OPR after Week Three Events
 

The OPR/CCWM numbers up to Week 3 events have been posted, please see

http://www.chiefdelphi.com/media/papers/2174

All events are now included. Enjoy the data!

If you find any error or have any questions, please let me know.

Re: OPR after Week Three Events
 

Thanks for all the work you put into this!

Re: OPR after Week Three Events
 

Thanks for putting this together. Always fun to use these numbers to compare teams between separate events.

Noticed that the WM results page was based on the Max CCWM for teams that have played in multiple tournaments, but the Query page posts and ranks based on the average. Is that an intentional difference?

On the Query page 2056 has an OPR of 85.6 for 2nd best, while the OPR results page has them at 86.8, which would be 1st. Also curious about the difference here.

Re: OPR after Week Three Events
 

The OPR and CCWM results page can use different criteria to sort OPR and CCWM. You can put a different number in cell G3 to use other options other than highest OPR and highest CCWM.

The Query page does not use the average. It shows the world OPR and CCWM ranking which is obtained by using all the match data from all the events so far and calculate OPR and CCWM. This way all interactions through teams that attend multiple events will be incorporated in the calculations. So there is no argument about strong or weak competitions.

Re: OPR after Week Three Events
 

Does anybody know the score of qualification match 72 at New York? It affects teams' OPR because I temporarily put in 0-0 to do the calculations.

Re: OPR after Week Three Events
 

Has anyone calculated Autonomous/Tele/Hanging OPR separately?

Re: OPR after Week Three Events
 

I actually discussed this recently with one of our other programmers, which wasn't too hard to write once you have overall OPR. Although it may not be important for alliance selection and such events, the results can be interesting and may tell something about the robots' general strategies.

Re: OPR after Week Three Events
 

It's in the spreadsheets; scroll over on any (completed) event page.

Re: OPR after Week Three Events
 

Thanks. Didn't realize that.

Re: OPR after Week Three Events
 

Just went through our qualification match videos for KC, and best I can tell our average climb and teleop scores are pretty close to our Teleop and Climb OPR.

(Hard to see everything when the video cuts away, but it looks like in match 72 we went 4/4 on 5 sets and still had time to hang for 20. :) Unfortunately an autonomous error kept us from beating our personal best performance from match 25.)

Our autonomous OPR is quite a bit lower than our average contribution to the autonomous. I wonder if that is because many teams practice lining up for the center autonomous and are moved to the wing when they play with us. If they are not as accurate/practiced from that secondary location that might explain the difference.

Moral to the Story... If you are not going to pick up the extra discs in auto please practice a wing location so that you are compatible with those that will. (We are developing an extra disc wing autonomous just in case we are paired with a 7 disc center partner too.)

Re: OPR after Week Three Events
 

There's also an overall World rank, in columns AA thru AC on the "World Rank" sheet.

Re: OPR after Week Three Events
 

Interesting. I would agree though, that your speculation is right. Teams that ordinarily score 3/3 or 2/3 from the center pyramid getting moved to the wings so you can run for 7, and missing their 3 would artificially deflate your Auto OPR.

I analyzed 4343's numbers from GTREast, and I would say that our Auto OPR is right where I guessed it would be (I guessed 80% x 3/3, 20% x 2/3 for 16.8 avg, and our OPR is 16.7). Our Climbing OPR is way low though, as is 2056's. Either we didn't hang in as many matches as I thought we did, or something is skewing that, because a hanging OPR of 4.5ish suggests we're hanging for 10 less than half the time. Our Teleop OPR is lower than I'd like, but thankfully there's lots of room for improvement. A slight change to our hopper design should prevent us from dropping our 4th disk out the top of our hopper (this happened about 60% of the time on the way back to the pyramid). Several times in the interest of time, the drivers elected to shoot for the 2pt goal when an alliance partner was using the pyramid. Hard to determine what effect that had on OPR though, since its still scoring points, but saving time.

Re: OPR after Week Three Events
 

This is so true. One of the worst things a coach has to do this year is talk to a team who can only shoot three in autonomous and they need the center back of the pyramid to start.

Re: OPR after Week Three Events
 

Here's a thought:

Put a trim pot on your bot. Read it with an analog input. Use that value to trim your autonomous shooter speed.

You can easily make tweaks between matches in the heat of competition. No software changes or uploads.

Re: OPR after Week Three Events
 

Good idea.

If you're worried about a trim pot getting bumped and screwing you up, you can put a very simple text file on the cRIO to adjust your speed. You can change the text file and FTP it to the cRIO in about 10 seconds. We call it our "fudge file" (as in the fudge factor to adjust or speed/angle/etc.). If anyone wants an example of how to do this in LabVIEW, let me know and I'll either post it here or e-mail it to you.

Re: OPR after Week Three Events
 

The twitter feed says 119 for red and 46 for blue

Re: OPR after Week Three Events
 

Yet another option:

If you have 3 free buttons on the operator console, you can use them to adjust values in the fudge file. Use button1 to cycle through the available fudge values in the file to select the one you want to adjust. Use button2 to bump the selected value up. Button3 to bump the selected value down. No need to connect a development computer.

Re: OPR after Week Three Events
 

Ed,

First off, this is an impressive collection of VBA scripts!

Second - when I try to run the refresh data script, I get an error, something about runtime error ' 13' Type mismatch.

I opened v3.0, go to the Dallas page, as thats the one I'm interested in, and make sure that I enable macros and a data connection. When I hit ctrl-shift-P it starts doing a lot of stuff but then stops at that error. When I run the step into the command and do it line by line, it seems it's because it's going through the teams list (which is blank at the time) until it hits the word "match" in C353.

Any tips? Am I missing a step? I don't see anything else suggested in the instructions page.

Thanks!
Andrew

Re: OPR after Week Three Events
 

Thanks. Do you know why the match score was not shown? Did they replay that match?

Re: OPR after Week Three Events
 

I am not sure what you are trying to do. While this can be used as a predictive tool in terms of match results, projected ranking at the end of qualifying round etc., it needs data to start. It is not that good a predictive tool when the match schedule has not come out yet. Try to run the macro again after each team has at least played 3 matches. Otherwise you may get an error saying insufficient data or something like that due to the ill condition of the matrix. For the OPR numbers to be meaningful, you need at least 4 to 5 matches completed for each team. I think some people have done that study so others please chime in.

Re: OPR after Week Three Events
 

Nope - match 72 was not replayed. The recording is here - http://www.ustream.tv/recorded/29852971

Re: OPR after Week Three Events
 

Heh. OPR After week 4 events, *all of waterloo shoots to the top*

Re: OPR after Week Three Events
 

Thanks I will enter the data and rerun the OPR with the Week 4 results. It will make quite a difference for those 3 teams with 119 points instead of 0 that I entered.

Re: OPR after Week Three Events
 

Have you seen whats happening at Wisconsin?

Re: OPR after Week Three Events
 

As of now, Waterloo's mean winning score (109) is 35 points higher than Wisconsin's, and the median winning score (96) is 23 points higher. The losing scores are also slightly higher.

Re: OPR after Week Three Events
 

Furthermore, Wisconsin's highest OPR is approximately 50% of waterloos top 2 oprs.

I see nothing terribly notable about Wisconsin. Waterloo is having possibly the deepest FRC event that's ever happened outside of MSC or IRI.

Re: OPR after Week Three Events
 

My buddy wasn't comparing waterloo to wisconsin. He was just asking about wisconsin's data.

But one thing I would refrain from would be to say that waterloo is the "deepest" regional. Out of all the matches that broke 100 points, 2/3's of them had one of the top three seeded teams on it: 1114, 610, or 2056.

Also, with such a low number of teams, the top three teams inflate the average OPR data.

Re: OPR after Week Three Events
 

They're actually not artificially inflating the OPR of the other teams as they normally would be, thanks to the 277 point match.

Can any other event say that it has 14 teams, or nearly half the teams there with OPR > 20. Yes, the 3 strongest account for most of the biggest scores, but you would expect that. With 64 matches and 32 teams, 1 of the 3 will be in about 50% of the matches. Just because the deepest end of the pool is really deep, doesn't make the remainder any less deep. Any of Waterloo's top 10 would have been a top 3 at Montreal last week.

Re: OPR after Week Three Events
 

By that metric (OPR >20), the West Michigan District even is stronger than Waterloo (23/40 teams) and Grand Blanc is about the same (17/40). Not to take anything away from Waterloo of course, but most District events have been in the same neighborhood by that comparison.

