Success of the #1 seeds at the Championship Event

[Joe Ross]

For reference, you should read Karthik's original thread here: http://www.chiefdelphi.com/forums/sh...ad.php?t=29678. The gist is that #1 seeds usually win their divisions at championship. In 2001, 2002, and 2003 3 #1 seeds won, and 2 #1 seeds won their divisions in 2004.


Now lets look at this year.

* 1 #1 seed won their division
o Archimedes #1: 245, 217, 766
* 3 division champions were #2 seeds
o Curie #2: 175, 33, 108
o Galileo #2: 56, 254, 64 (Championsip Finalist)
o Newton #2: 330, 67, 503 (Championship Winner)
* 2 #1 seeds were division finalists
o Curie #1: 337, 1510, 85
o Newton #1: 910, 135, 537
* 1 #1 seeds were division quarterfinalists
o Galileo #1: 1089, 1280, 365

Anyone have any ideas why the #1 seeds didn't do as well this year as in year's past?

[Alex Cormier]

Originally Posted by Joe Ross

Anyone have any ideas why the #1 seeds didn't do as well this year as in year's past?

one thought i had was that the divisions had so many good team this year, the alliances had a good number of teams for amazing alliances other then just the first 2 being power alliances. It also was all about the third pick this year, that was the breaker for many alliances.

[Eugenia Gabrielov]

I think it's all about compatibility. Not to slight any alliance at all, but I think at the championship event there are so many good teams that when alliances are picked, people aren't just picking the powerhouse teams, they're picking teams they're compatible with. Seed #2 may simply have chosen more compatibly than their opponent in Seed #1. Then you take into account which seeds play each other at which point in the finals...this could be a factor in survival and essentially sets Seed #1 and Seed #2 on even ground.

[Matt Leese]

I think that this year being number one seed didn't mean what it did in past years. When you have three robots per alliance, the individual contribution of each robot is less. Therefore, a lot more of the ranking came down to who your alliance partners were in the seeding rounds than to just how successful your robot was.

Matt

[Arefin Bari]

I have to agree with Matt Leese. The most important thing in this year's game was scouting. The seedings this year totally depended on wins and loses. So if one team is paired with 2 other bots which can't cap, you lose matches, so your seeding goes down. That is when the scouting comes in play. I think it was team 66 in Curie Division who kept track of how many each team capped each match. At the end of the qualifying in Curie Division 108 was ranked after 60s... but according to team 66's scouting system 108 was 7th in capping most tetras. Team 175 (BUZZ) and team 33 (Killer Bees) took advantage of the scouting system and picked 108, and went on to winning the Division.

... As I pointed out before, Triple Play was about teamwork, scouting and making relationship with other teams.

[tkwetzel]

Originally Posted by Matt Leese

I think that this year being number one seed didn't mean what it did in past years. When you have three robots per alliance, the individual contribution of each robot is less. Therefore, a lot more of the ranking came down to who your alliance partners were in the seeding rounds than to just how successful your robot was.

Matt

Wow...my thoughts almost exactly. In past years, one robot could dominate the score during seeding matces and get ranked well. This year you had to have at least one good alliance partner. The seeding matches this year did not mean nearly as much as in the past since there were 3 robots per alliance.

[Karthik]

There are a couple of factors that caused the lessened success of the number one seeds.

As Matt pointed our earlier, in three on three matches, it's very hard for a single team to control their own destiny in a match. There are five other robots out there who affect things. In the long run, this will average out, as you'll have matches with good partners/opponents, and matches with bad ones. But with only 7 qualifying matches, in a division with 85 teams, teams were victims or benefactors of their schedules. So, many of the "best" teams did not fair very well during qualifying at the championship.

On Friday afternoon, I predicted that each division would be won by the highest ranked "elite" team. If one of the really strong teams had a good schedule, and seeded high, and was able to pick another "elite" team and in the second round get a "very strong" team, this alliance would be hard to beat.

Take Archimedes for example. After Friday, there was a general consensus that there were about 9 dominant robots. But when it came to alliance selection time, only 3 of them ended up picking. The other 6, were the first 6 teams selected. So who's going to have the best chance to win? Clearly the alliance with two dominant robots.

So after writing all that, I was about to hypothesize that the #1 seeds did extremely well during the regionals. I believed this because, I figured with the immense number of seeding matches this year at the regional level, a dominant robot would always end up rising to the top. I was wrong, here are the stats:

Of the 2005 Regional Champions
- 15 were #1 seeds
- 12 were #2 seeds
- 2 were #3 seeds
- 1 was a #4 seed

That's a good success rate for the #1 seeds, but not as high as I thought it would be. Further study is required. Alex's point about the importance of the 3rd partner is very true. This game could be won or lost on that final selection.

[Bill Moore]

Originally Posted by Joe Ross

For reference, you should read Karthik's original thread here: http://www.chiefdelphi.com/forums/sh...ad.php?t=29678. The gist is that #1 seeds usually win their divisions at championship. In 2001, 2002, and 2003 3 #1 seeds won, and 2 #1 seeds won their divisions in 2004.


