Thoughts on nationals...

Correct me if I’m wrong, but I don’t recall seeing a thoughts on nationals thread.
Here are my thoughts.

The Venue: Overall I loved it! The pits had a pretty good layout, but I would’ve changed a few things if I had organized it. The pits were easy to find teams from your division and it was nice because you did not have to walk far to find your partners or opponents. This made scouting easy. However, I did not like how the teams tables were split up. They were divided by divisions and I think it would have been easier if teams from the same schools had tables that were next to each other. I’m sure schools that brought more than 1 team per school would agree. I liked the tournament area also. The seating was good during qualifications and divisional eliminations but during finals it got a bit crowded. I liked the mounted chair setup rather than the pop-up bleachers that worlds has. But worlds needs more fields so that’s understandable. Ohh and one more thing on division. I did not like how the teams were split up. It went that the first half of your clubs teams (fake ex: 1200A-1200C) were in the red division, and the second half (fake ex: 1200D-1200F) were in the blue division. This made the blue division significantly easier than the red division since the best teams from a school are typically in the first half of the alphabet. Also, since the teams were split up like this, you could pretty much tell which division was going to win just by the division list.

Robots: There were some great robots there! I didn’t watch any of the Red division qualifications or eliminations so I only saw the 12’s team up to win in the finals. But in the blue division 323Z was definitely the best scorer. 1045F also had a great robot. It was one of the best descorers in the division and if they had a scoring autonomous (in semi’s they didn’t use it, so not sure if they have one or not) they possibly could have beat us. If they picked us for a partner the outcome might’ve been different as well ;). Those are the top 2 teams from my division but from what I heard 1064a and 3018 were good in the red division besides the 12 teams.

Suggestions for teams going to worlds: Have a good autonomous!!! This was a game changer in half our matches. We had some close matches and without the autonomous bonus or sacks that were scored in the beginning of driver we could’ve lost. And parking bonus. With matches that good teams participate in, the top score is usually around 175 and that extra 20 points is about 11% of your overall score, it definitely helps. I’m sure that has already been talked about on the forum before though.

Complaints: My only complaints are what I talked about before about the not having teams from same schools pits next to each other and splitting up divisions by the first and second half of the alphabet. I think it would be better to have it organized by every other team.

This is random but do the 12 teams or 1045F have any members actively on the forum? I am just wondering because they were great teams at nationals and I have not seen them on the forum. I have not been active on the forum this season so I can’t exactly say I haven’t seen them on the forum.

I asked some members of the 12 teams this a while back (read: last season), and it seems that their coach is the only one on the forums. He’s under the username BillW.

That’s surprising that they don’t. They have a youtube account with a lot of videos from last season. 12A is a very successful team and has been great the past 3 years.

I’m interested in seeing what they’re build area looks like.

Yes, starting a thread such as this is an excellent idea—if for no other reason than to acknowledge the tremendous job that Jim Schulte and his many able volunteers (including team 3018) as well as the Create Foundation did in running this year’s championship. Practice makes perfect, and these folks have learned how to host a world-class event. This season they moved to a new venue that was not unlike the one used for the finals at Worlds. The seating and viewing angles were all tremendous as was the proximity of the fields to the pit area.

As noted in a post above, the team split was a little odd. This resulted in a few more teams in the Blue division and probably did add a little unintentional bias. I am not sure why they used this system (I also saw this approach at the mid-west Championships), but it was a minor issue that I am sure they will review prior to next season. The extra teams may also be the reason that the Blue division had only 13 qualification matches, while the Red division played 14.

BUT did you read that carefully? Fourteen matches! IT was great. When, if ever, have you played fourteen qualification matches? Statistically, things shouldn’t change much after ten or eleven matches, but every last one of these was critical—for ranking or for schedule points. There were many excellent teams on hand, and it was simply tons of fun playing that many matches with and against them.

Better yet, was that the qualification rounds finished ahead of schedule. (How is that even possible?) This gave all the teams plenty of time to prepare for alliance selection. Having this time is really unusual but so beneficial for teams who want to have some intelligent discussion before rushing into alliance selection. (I wish all tournaments, large or small, would adopt this approach–even if there is less time after alliance selection).

As part of our scouting, we use a system generally termed OPR (Offensive Power Ranking) to mathematically assess each team’s offensive capability-- independent of partners/opponents; it is not perfect but it generally correlates pretty well with observed behavior.

