1st Seeds Win

This is really the first year that I’ve actually paid attention to who wins regionals other than what I attended personally, and so I was wondering:

This year it seems like almost always the first or second seeded alliance wins the regional. I think I’ve seen only a couple of exceptions to that so far. Is that the normal thing, or have other years had more of a variation in which alliances win?

Think about it, and mathematically/statistically it makes sense. If the first seed picks the second, then the first alliance has the top two teams in the regional. This makes for a powerhouse alliance.

This is usually a normal thing, because of the strengh of the number 1 and 2 alliances… Usually the number 1 alliance is composed of the best two robots at the event. Depending on qualfiying matches, a not so dominate team can really throw a wrench into picking and really make it anyones game. (Example the 2008 great lakes regional, where the number one seed was declined 3 times, breaking up the picking of some other top 8 teams). And sometimes you just get that alliance that catches all the breaks and pulls a upset.

Come St. Louis, any alliance really has a shot at winning a divison, there will probably be 24 very good robots in each divison, making things very interesting.

This isn’t always true. It ONLY makes sense if the ranking system is an accurate measure of robot performance, which it never is. W-T-L systems, or even any systems, don’t take into account robot improvement, lucky or unlucky qualification pairings, or a number of other factors…

This is true, but the overpowering 1 is usually to much. i don’t remember the statistic, but I think it was 70% of number 1 alliances win the event.

That’s definitely true. For example, this past weekend at Niles, team 1941 ended up being the 4th place alliance captain. The problem was that their robot was basically a cart with four wheels. They didn’t even have a minibot*.

*I’m not trying to be cruel, rather I’m stating the facts. Their robot never (to my knowledge) scored a single point for their alliance. Considering that the GDC wanted Logomotion to discourage defense, I find this ironic.

Thats what I always figured. I always assumed that since the rankings never put exactly the best people in the top eight (not that the ranking system is bad, it just is impossible to account for every factor), that every one there had some chance to take it all.

Another factor is that there are only 2 minibot poles per alliance and ‘bonus’ points given to faster minibots. The top seed typically teams the team with the fastest minibot, so they typically get 1st and 2nd in the minibot race. In past years ('07, 09) three good scoreres could beat 2 elite scorers, but not this year because there are only 2 minibot poles and a limited amount of pegs to score on. Its a lot like '08, which had only 2 trackballs. If there were 3 minibot poles per alliance and more scoring pegs (so 2 robots can’t almost score on them all), you would see more upsets.

At small regional tournaments where there are only a few teams that stand out above the others, it is very difficult for low-seeded alliances to win.

However, at regional tournaments where there is much more depth in the field, it becomes more likely for lower-seeded alliances to win. The Week One Granite State Regional is one of the tournaments with a deeper field composed of many excellent teams without any dominating powerhouse teams. With a deep field, the second pick can still be a good robot, particularly with a serpentine draft which gives the lower-seeded teams the potential to make a better second pick than the top-seeded teams.

Our team (1519) has won the Granite State Regional three times, and never as the #1-seeded alliance.

In 2006, we were the first pick of the 6 seeded team, 1276, which had been promoted to the captain of the 5 seeded alliance. The alliance, composed of 1276, 1519, and 133 were the 6, 8, and #13 seeds of the tournament. In order to win, we went head-to-head with the 4 seeded alliance, 1 seeded alliance, and 2 seeded alliance.

In 2010, we were the first pick of the 2 seeded team, 1073, and joined by 1058. We went head-to-head with the 1 seeds in the finals.

In 2011, we were the first pick of the 3 seeded team, 175, and joined by 176. We met the 1 seeds in the finals.

Each of the above “upset” wins was enabled by having a strong alliance of three capable robots. Having a deep field and excellent scouting (to enable a good second pick) are key aspects of winning from a lower-seeded position. Three excellent robots playing with a good strategy can often upset two exceptional robots who have a weaker third robot.

PS: The above history of our team at GSR makes me wonder how often the 1 seeds have won in New Hampshire…

It is always possible, but unlikely. This year at FLR, the number 2 seeded alliance gave 217+2056 a run for their money with a tie game (easily the most intense match I have ever seen)

At Philly, last year, we were the 8th seed, moved up to 5th, and gave the first a run for their money with a 7-8 game.

I feel like some of the problem is that 1st always plays 8th first, which makes sense for a viewers standpoint (if the seeding was accurate, the finals will be seed 1 vs seed 2). But I have never in my life seen an eighth seeded alliance win against the first (well, except at FLR this year, where there was a red card involved)

This year at the Autodesk Portland regional, the Eighth seed beat the first seed after 3 ridiculously close matches, before losing to the fourth seed Alliance, which went on to win the regional. It was an amazing match to watch.

2009 North Star regional 8 blew out 1 in 2 matches. While our number 8 alliance was decent, it was more a factor of a team sneaking into the 1 spot and not being very prepared to make a good pick.

What I’ve noticed is this: In the past couple seasons, two good robots could hold up an alliance and win a match. The third one could play some very effective D (and some finals matches have been decided by how good the third robot is at playing defense) but especially in the quarters and semis the overall firepower of the top two robots in the top two alliances is too much for the others.

