Quote:
Originally Posted by IKE
IRI seeding algorithm:
I was actually a big fan of the 2010 algorithm with a few tweaks applied. I thought it had the highest potential for doing a good sort on teams. Here is how I would do it for the IRI:
Winners seeding points: Winner score + Loser score + Constant
Loser seeding points: 2xLoser score
Tie score: 2xTie Score for all
What about the Co-Op bridge?
Co-Op is worth 10 pts. for a single balance to both sides. Co-Op is worth 25 points for a balance with 1 red and 1 blue member.
I personally think this carries the right balance for teams. The Co-Op gets doubled for both sides of the field. If the other alliance stands you up at the bridge, you can still get substantial points for it.
Close matches will have close qualifying scores. High scoring matches will provide high qualifying scores. There is a general dissincentive for reducing your opposing alliances score (this dissincentive is adjustable by moving the value of the Winning constant up or down).
This system also eliminates the incentive for 6v0 which was controversial in 2010.
I would award the Co-Opertition award to the highest Co-Op score that is not an alliance captain (possibly alliance captain or higher seed than the lowest seeding alliance captian).
I also think that this can serve as a future scoring model for future first games if they want to continue with the "Co-Opertition" aspect. It must be mutually beneficial to both sides, it must be more valuable if both sides participate. It must have some value if only 1 side participates (this should reduce hurt feelings of getting stood up to the prom).
This style of play would work for many first games. Having a common central goal. As it ties into both teams points, with my ranking system, the common goal is a doubler for both teams. This give it equal precedence for the Loosing side (loosers get 2L), and higher precedence for the higher scoring side (winners get W+L, therefore Co-Op scoring is 2x the value of W only scoring).
Co-Op points could be added in real-time to both scores, or Co-Op could be a seperate entity doubled up for each side at the end of the match.
For this years game, I would put the "winning constant" around 25 points. This should be a high enough value for teams to go for the win.
This is apretty big tear-up to the seeding algorithm this year, but I think it would be oworth trying out at a high caliber event.
|
I like the idea, however I think it punishes defensive quite seriously. I saw multiple teams at the events that I went to with great defensive robots that shut down the competition.
The argument "If they have good defense then their ranking won't matter because any good scouting team would note their abilities." doesn't cut it either. This algorithm hurts any alliance that tries to play defense during qualifications. This may cause any potentially great defensive robots from showing their abilities off for fearing of losing valuable points.
It also brings up whether or not you want to play defense to win. For example, In a match against Team 548 Robostangs, we were forced to play defense against them to keep them off the fender, simply because we knew they could out score us. This algorithm would have me thinking twice about play defense for the win because It limits the amount of points you could obtain.
Purely Theoretical:
Option 1:: Play defense
Your score(red) - 60
Opponents score(blue) - 40
Reasoning: You play defense against the opposing alliance, starving balls and attempting to force penalties. Your score is lowered slightly due to losing a robot that could be scoring, while your opponents score is lowered severely.
Option 2:: No Defense
You score(red) - 65
Opponents score(blue) - 70
Reasoning: With no defense played against them your opponents outscore you, however you gain additional points due to having a 3rd scoring robot.
In option 1, the Winning alliance would recieve 60 + 40 + constant (let's use 25) Which totals to
125. The losing alliance receives 40 x 2 which equals 80.
In option 2 the winning alliance receives 70 + 65 + 25 which equals 160 points, with the losing alliance receiving 65 x 2 which equals 130.
As you can see, even though red lost in option 2, they obtained more points then if they had won. This is purely hypothetical and reflects no matches I've watched. I like the algorithm, I just figured that someone needed to play the devils advocate to get some discussion started.