Week 4 OPR - All regionals and districts

[Jaxom]

Originally Posted by Jared341

This is definitely a limitation of OPR. I am not sure how valid it is to compare OPRs across events at all, for that matter. I am a much bigger fan of CCWM (OPR+DPR), since it better reflects contributions to score DIFFERENTIAL rather than raw score itself. Many top teams leave a lot of potential points on the field once they see they have built a big lead so that they can ensure a higher success % on the Coopertition Bridge.

Don't the alliance partners also influence the OPR? For example, if you're at a tournament with a large number of robots that can't score, HP & TP will be lower than they would be at a different tournament. I will admit that I don't know exactly how OPR works, but this seems to be the case, at least for our results. Our GCK OPR is much higher than what we showed in STL and I see similar, lower, OPRs for other teams that went to the same two tournaments.

I in no way want to demean any particular tournament. St. Louis is a great regional -- well-run and in a nice facility. The teams there were all fantastic; it's just that there were a higher percentge of teams that couldn't score.

[Chris Hibner]

Originally Posted by Jared341

... I am a much bigger fan of CCWM (OPR+DPR), since it better reflects contributions to score DIFFERENTIAL rather than raw score itself...

I usually take CCWM with skepticism, unless I'm looking for a defensive team or strategic specialist. I've seen instances that strength of opponents really impacts CCWM, but doesn't have much affect on OPR (unless it's a highly defensive game). In other words, if you play a tough schedule and lose all of your matches 85-80, your CCWM will suffer because of the number of losses. The OPR will show that you can put up a lot of points and the losses aren't likely your fault.

[Thomas DeSilva]

Originally Posted by Jaxom

Don't the alliance partners also influence the OPR? For example, if you're at a tournament with a large number of robots that can't score, HP & TP will be lower than they would be at a different tournament.

This is partially true. Your team's OPR is the least-squares approximation of how many points your robot can score in a match, which is based both on the number of points each of your alliances scores, and who was on each alliance. This basically means that if you're always on high-scoring alliances, you'll probably have a reasonably good OPR, but your OPR will be higher if your partners have lower OPRs and vice versa. (I can give a more detailed mathematical explanation of how OPR works if you like, but I think that post has probably been made already somewhere around here.)

[MisterG]

WI regional standouts (condensed from OP data):

OPR
2481 Roboteers 33.247 4th overall 1st WI
2169 KING TeC 30.098 9th overall 2nd WI

Hybrid
2481 Roboteers 11.7361 1st overall 1st WI
2062 CORE 9.89023 9th overall 2nd WI

Tele
2169 KING TeC 22.3849 1st overall 1st WI
2826 Wave 12.5662 13th overall 2nd WI

Bridge
967 Iron Lions 11.2864 4th overall 1st WI
1732 Hilltoppers 11.2684 6th overall 2nd WI

Nice to see a high level of competition in the Miltown!

[Travis Hoffman]

Originally Posted by Lavapicker

Even Lebron and Koby don't win it every year which gives us some solice.

LeBron has yet to win "it" in *any* year.

Cue Koko Ed...

I'd love to know a way to determine elimination round OPR, but it's probably just as easy to post videos of the matches and say "Here, this is what we did". Go figure, I did that.

We serendipitously "found something" in the semis and finals in Wisconsin, and I think our driver is eager to continue it. I'm very much looking forward to Queen City.

[Ether]

Originally Posted by Travis Hoffman

I'd love to know a way to determine elimination round OPR

For any given event, the alliances in the elims don't change. If you attempted to setup and solve Ax=b, you would find that A is rank deficient. There is no unique solution.

Translated into English: elimination data provides no insight into which team on the alliance is doing the scoring. So you can't calculate OPR from that data alone.

[Nemo]

Originally Posted by Travis Hoffman

LeBron has yet to win "it" in *any* year.

Cue Koko Ed...

I'd love to know a way to determine elimination round OPR, but it's probably just as easy to post videos of the matches and say "Here, this is what we did". Go figure, I did that.

We serendipitously "found something" in the semis and finals in Wisconsin, and I think our driver is eager to continue it. I'm very much looking forward to Queen City.

Given good enough videos, you could simply calculate the exact number of points scored by each robot.

[Travis Hoffman]

Originally Posted by Nemo

Given good enough videos, you could simply calculate the exact number of points scored by each robot.

Indeed, but such good videos aren't always available for all events.

It's fairly clear how many we score in Wisconsin in the vids I posted elsewhere, even though the angle is blar. The webcast folks did say they hope to share an archive of the event footage at some point in the future. Can't wait for that!

I guess the solution is for everyone to share more elimination round videos so we can see exactly how good teams are during crunch time!