Re: OPR after Week Three Events
 

Ed,

For future reference in case you encounter this issue again, you have the following 2 recourses:

1) Twitter data for that match (for example, see here).

2) If there's only one match score missing, pick a team from each alliance that was in that match. Using the Qual Match Results data, add up the scores for the alliances that team was on. Subtract that from the sum of Auto+Climb+TeleOp for that team in the Team Standings data. That should be the alliance score for the missing match.

Example calculation, using Team 1660:

Here are Team 1660's alliances and scores at New York, with Match #72 set to a score of "0":

... and here is the Team Standings data for Team 1660 from the New York event:

142.00 + 90.00 + 203.00 = 435; ..... 435 - 316 = 119

Re: OPR after Week Three Events
 

What is notable about Waterloo is that the top 5 teams all have OPR > 50, and the top 10 are all OPR > 37. Most events from this week I am checking have OPR < 37 by the time you are out of the top 5 showing a clear split of the top teams from everyone else. Of course Waterloo still has that, with the top 2 teams having an OPR > 90 :eek:

Wisconsin has a very strange OPR set. It has 29 teams of OPR > 20 and the top 10 all have OPR > 36, but no one has an OPR > 50. And I already noted in a different thread that OPR compared to average individual robot scores was wrong by at least 5 places for most of the teams I checked.

Re: OPR after Week Three Events
 

Ed,

What is the significance of the shading of the cells in columns O & R (OPR & CCWM) in the individual event pages?

Re: OPR after Week Three Events
 

They are the top 24 OPR/CCWM at the event. This provides an interesting way to compare OPR to alliance selections.

Re: OPR after Week Three Events
 

Sorry Joe, that is not what it is. A lot of times it is true because teams can tell which teams are good without looking at any numbers. It correlates quite well to which teams were drafted.

I put in those conditional formatting back in 2008 and never took it out. Basically, when OPR is bigger than average score, or CCWM is bigger than average winning margin, it will turn green. Green is good. It means those teams contribute more than their partners to get the average score. Somebody with a lucky schedule can have very high average score but OPR/CCWM which are the calculated contribution will tell you the true story in most cases. If you only look at OPR/CCWM and ignore average score and average winning margin, the green color does not add any information.

I am surprised nobody has ever asked this question. I can take it out in the future.

Re: OPR after Week Three Events
 

Tell that to: 1310, 1334, 4039, 1241, 772, 2702, 2852, 3683, 3161, and others...

This was the highest quality FRC Regional i've ever seen, all of those teams listed would do well on a field at championship. The field of 32 teams at Waterloo was historically deep, we got a robot that could do 3+ cycles a match with the 23rd pick, 1114/2056 got a long range shooter with the 24th pick, I call that depth.

At most regionals during qualifications the event can get a little monotonus with a couple poor matches in a row. At Waterloo there was rarely any slow matches during qualifications, everyone in the stands were engaged, I even saw arena staff and regular Waterloo students getting into the games during qualifications.

Re: OPR after Week Three Events
 

One thing that can explain this is the top teams at Wisconsin all had a couple really down matches (1732 got stuck under the pyramid once, had their intake bent and jammed, 2826 jammed once and I think tipped another match, 111 jammed several times, 4212 had matches where they just couldn't get locked in). I think that also explains why no team had a record better than 8-2. When those teams were working, they were far better than their OPR indicated.

Re: OPR after Week Three Events
 

I don't think Waterloo's matches "rarely being slow" has anything to do with depth of field. When there are only 32 teams, and about 20% of the robots at the event are on the field at the same time, chances are there will be at least one good robot.

Ignoring the scores of 1114, 2056, and 4069 during elimination, the average elimination score at Waterloo was 101 (121 if you included them).
Compare that to Boilermaker which had an average elimination of 97 (or 88 if you removed 359, 868, and 1747), I'd say that's similar, but I wouldn't say that BMR had that much depth.

My guess is that you'd find similar results for other regional competitions.

Re: OPR after Week Three Events
 

Please don't. If nobody has ever asked, then they weren't using it. Our team, on the other hand, is probably going to use the heck out of it in our upcoming District Event.

Thank you, Ed, for what is looking like a truly wonderful tool.