Highest Seed Statistics?
 

I tried to see if I saw these stats in any other threads, but didnt see them, so I put them together (Im sure someone else has but again I couldnt find them).

So I was reading the threads about intentionally loosing matches and trying to think about the importance of Rank this year and how I had seen regionals play out. So my quick pass seemed to reveal the following stats for the winning alliances:
1st Alliance Won: 22 Regionals - 54%
2nd Alliance Won: 8 Regionals - 20%
3rd Alliance Won: 4 Regionals - 10%
4th Alliance Won: 2 Regionals - 5%
5th Alliance Won: 0 Regionals
6th Alliance Won: 1 Regional - 2%
7th Alliance Won: 2 Regionals - 5%
8th Alliance Won: 2 Regionals - 5%
(yes I know my rounding leaves me off 1%)

So more than half the time the First seed won the Regional. Nearly three quarters of the time the First or Second seeds won the Regional. If you look at the alliance selections, I would bet that most all the teams picked a hurdler as their first selection.

Is this game so heavily offense weighted in the regionals that its nearly impossible for the lower alliances to form strong enough alliances to be two strong robots? It would seem so by the numbers.

I realize that obviously there are ways and strategies for these lower 4 seed teams to beat the higher teams, we saw it in 5 regionals. But 5 out of 41 makes me think that its not just strategy, that its really hard to beat two strong offensive robots.

Im looking forward to championships where perhaps the fields will be deep enough that we will see the higher seeds de-seated more often (yeah they deserve to be there, but dont you always have the urge to root for the underdog??).

I was wondering if anyone else examined this and if anyone had comparisons for past years... is this game really TOO offense based? or is it something else? or is this typical?

Re: Highest Seed Statistics?
 

That is very interesting as you have shown.
I'm sure if you compare it to last year, might be more skewed towards the lower seeds a little.
Bonus points was a major factor last year, not only in winning matches, but in design of robots as well.
Overall, it does not look so surprising other than the fact that a couple of 8 seeds won. I was there to witness MORT do it at Chesapeake. But that was because their robot and driver were on steroids during elim. matches. :P
Just kidding, it just improved so vastly!

Re: Highest Seed Statistics?
 

With only 2 trackballs on the field, the first pick which is usually another hurdling robot is a lot more important than the second pick which is usually reserved for defense or lapping this year. I would assume that it's a lot easier for, hypothetically, 1114 and 217 to team up with Rookiebot for the laps and beat a number 8 seed who got to pick 148 in the second round but had the last pick in the first round for the hurdler. Since the first seeded alliance is probably going to be a strong hurdler (probably) you'll see them pick the other top hurdler 9 times out of ten and make a very strong top seed. I think that makes the most sense.

It would be interesting to see if this was also the trend in 07 and other previous years though...

Re: Highest Seed Statistics?
 

Leaving out the discussion of whether it's actually a problem that the #1 seeds are winning, which is another massive debate in itself, I don't think it's because the offensive nature of the game entirely. While 2007 may or may have more success from lower seeds (our 2007 Champs were #8 seeds), I'd bet (and will try to prove once I have free time) that even more #1 seeds won in 2006. With the nature of the Aim High game, a 2 shooter/1 defender alliance was very very successful, and while a few defenders excelled, you could typically get a very worthy defender at the last pick of the draft. It's not the offensive nature of games, it's just the way the games play out.

Re: Highest Seed Statistics?
 

Obviously varies by year but if my memory serves me well this year is much more weighted towards the top alliances winning versus the last 2 years of serpentine Alliance selection. My impression is last year was much more spread out and 2006 was somewhere in the middle. I would also say the seeding is better (top teams seed higher) this year without the algorithm and heavy defense interfering. 2 great robots can do a majority of the scoring this year (since there are only 2 trackballs) so a lower third partner pick doesn't matter as much this year. I am sure some numbers exist (maybe not for whole years) from the last few years.