At Nationals, my recollection is that the top three three teams in the Blue division did a good job with their first picks, and ended up forming pairs such that the top 6 teams ranked by OPR made up the first three alliances. The second picks weren’t quite as accurate by OPR but this seemed to be true of the rest of the division also. Regardless, as might be expected, the top three alliances did not really struggle in the division elimination rounds, and the alliance led by 323Z ended up on top.

In the Red division, alliance selection was not quite as predictable. Team 1069B allied with 3018 and 1028A, creating an 8/12/23 OPR-ranked alliance. This allowed 12A and 12B to ally with 6813M, creating a 1/2/9 OPR-ranked alliance. Perhaps most surprising was that team 4495A, which ranked 16 after qualification, but 3rd by OPR, was never selected onto an alliance. Perhaps not as surprising was that a 1/2/9 OPR alliance would end up winning the division and the championship.

In the end, though, there dozens of smiling faces on the field and in the stands, which is what really counts. These included those of the ACME #12 teams, all of whom had a wonderful time and look forward to seeing many of the same great teams at Worlds.

If you have the opportunity to go to Nationals next season, but have some doubt about whether to go, erase it and attend. It is a really, really well run event and fun championship.

There are a couple reasons that our teams refrain from posting: One is that because we have a large organization with 20+ teams, we can generally find the answer to most questions internally. Another is that it becomes difficult for anyone on the outside to know who is speaking for whom and whether they are speaking for all teams or just one. By keeping our posts limited, we avoid these complications both internally and externally.

For the record, 12A has been composed of entirely different team members in each of the last three years. In addition, 12B is every bit as good as 12A. Lest there be any doubt about this, consider that at Nationals, we ran 12B, not 12A, in the second match of the division semi-finals, finals, and championship finals. We did the same thing at the Mid-West championship.

It is nothing special–just a large physical science lab with a permanent field and seven lab tables. The hood has been repurposed as a dremel station; it also holds a small drill press. The only other significant piece of equipment is an up-right variable-speed bandsaw.

If the 3rd OPR team was still available when you were picking your 3rd partner, then why not have a 1/2/3 OPR alliance?

I am confused on what you mean with OPR. Does that mean ‘‘Out of Picking Rank’’ or is that a specific team? Well in our division there were a lot of teams who ranks were undeserved I think; In both ways, having some teams who were good be ranked bad and vice versa. We scouted out a small amount and from that small amount we found out that 12D and 4184 were our best chances. Either would have worked and I think we would have got to finals with either of them. Luckily, our other partner 1028 had an undeserved rank (IMO) so he was good for a 2nd pick. I am happy with our alliance but 12D would’ve made an excellent partner also.
Does that answer your question?

EDIT: I completely missed BillW’s post. I thought you were quoting my post, so sorry for posting my answer! I reread his and now understand OPR. I also am wondering the same thing.

I do have a question for BillW though. Does the OPR take into account the amount of points robots individually scored? I know you don’t like to reveal much but what exactly does it take into account, since the robots are assessed individually?

I think the 6813 pick was because they were able to fill in one of the weaknesses of the 12 robots, high goal scoring the match loads. They were also just a strong team overall. We won the Nebraska State Championship with them and I’m not surprised that they did well.

OPR is pretty widely used in FIRST I think so it’s easy to read up on on Chief Delphi if you want to. There’s also a vexforum thread on it here, and a detailed explanation of how it is calculated here.

Here is a simple explanation of what it is:

Notice that there are many more equations than variables. This means there isn’t an exact solution. However, there are ways of finding the best solution.

What you end up with is an estimate of how many points each robot scores, for its own alliance, in an average qualification match. Or more accurately the number of points it contributes to its alliance score, since helping your partner to score more points than they otherwise would also increases your alliance score and therefore your OPR. Unfortunately, points scored by the opponent are indistinguishable from other points so they also count towards a teams OPR. Apart from the SP-scoring issue, though, I don’t see much reason to doubt this estimate. So if you want to know who are the fastest-scoring teams in the competition then OPR is probably the next best way to do that after robot skills, and not everyone does robot skills.

However, I don’t think the number of points a robot contributes to its alliance’s score is always a good measure of team quality. Closely fought Sack Attack matches are often low-scoring. In one-sided matches, both the winning and losing alliances often get high scores because it is in both the winning and losing teams’ interests to increase the losing team’s score. Teams that play defensive strategies, including teams that use descoring a lot in qualification matches, will also end up with low OPR even if they are high quality teams.

My teams use two calculations, OPR and CCWM, to aid in scouting. You can find a brief description of each on our web site at www.cornerstonerobotics.net/power-rankings.html. From there you can view the record and power rankings for all teams at a single competition, for all competitions for a single team, or select Scouting By Competition to view all the information for teams attending an upcoming competition.

