Re: When will Divisions be released?
 

I think it would be more accurate to say that OPR accounts for *no* way of scoring. OPR only compares total points scored to the teams in the match. OPR can't be used this year to accurately approximate the actual number of points scored by a robot in a match, due to the inseparable nature of this game and cooperative scoring. Whether or not it's a useful approximation of match contribution is the subject of much debate :)

Re: When will Divisions be released?
 

This is what threw me when looking at the equations for OPR. It seemed to me the game is too integrated to get a number for each team.

Re: When will Divisions be released?
 

Right, technically you aren't getting any numbers on the team, so it literally does not account for anything the team actually does because it is a projected statistic based off of the alliance score. So you could very well have a team who had favorable alliance partners all throughout qualifications, which results in an inflated OPR. This actually happens more often than you may think.

Re: When will Divisions be released?
 

Having watched a bunch, it happens a LOT! Which is why I believe OPR (having just tonight been exposed to it) is not a great way to sort the teams this year. I'm not sure my preferred way to rank is better, but I would rank by assist points.

Re: When will Divisions be released?
 

It's the team's rank based on opr. I updated the graph.

Does it add to the curse if I say this proves it *has* to be the year to end the curse....? :-)

Re: When will Divisions be released?
 

Although OPR doesn't tell you anything about a team's specific contribution to the alliance score (truss, finisher, auton, etc.) it does a decent job of showing which teams bring success with them when they show up on the field. Having strong alliance partners may improve your seeding, but it does not necessarily raise your OPR since the math may "credit" the scoring to your partners instead of you. Since the scoring in this game depends so much on how the alliance works as a unit, a team could achieve high OPR by planning effective match strategies that makes their partners contribute more effectively than they might do otherwise. Having a good robot helps, too.

Re: When will Divisions be released?
 

It is interesting how this is both the quantity of teams greater than <average> + <1 standard deviation> AND the number of teams who will be in elims on Saturday.

Re: When will Divisions be released?
 

So, does that mean the wait list spots are all taken??

Re: When will Divisions be released?
 

Yep, FIRST is showing that 400 teams are now registered for the 2014 FRC World Championship. The lucky 400th team was none other than Team 48, Delphi E.L.I.T.E., from Warren, Ohio. They are in the Newton Division.

Re: When will Divisions be released?
 

Statwistics. :-)

Seriously, why is thread still so active.
This is great analysis, but shouldn't it be in the Einstein 2014 thread?

Re: When will Divisions be released?
 

Why would 100 teams per division be the target number? It requires 2 teams to play surrogate matches. Would much rather see them cap the event at 96, or 102. When you have a surrogate match your stats lag behind the other teams in your division in real time. Real time stats seem to be important to FIRST considering they now include current rank on the live match scoring display.

Avoiding surrogates does not appear to be a goal however. Of the 6 years we have qualified for champs, the number of teams in the division has never been divisible by 6.

Archimedes 2009 (Atlanta) = 87 teams
Curie 2010 (Atlanta) = 86 teams
Curie 2011 (St. Louis) = 88 teams
Curie 2012 (St. Louis) = 100 teams
Newton 2013 (St. Louis) = 100 teams
Newton 2014 (St. Louis) = 100 teams

Re: When will Divisions be released?
 

anybody know for sure how many wait list slots there were?
We were finalists in two regionals and won the quality award and
the industrial design award this season.
Thought we had a pretty good shot at the wait list.

Re: When will Divisions be released?
 

Based on these calculations, it looks like there were 6 waitlist slots. A small number was expected, and a few weeks ago, Frank even said there might be none. As far as I know, the waitlist is generally first-come, first-serve, meaning that how you perform throughout the season has no effect on whether or not you'll get in off the waitlist (unless you qualify by performance, of course). It all comes down to who has the quickest trigger-finger on registration day. However, they might automatically filter out some teams, since none of the waitlist teams that got in this year attended World Champs last year. I know that 5 out of the 6 waitlist slots this year went to 4176, 357 714, 2073, and the lucky 400th team, 48. I'm unsure who the 6th waitlist team was, but they're registered in the bunch somewhere.

Re: When will Divisions be released?
 

For example, we had an OPR of 107 at SVR, but we averaged making only one high goal teleop shot each match. The OPR reflected our focus on assist scoring on the total score.

Re: When will Divisions be released?
 

An inflated OPR happens if you have a favorable set of alliances AND your alliance mates score at least their average amount in those matches. However, we had a favorable schedule at Curie last year but our alliance mates scored below their average (about 29 points per team) so our OPR was depressed well below our actual goal score. Another is if the entire field at a regional is scoring well above the FRC average, which could be occurring for any number of reasons, including quality of play or more fouls.

Re: When will Divisions be released?
 

My guess is that the pits are too small to safely accommodate 408 teams, but FIRST doesn't feel it's fair to cap Championship at 384 teams.

Re: When will Divisions be released?
 

Looking at relative ranking through the entire distribution (1,2,3,4), it looks like Newton is the strongest, never falling below 3rd and usually 1st or 2nd. Next is Galileo which is often 1st particularly in the mid-range and only at 4th for a couple short periods. Archimedes is the weakest never rising above 2nd and usually 3rd or 4th. Curie is 3rd with the strongest upper end, which may contributed most to its performance on Einstein.

Re: When will Divisions be released?
 

Considering that elims are played almost exclusively by the top 24 teams in each division, having a better top 25% of teams in your division seems to matter much more than how strong your division is below the 24th spot. Curie clearly has the advantage when it comes to comparing the teams in each division that will likely make elims.

Re: When will Divisions be released?
 

My guess is that someone thought 100 teams per division was too nice of a number not to use.