I too look forward to seeing how the deeper field of hurdlers works and less qual matches (thus worse seeding) affects the number of upsets in the elims. The deeper field seemed to lead to more quarterfinal upsets at GLR. I still think the top teams will be favored quite a bit, but not as much as regionals.

Edit: Here is a #1 seed thread from 2006 (many of the people who posted here posted there Kim, Sean & me). From 18 regionals, #1 Alliance won 9 (50%). I would guess that some the non-#1s were more spread out.

Re: Highest Seed Statistics?
 

Without looking at data for 2006, I will agree about more top seeds winning.
Raider Robotix had a lot of teams play defense of them (my shining example) being successful many of times for most of the match........too bad it only took about 10-20 seconds in certain instances for them to beat you. Once they locked and loaded, it was over. The way that teams scored the bulk of their points that year happened in spurts rather than a continual thing throughout a match.

Re: Highest Seed Statistics?
 

Ok, I found a real quick way to get some data from the last 2 years. I pulled up Karthik's Championship scouting sheets from 06 & 07 (amazing how they are still useful:) ) I sorted by Results and Draft Pos and counted the alliance captains. It doesnt cover all the regionals, but at least the majority that go to Championship. Here are the numbers
Alliance 8 never seems to win (or goes to champs) which is odd. I'll let someone else go searching through TBA (2007) and FIRST History (2006 *shudder* that will take a while) for the full data set.

Re: Highest Seed Statistics?
 

Alliance #8 came within inches (literally; a successful 30 pt. lift would have done it) of beating alliance #1 in the quarterfinals at Davis in 2007. Had they beaten alliance #1 they would have gone on to win the regional. The quarterfinals at Davis had the highest tie-score of the year, 92 to 92.

Still more amazing was that they created such a strong alliance. Davis is not a very competitive regional. (I'm not even sure alliance #8 had a hurdler in 2008). The fact that they created such a strong alliance is testament to the importance of scouting and strategy.

The championship is a whole different story. At least it was in the past; this year it seems the two top hurdlers will be able to dominate.

Re: Highest Seed Statistics?
 

A-HA! Now we know who it is!!!! :)

-dave



.

Re: Highest Seed Statistics?
 

Dave,
Its so funny you pointed it out. When I read that post, I thought about that for a second.:D

Re: Highest Seed Statistics?
 

Is this data for going to championships meaning as in the Finals?
Or actually winning a regional/Championship Event?

Re: Highest Seed Statistics?
 

The data is taken from Karthik's Championship scouting database for 06 and 07. So only teams that are going to the Championship Event are included in the spreadsheet (not all regional winners but most of them). I only looked for regional winner alliance captains and counted them.

Note: I try to post "Championships" instead of "Nationals" but here it has lead to confusion.

Re: Highest Seed Statistics?
 

The dominance of the top-ranked alliances certainly puts the hammer on the argument that the serpentine draft was taking away the benefits of being highly-ranked in qualifying, doesn't it?

Once again the Pareto principle rears its head. Eighty percentage-ish of the tournaments are won by 20%ish of the alliances. More or less, YMMV.

Re: Highest Seed Statistics?
 

I don't think so. After championship sometime I'll try and collect full data sets for 2005 as well for this, the last year pre-serpentine. I think the #1 seed winners would actually be much much higher than the ~50% we're seeing during the three years of serpentine.

Re: Highest Seed Statistics?
 

As a fan of the leveling effects of the serpentine draft, I would be a hypocrite to not agree with you. It's just that if we want the top alliance to win every single time, why bother even having eliminations? I would much rather watch an event where the outcome was at least slightly in doubt, and the serpentine draft enhances that. My point was that the serpentine draft has not eliminated the advantage of the top-rated teams, it has just slightly reduced that advantage. In sports terms, I like salary caps, giving the first draft pick to the worst team, and making all Yankee pitchers wear ten-pound weights on their pitching hand.