NOTE: These calculations are based on the results posted on RobotEvents.com and far from perfect. OPR and CCWM are based only on qualification matches. The results for the elimination matches are not reliable enough as far as which teams actually played in each match. Also, the automated scripts which pull the data down to my database are not correctly handling competitions with multiple divisions. Only the results from the first division shown on RobotEvents is used. I’m working on correcting this but don’t know that I’ll be able to have it competed in time for Worlds.

I believe OPR and CCWM are great additions to DIRECT, PERSONAL/OBSERVED SCOUTING. Nothing beats watching and talking directly to teams. There are many aspects to scouting that are not easy to quantify and analyze.

I am looking forward to meeting many of you at Worlds!

Jay

VexMundi and JayM thanks for providing more detail on OPR. As you note it is not infallible and simply tries to find a “best fit” solution for each team’s offensive contribution. It is not able to tease out the error introduced by teams scoring for their opponent, but fortunately it is typically not in either side’s best interest to descore during qualification rounds, which lessens the distortion that would be caused by aggressive defensive play. Rankings can be further optimized by adjusting for opponent and teammate OPR, which removes some of the schedule-strength issues .

It is also possible to calculate OPR for many teams across multiple tournaments even though many teams are never present at the same competition. This calculation is subject to early vs. late season results and other complications, but it is a way to get a handle on which teams to keep an eye on before attending an event.

OPR is simply one scouting tool that has the benefit of providing a consistent metric. In this case, the 12 teams felt that 6813M would better compliment their style of play. And yes, the scoring on the high goal was one capability they were looking for (even though they can score it also; they just rarely choose to do so).

One of the factors that I have noticed about the OPR system is that it has some difficulty in accounting for some teams that do de-score aggressively in qualification rounds. I agree that it is not in the best interest of any of the teams during quals., but some teams do not realize this and pursue that strategy anyway. If you run into some of these teams, it can really effect the final scores, especially if your alliance is not a high scoring alliance.

This is why we almost completely ignore rankings. We just watch as many matches as possible and pick the teams that we know will be the most useful on an Alliance.

If you descore aggressively and win consistently, who cares if your rank is 18/80? If you have a killer wallbot that makes the scores really low, that’s awesome. If you get carried to the top because you get lucky, why would we want to pick you?

Nothing beats scouting for yourself.

It is my experience, in events we have been to, that almost all teams (us included) will attempt to descore if they believe they are obviously behind in score during qualifying games. To not descore in this situation, teams must make a conscious decision that the win points are not achievable, so they effectively are raising the white flag. Many people are not wired this way, and will fight for the win regardless of how hopeless the situation might seem, as this is perceived as giving it your best effort.

Cheers, Paul

I meant this discussion in reference to using a mathematical approach to ranking teams and how scores can be effected by specific game play. I agree there are times when descoring is essential to the win, I have even counseled my team to descore from the beginning of a qual. match, which normally would be something I would not recommend (We knew we were set against a high scoring alliance with a weak partner). I also fully agree in some instances as your alliance falls behind, descoring is definitely required to win, which obviously is the first priority.That being said, I know of some teams that come out descoring from the beginning of every match.

THANK YOU everyone always goes by rank >.<

I also agree with you. 6252 (our first pick at Nationals) came up to us worried after qual. because they were the 4th seed. We said we didn’t care about rank, just robot performance. Teams can get some bad luck once and a while.

Tough losses are easy to come by, especially in 13 or 14 qualifiers. We won our match in the qualifiers against 6252 despite them probably being a superior etam because they had a dead controller battery, then turned around and lost to a weaker team then us because of a forgotten rubber band. Making an opinion off of just one match or just one look at the rankings can lead to picking the wrong team.

Back to original topic, 1045F isn’t active on the forum. I think this is their second year and they just do robotics as a class, not in a club or anything. They’ve had the same design all year and drove it to the top.

void rant()
{
[INDENT]In a good scouting system you should take rank into consideration, but not as the decision maker. Consider the strengths and weaknesses of your robot and try to complement them with your alliance pick. Look at their driving skills score to gauge how efficient they are at scoring. Do their autonomous routines conflict with yours? The only thing rank really shows is consistency, because it takes a robot who can preform well under sometimes adverse circumstances. Chalk it up to bad luck if you will, but if team can succeed in spite of poor luck they should be under high consideration.[/INDENT]
}

If you want a result from your rant, i’d suggest writing it in a form that isn’t preluded by void :wink: