances should make it further than the lower alliances, but is it really fair that the 8th alliance be forced to play the 1st alliance? Maybe the 1st alliance should play the 5th alliance, 2nd plays 6th, 3rd to 7th, and 4th plays 8th, then going down from there. I don’t know, interesting idea at least.

I mean, ideally we would do a Round Robin style, where everyone plays everyone else. But this would take 28 matches (I believe). In the current setup, the maximum amount of matches would be 21, with a minimum of 14. So a Round Robin setup would take up to twice as long.

2 Likes

I think gellnick was taking the statement ideal game and taking it to the logical extreme as a joke. I doubt he actually thinks there will ever be a game where it is 100% equal.

1 Like

Yes. If you have too many alliances for a round robin, folding seeding brackets are the way to go. Teams that did best in the “regular season” should have the easiest ride, those that just barely qualified should have the hardest. Look at it this way: why should the #5 alliance have a harder first round than #8? NCAA March madness uses them, tennis uses them, just about all one-on-one sports* use them for playoffs. NFL modifies the brackets so that matches are dynamically assigned based on the original seeding, but this actually enhances the concept that it’s highest against lowest, just above middle against just below middle.

* That is, pretty much everything but measured and judged events. (I include races as measured against the clock.) Looking back at FRC games, 2001 and 2015 were “measured”. All other games since 1999 were “one-on-one”, at least in elims/playoffs.

4 Likes

The #1 seed is supposed to be the best robot at the competition based on merit as defined by the ranking system. Theoretically, the #1 seed is supposed to win every competition assuming every alliance is roughly balanced by the pick order.

That said, the competition is meant to give every alliance a fair opportunity to win. Otherwise, competitions would end after qualifying rounds and give the 1 seed an automatic win. So, the answer to OP’s question has to be somewhere in the middle of these extremes.

I don’t think there’s any “perfect” percentage. In the end, one alliance will win and it’s usually the alliance that plays the best over a relatively small number of matches. Why change the system? Why is there any justification to say “1 seeds should win more or less than current”? (Unless the 1 seed won almost every competition, then I could see a good reason to change the system)

1 Like

My big requirement would just be that the average win probability is strictly decreasing with seed. If you don’t have that then you get large amounts of sandbagging and accusations of sandbagging.

As long as that property clearly exists I’m kind of meh (but note that neither 1 seed win rates of 100% nor <1/8 can have this property).

7 Likes