Thread: Best Stragtey
View Single Post
  #15   Spotlight this post!  
Unread 15-03-2009, 20:56
Lil' Lavery Lil' Lavery is offline
TSIMFD
AKA: Sean Lavery
FRC #1712 (DAWGMA)
Team Role: Mentor
 
Join Date: Nov 2003
Rookie Year: 2003
Location: Philadelphia, PA
Posts: 6,654
Lil' Lavery has a reputation beyond reputeLil' Lavery has a reputation beyond reputeLil' Lavery has a reputation beyond reputeLil' Lavery has a reputation beyond reputeLil' Lavery has a reputation beyond reputeLil' Lavery has a reputation beyond reputeLil' Lavery has a reputation beyond reputeLil' Lavery has a reputation beyond reputeLil' Lavery has a reputation beyond reputeLil' Lavery has a reputation beyond reputeLil' Lavery has a reputation beyond repute
Send a message via AIM to Lil' Lavery
Re: Best Stragtey

Quote:
Originally Posted by writchie View Post
I thought that the Alliance selections at the Florida regional were a bit bizarre. If you look at the elimination results, versus the "average" seeding rank, the pattern is clear. The finals were a close match between the teams with average seeding ranks of 12.7 and 12.3. The teams that lost in the semifinals had average ranks of 18.0 and 17.5.

Spoiler for hidden to reduce length:

The way the alliance selection works, the expected average seeding for all teams should be 12.5 or better and the first seed picking 2 and 17 could average as low as 6.7, even less if teams below 17 were still available. Even the 12th seed (in the 8th alliance) can do no worse than picking 14 and 24 for an average of of 16.3.

As it turns out, the average seed of alliances in the Florida regional was 19.3 and the lowest were 12.3 and 12.7. The highest were 29.7 and 25.6. 9 Teams ranked 14 -24 (average 16.1) were left unpicked while 8 teams from 25 - 47 (average 35.2)

I would have expected teams ranked 14 - 20 to be picked, especially over teams ranked (36 - 47 etc). Several alliance captains appeared completely unprepared. Going by name recognition, low team number, loudest shouts in the stands, color of t-shirt, etc. is not likely to be a successful strategy.

Seeding for sure is not the sole reason for picking a partner. The alliance must be balanced as well. And there are often mechanical breakdowns or other reasons which explain poorer performance. But, if this were the case we would see teams with higher average ranks doing better in the eliminations. After 9 matches, the seeding is likely to reflect much more than the luck of the draw - things like pinning ability, consistency, human player ability, driver ability, penalty avoidance, etc. I think the alliance selection at Florida left a lot on the table.

The teams that have a possibility of doing 6 wins or better have to be prepared for alliance selection immediately following the last match. Your first pick can always be a team ranked 14th or higher and your second pick can always be 24th or better. You will never be able to pick a team ranked higher than you. Don't worry about them. They won't be on your list. You should know which robots ranked below you that are complementary to you. Unless you have a good reason otherwise, you may want to pick the lowest seed that is complementary.

IMHO alliance selection should account for 1) proven performance against the field (an objective measure based on final seeding rank), 2) balance for the team (shooting/defense/empty cell) and 3) experience (often reflected in lower team number).

For lunacy, it appears to me that the most successful alliances consists of an excellent shooting bot and a good defensive bot that can reliably play defense on the opponents best scoring bot and pin opponents for easy human scoring. During the early phase of the game, pinning for human scoring is essential. In the latter phase, when humans run out of moon rocks, neutralizing the opponents top dumper (or equivalent shooter) is essential. In all phases, penalty avoidance is essential. Teams that draw penalties or waste moon rocks on poor targets are usually reflected in the standings.

Just my $0.02
I think this clearly shows, as any FIRST vet will tell you, rankings are not what's important (especially in an event that runs less than 11 or 12 matches).
Last year's Championship winning alliance was constructed of teams ranked 1st (1114), 12th (217), and 57th (148) out of 86. An average ranking of more than 23. In 2007 the Championship winning alliance was ranked 9th (190), 37th (987), and 50th (177), or an average rank of 32nd.
Even in Lunacy this still applies. Look at the Cass Tech event, which ran 12 qualification matches. 469 started out 1-7, finished 5-7, and ranked 27th out of 40. Yet 469 was selected 2nd and reached the finals.

Seeding doesn't equate directly to robot quality.
While I'll agree some captains were unprepared, as they always are, most of them did a fine job in picking out who was ranked higher and lower than they should have been.
__________________
Being correct doesn't mean you don't have to explain yourself.
Reply With Quote