Re: When will Divisions be released?
 

Top 32% this year. Unless you believe that the 3rd pick is inconsequential.

Re: When will Divisions be released?
 

I said top 24 because only those teams are guaranteed to play in elims. Chances are that no back up teams will be called, or only 1 - 2 at most so I figured it wasn't worth projecting elims based on robots that probably won't play.

Re: When will Divisions be released?
 

You may be right but just as a counterexample, we were picked for elims twice at IRI. Both times we were the final pick on the alliance. Both times we played the first match.

Wait list registration first opens to teams that did not attend Champs the year before. It then opens to all teams a few weeks later. So, that's the "filter".

Seems like a safe bet that we will rarely if ever get to that second group again, unless they move Champs to some location that causes a lot of invitation declines (which would be a shame).

Re: When will Divisions be released?
 

Einstein isn't decided by putting the top 25% from each division on the field at the same time (now, THAT would be interesting to watch!). Each division sends the alliance that demonstrated the best ability to win. Perhaps that was the two best robots in the division plus a sleeper, perhaps it was the combination that worked best together. Perhaps they just got lucky. Most likely it will be a combination of those. Winning alliances don't reflect the average quality of the top 25% of the division. They just reflect the ability of the 3 bots they put on the field. Any division can win it.

Re: When will Divisions be released?
 

400 being the target never made sense. You think you would want 408 teams, especially evidenced by the fact FRC pits last year had at least 8 empty spots in the back of Archimedes.

Anyone know if we are getting preliminary schedule this year?

Re: When will Divisions be released?
 

In 2012 they played 9 qual matches so there were no surrogates. Now with 10 matches there are 2 surrogates per division in 167 matches.

If they like the 10 match format with the all day Thursday quals they may just plan for 102 per division. I don't think FIRST came into this year planning on 10 qual matches at Champs. Last year they were clearly planning on 9 matches but had to cut it to 8 due to belaying, manual disc count, etc... I think the pits can take 8 more teams, and the new schedules can take 3 more matches for an even 170 matches per division. MAR championships just did 83 qual matches on a Friday (which I am pretty sure is a record for qual matches in a day).

Re: When will Divisions be released?
 

Well I have heard that it is not first come first serve so I dont know what is the truth on that.
We did not attend champs last year and I got on the wait list in the first 15 seconds after it opened.
But, still no luck.
Best of luck to all you attending and above all, have a great time!!!

Re: When will Divisions be released?
 

Very much agreed with those who have said that OPR is pretty flawed for comparing team strength and those who have said predicting the World Champion by comparing divisional strength is silly.

As a result, I updated the aWAR spreadsheet so I could use that to compare divisions... :-) See the plot of the top 32 here (Attachment 16869).

By aWAR, the best division is almost unanimously Galileo. It doesn't matter if you look at the top 16, 32, or the entire division... aWAR strongly favors Galileo. Curie and Newton are nearly identical, regardless of which metric you look at across what population. Archimedes is the 2nd best division if you look at the top 16 or 32 (except by median), but if you look at the entire division it drops to last. The main reason for this is that it is comparable with Galileo in the top 4, comparable with Curie and Newton for the top 7-15, but then drops to last for the rest of the division.

My spreadsheet for looking at the divisions is attached here (Attachment 16868).

My post with the updated aWAR spreadsheet is here (http://www.chiefdelphi.com/forums/sh...2&postcount=32).

Re: When will Divisions be released?
 

I have a day job, so I haven't put the numbers together -- and someone's probably already done the analysis somewhere anyway -- but I'd be *very* surprised if the 24 teams with the top OPR in the "regular season" are always the teams in elims. Probably a large percentage of them, sure. But even if you discount high-OPR teams that break during quals & therefore aren't available for elims, not all of those teams end up in elims.

Also: I don't think that the 4th alliance robot should be considered a "backup". They're going to be a full-fledged member of the alliance, and I would expect (especially on the top 2 or 3 alliances) that there's going to be very little difference between the 2nd & 3rd pick. Seems to me that there's a lot of room for picking for varying tactics, not just emergencies. Having said that, I've never been to IRI, so I'll ask -- how often does that 4th robot not play unless one of the first 3 are broken?

Re: When will Divisions be released?
 

I know we rotated our 4th bot into play at the 2013 IRI even though the first three were still fully functional.

Re: When will Divisions be released?
 

That would be entertaining, but I think we'd need a bigger field.

My statement was not absolute at all, it's simply a trend. Robots that have a higher OPR tend to win more often, I think we can all agree this is true. Curie division has the highest OPR's in their top 24 robots between all divisions. This is also true. Therefore it follows that it is likely that curie division will on average produce the alliances most likely to win. I think it logically follows that this statement is cogent and is more likely true than untrue.

Re: When will Divisions be released?
 

We were the 4th pick at IRI last year and played in all 3 of our QF matches. Our partners did have problems but we would have been in to block 148's full court shooting either way.

The divisions are very deep with teams that can really help an alliance. There may only be a few world class power forwards but there is no shortage of teams that can help get to three assists and play defense. The 4ths on a lot of alliances will get playing time for a variety of reasons. On the top seeds it makes sense to draft two teams with similar abilities. They can give them both a chance in the QFs and figure out which one fits better with the other two robots. Lower seeds might draft for riskier strategies such as catching or a robot with a lot of potential but lacks consistency.

Re: When will Divisions be released?
 

Last year (IIRC) they released match schedules (with all the caveats that they were preliminary, tentative, subject to change, etc.) prior to the event. Does anybody know if they plan to do that again this year? Does anybody recall when the match schedules were released last year?

Re: When will Divisions be released?