[Austin2046]

the spreadsheet says our overall OPRs are 17.67 and 17.04 but the week 3 thread says our OPR was 11.25 for portland, which is what OPRNet says... and the first post in this thread says our OPR was 23.37 for seattle, which is also what OPRNet says...

so why are the numbers all so different?

we performed much better in seattle since we were working fully for most of our qualification matches, whereas in portland we were fully working only for our last 3 matches... so i can understand the 11.25 to 23.37 jump, but i don't see the minimal difference from 17.04 to 17.67...

[Thomas DeSilva]

17.31 is the average of your OPRs from Seattle and Portland, and the overall OPRs it gives you are your average, roughly, plus or minus .315.

[Ed Law]

Originally Posted by Gaurav27

Please see the attached Excel spreadsheet for weeks 1 through 4 OPR. (Excluding Hawaii)
It has filters to rank by hybrid, bridge, teleop. and total. (Excluding Hawaii, even for the 'total')
It includes multiple entries for teams that have competed at more than one regional.
Feel free to change it up and add any functionalities. (i.e sort by regional, etc.)

Once again, thanks to Bongle for his work on OPR over the years to make it very easy to use and comprehend.

The data you posted on total OPR are wrong. For teams that have played more than one event, you sometimes add up hybrid OPR, tele OPR and modified bridge OPR from different events together. That is why team 2046 numbers are quite a bit off and other teams are off too.

There is quite a bit of data pubilshed on CD already and they match each other perfectly. Everybody have access and choose the one that is more convenient for them to use. I like how you format your data, but please verify it before you publish so people are not confused.

[Ether]

Originally Posted by Ed Law

Where did you get the data for Hawaii? The reason I have not posted this week is I was waiting for them to post the Hawaii regional standing. Without the total HP, BP, TP and CP, how did you calculate the OPR of those for Hawaii?

Does anyone know why the Hawaii "Team Standing" data has STILL not been posted?

Attached is the Twitter data for Hawaii, from which the HP, BP, TP, and CP could be extracted. The usual warnings about Twitter data apply.

[Joseph Bisch]

I have not seen the OPRs for Hawaii posted anywhere, so I am posting them here. This was calculated with just the qualification matches. I am working on an addition to The Blue Alliance that will calculate OPR automatically. The OPR data was generated using the code I wrote for TBA and match data from TBA. If someone can verify these OPRs, it would be much appreciated.

Team	OPR
1056	28.51
359	26.57
368	16.11
3008	13.67
2090	13.48
2443	12.79
2439	11.91
3880	9.59
2465	9.39
2438	8.49
4158	7.67
2348	6.88
2455	5.34
2460	5.23
1378	5.1
4374	5.01
2896	4.15
3878	4.12
4253	3.85
2437	3.76
2853	3.4
3306	2.97
2445	2.66
2459	1.43
4218	1.35
3685	1.32
1920	1.25
2441	0.82
2504	0.76
3881	0.75
3879	0.56
4270	-0.54
3800	-0.73
3837	-0.77
3721	-0.9
2444	-0.92
3882	-5.4
2477	-5.58

[Ether]

Originally Posted by Joseph Bisch

If someone can verify these OPRs, it would be much appreciated.

Looks correct. See PDF attached.

[Ether]

Originally Posted by Ether

Does anyone know why the Hawaii "Team Standing" data has STILL not been posted?

Attached is the Twitter data for Hawaii, from which the HP, BP, TP, and CP could be extracted. The usual warnings about Twitter data apply.

Since the Team Standing data for Hawaii has still not been posted, and apparently there is no effort underway to try to retrieve the data and post it, I thought I would try to reconstruct the Team Standing data from the Twitter data for the Hawaii Qual matches.

Toward that end, and given the unreliability of Twitter data, I invite everyone interested to review the Hawaii Qual match Twitter data and report errors here (based on videos or personal notes taken at the event).

For example, the Twitter data has two tweets each for Qual Matches 8 and 16, each with different data:

Event	Type	M#	red	blue	red1	red2	red3	blue1	blue2	blue3	red	blue	red	blue	red	blue	red	blue	CP
			score	score							bridge	bridge	foul	foul	hybrid	hybrid	Tele	Tele

HI	Q	8	14	0	3008	3837	2459	3879	3881	2445	0	0	0	0	11	0	3	0	0
HI	Q	8	8	11	3008	3837	2459	3879	3881	2445	0	0	3	3	0	6	5	2	0

HI	Q	16	16	26	3306	3878	368	2444	2443	3880	0	10	0	3	14	12	2	1	0
HI	Q	16	0	24	3306	3878	368	2444	2443	3880	0	10	0	3	0	5	0	6	0

Does anyone know which is correct?