This perhaps is a side effect of FIRST’s decision to design really offense-oriented games since 2007. That year was crazy: an alliance could score 256 points in one match and 0 the next, all depending on how much defense was played against them. Rack 'n Roll was, in my opinion, the only game where three good robots could beat two great robots and one not so great robot, which resulted in a lot of 8 over 1 upsets. This phenomenon was augmented by the fact that since defense could shut down many offensive teams, the best teams often would not seed first. Even when they did, they often lost if the regional or division was stacked enough to provide power to the #8 alliance.

Look at the TBA results for GLR and West Michigan… more wins for blue than red in the elims. Oh look… 1114 and 67 lost in the semis due to amazing defense and the fact that 57’s robot could not provide the defense necessary to stop the opponents from scoring. In 2007, none of the #1 alliances at Champs escaped the divisions and reached Einstein. The #8 alliance of 190, 987, and 177 won in the finals. This could be interpreted as proving my point that 2007 was the only recent year where the #8 alliance could be the best alliance at an event.

However, back then there was this huge discussion about how overpowered defense was in FIRST and how boring it was to watch robots bump into each other instead of score. There were also instances where teams played defense that was too rough and complaints about how the serpentine draft gives an advantage to the #8 alliance and should be done away with. Perhaps as a result, FIRST has designed games in recent years to be based on offense, so the good teams can seed high in qualifications and be less hindered by defense in the elims. These offense-powered games in recent years have given the #1 alliance more of a chance to dominate a competition.

Very good points here! Depth and alliance partners are key to the outcome of a regional and alliance and whether or not the number 1 seed makes it all the way through!

Come to think of it, the only time the number 1 seed didn’t win GSR was 2006, 2009 (team 238 was number 1 seed, picked 319 and 562, and were semifinalists), 2010, and 2011. All other years that I can remember (03, 04, 05, 07, and 08) were won by the number 1 seed alliance. 5/9 chance the number 1 seed wins is why I love attending GSR!

This happened at Waterloo this weekend too. 2361 was the 5 alliance captain, despite being a box on wheels with NO means of scoring. I’m not sure if they had some scoring mechanism that they took off at some point over the weekend, but when I saw their robot, I couldn’t believe they had made 5 alliance captain. Luck of the draw I guess.

Apparently the number 1 seed has never won the West Michigan District event. Granted, FiM has only been around for 3 years.

This analysis is quite interesting about 2007.
I made a similar one back after that season, but for different reasons.
In 2007, I noticed that a lot of the lower seeds were winning due in large part that they had 1st pickings of the “ramp” bots that gave a relatively large bonus for winning matches. With proper defense and pinpoint scoring to block the multiplier on certain pegs of the rack, they could come back and beat you on the bonus.
The bonus back then was much more difficult than this year.
This season, your bonus is independent of an alliance partner and could be done entirely by yourself. Everyone can technically have the same type of 1 to 1.x sec minibot. The formula to creating one is out there and can be done independently of your robot between regionals. You would not (nearly impossible for most) be able to go from a non-bonus bot to one all of a sudden in your next tournament.

IMO, that game gave every alliance the best chance to win a regional. You had to be very, very careful on who you picked in trying to beat out the other alliances.

Looking at the data for #1 alliances going back to the Palmetto Regional’s inception:

2004 (343/1402/665): Out in quarterfinals (three matches), #2 won
2005 (1251/25/301): Finalists (three matches), #2 won
2006 (68/180/1028): Semifinalists (two matches), #6 won
2007 (1251/1758/1626): Finalists (three matches), #2 won
2008 (343/342/393/804): Finalists (four matches), #7 won
2009 (3025/2815/1379): Finalists (two matches), #3 won
2010 (343/1261/1398): Champions (three matches)
2011 (180/2363/2815): Champions (three matches)

2005 was the start of 3v3 play, and 2006 was the start of the serpentine draft. Take from this what you will.

Here is some previous years threads with data:
My 2009 prediction & results of 50% Alliance #1
2008 with some '07 & '06 data

The Philadelphia Regional is an interesting case as far as #1 seed vs the field. Last year *(http://www.chiefdelphi.com/forums/showthread.php?p=941786#post941786) due to the new seeding algorithm (also contains my 09 spreadsheet). This a major streak breaker because before 2010, no #1 seed had won Philly since 2001. Philly tends to have a large similar top tier of offensive robots and lots of defensive bots that causes a lot of upsets during quals and elims.

I don’t have the % of #1 Alliance champions in 2010, but I suspect it was higher than usual. The scheduling algorithm (as much as I disliked it) did a good job of seeding the top robots high regardless of schedule difficulty. This year with a return to WLT system, I have been more interested in the how OPR correlates to event winner. The team with the highest OPR during the quals almost always wins the event even if they are not in the #1 alliance (won all the Week 1 events). In Chesapeake where the top team didn’t win (one of the few cases), the alliance with the highest combined OPR did win.*

No, it’s not always like this.

This year, especially in many of the weaker regionals, you’ve seen only a bare handful of robots that can deploy minibots consistently. With the minibot race consisting of more than half the score of a normal match, putting two teams together who can both deploy minibots is an advantage that is nearly impossible to overcome.

As the season progresses the games will begin to rely more on hanging because more robots will have minibots. I would expect Michigan State Championship to be that way, however I would expect the National Championship to be slightly less so since so many teams attend because they can pay the fee without having to qualify based on robot quality (not that there’s anything wrong with that system - I’m just saying).