Now lets look at this year.

* 1 #1 seed won their division
o Archimedes #1: 245, 217, 766
* 3 division champions were #2 seeds
o Curie #2: 175, 33, 108
o Galileo #2: 56, 254, 64 (Championsip Finalist)
o Newton #2: 330, 67, 503 (Championship Winner)
* 2 #1 seeds were division finalists
o Curie #1: 337, 1510, 85
o Newton #1: 910, 135, 537
* 1 #1 seeds were division quarterfinalists
o Galileo #1: 1089, 1280, 365

Anyone have any ideas why the #1 seeds didn't do as well this year as in year's past?

I'm not sure about the other #2 seeds this year, but in Galileo, the #2 seed (Team 56 Robbe Extreme) declined an invitation to join the #1 seed. They then went on to win Galileo, and were Championship Finalists. Did any of the other #2 seeds also decline? With so many strong teams in the division, it was easy to decline and still have a dominant alliance.

Maybe, as FIRST grows, this is a preview of how the Championship will be in the future -- so many strong teams that any alliance, 1 through 8, can win it all.

[jgannon]

Originally Posted by Joe Ross

2 #1 seeds won their divisions in 2004.

Yes, but keep in mind that the other two division winners were #4 and #5. I assume that this means that the Archimedes #1 lost in the semis, and I know that the Curie #1 lost in the quarters. The national champion was a #5. It looks about even to me.

[ScoutingNerd175]

Originally Posted by Arefin Bari

I think it was team 66 in Curie Division who kept track of how many each team capped each match. At the end of the qualifying in Curie Division 108 was ranked after 60s... but according to team 66's scouting system 108 was 7th in capping most tetras. Team 175 (BUZZ) and team 33 (Killer Bees) took advantage of the scouting system and picked 108.

Actually, Team 175 did all of our own scouting. We developed our tetra counting system at the Chesapeake Bay Regional. At nationals we had imputed the data into a team members PDA and we were able to rank it using that. Yes, when we picked 108 they were the highest scoring team left.

We did find that in this game alliance partners do make all the difference. Our first alliance pick (Team 33) was the highest scoring robot there but they had had quite a bit of bad luck and they ended up seeded around 11th or so. On the other hand, many robots that weren't as good were seeded quite highly because they had good luck.

[SoftwareBug2.0]

Originally Posted by Arefin Bari

... As I pointed out before, Triple Play was about teamwork, scouting and making relationship with other teams.

I think that teamwork and strategy were very important. I don't really know about how much it played into the 2003 or competitions before that, but I think that this year, it was much more important than last year.

When strategy becomes more important, outcomes are much less predictable. This makes scouting more difficult, as you not only need to judge the robot on its performance, but also on whether it has the potential to improve its performance significantly based if it follows the direction of your team.

[Elgin Clock]

This years game had more to do with luck in alliance pairings than any other year did. Thanks to the three team alliance.

I saw many kinds of "bad" robots that tipped or didnt do so well in their own matches gain high qualifying points thanks to their alliance partners.
But, when they were forced to pick a team in the finals, they were underqualified in scouting, and did bad.

[slickguy2007]

Originally Posted by Elgin Clock

This years game had more to do with luck in alliance pairings than any other year did. Thanks to the three team alliance.

I saw many kinds of "bad" robots that tipped or didnt do so well in their own matches gain high qualifying points thanks to their alliance partners.
But, when they were forced to pick a team in the finals, they were underqualified in scouting, and did bad.

It happens. We should be happy for those who get into a position where they are forced to learn. They will hopefully identify strategy as a weak point and work to correct it. When teams learn from their mistakes, FIRST is doing its job. For those that know their lessons and are effected by something like this, sorry to say but it just happens.

I personally liked the 3 teamed alliance because it adds to the unpredictableness of the game. It certainly is harder to work with two other teams, but then again who ever said that FIRST was going to make it easier for us?

GO 1403!!!

[Koko Ed]

Penalties played a very big part of the competition this year. You could have a domiant robot but if you're saddled with a clueless partner (or are clueless yourself) it doesn't matter how fast it goes, how high it reaches or how quick it stacks.

[JackN]

[quote=Karthik]
Of the 2005 Regional Champions
- 15 were #1 seeds
- 12 were #2 seeds
- 2 were #3 seeds
- 1 was a #4 seed
[quote]

Um.... Karthik at west Michigan the winners were the fifth ranked alliance, sorry this isn't the time or place for that.

Anywho the reason for all the #1 seeds not doing so well is that it might not necessarily show how good a team was. We were undefeated through Friday despite having mechanical troubles. I guess what I'm trying to say is that you could get put with good alliances and make it through. Also I'd have to say that a team might have picked a robot that could cap several at a time which was a questionable strategy.(except for pink you guys pulled it off sweet like butta)