How will the towers affect OPR, DPR, and CCWM?

In the past, teams have used statistics like OPR, DPR, and CCWM (Offensive Power Rating, Defensive Power Rating, and Calculated Contribution to Winning Margin) to quickly compare teams with reasonably high accuracy.

This has worked really well for some games (like In The Zone and Nothing But Net) and slightly less well for others (like Turning Point because descoring was allowed).

It seems like these formulas operate under the assumption that a higher match score is in all cases better than a lower match score. (Full disclaimer, I’m not super sure on how these formulas work so I might be wrong about this.)

This would mean that the formulas significantly underestimate the OPR of teams with a neutral tower strategy (i.e., focusing on an even distribution of colors instead of going after one specific color and using the towers). It would also mean that the OPR of teams that try and bring the score as high as possible (through the use of towers) would be overestimated. DPR and CCWM would be similarly skewed to favor certain strategies.

Am I correct in my assessment of this? While a holistic approach to picking an alliance partner and comparing teams is obviously ideal, the raw CCWM statistic helped teams quickly figure out which teams were better than which others. Is there another metric we can use similarly to CCWM, perhaps one that considers the ratio between scores instead of the raw number?

8 Likes
These are some links I found in the past to help explain the formulas. I couldn’t find one on DPR though.)

(Disclaimer: I did not make these credits to amazing people who did )
https://www.chiefdelphi.com/t/paper-new-scouting-database-from-team-2834/91560/48

https://www.chiefdelphi.com/t/incorporating-opposing-alliance-information-in-ccwm-calculations/144460/14

https://www.chiefdelphi.com/t/opr-formula/117688/18

https://www.chiefdelphi.com/uploads/default/original/3X/f/c/fc16640bd978382f349eadbcdd92c2eeffa69236.pdf

(I couldn’t find a formula for DPR but if you might be able to find it by using the proof CCWM=OPR-DPR)

This is correct. The formulas give higher scores for CCWM and OPR if your score is higher. Some teams however, purposely score lower for more SP points, which could affect the matches CCWM, OPR, and DPR because the robot is not scoring it’s full potential of points for its own alliance.

And in regards to the towers, yes, they will affect these statistics. This is because I believe your CCWM lowers when you score anything for the opposing alliance. Because towers affect everyone’s score, I believe if the cube you place in the tower doesn’t help you much more than your opponent your CCWM will be negatively affected or perhaps because Towers will help the opponents, maybe the CCWM will be on average lower this year. (Ps, just checked on vex via and there is only one with posted results and yes, there are a lot of low CCWM’s.
But, the CCWM’s of close ranked teams was drastically different. In one competition, the second place had a 16.6 ccwm but third place had a 3.0, both these teams however had a 7-3-0 score. (Third place did have less ap and sp points tho).
Perhaps based on the large effect on the statistics, teams with stats in a certain range will have a certain strategy such as more inclined to placing in towers or primarily stacking tall stacks or another range of stats would be a robot capable of making stacks and placing certain colors in towers to benefit primarily their own alliance? Hopefully after more competitions are played it will be easier to have an idea of what the stats for this season mean.

Edit: in regards to your final question, I looked and was unable to find anything like it. If I do find anything I’ll be sure to post it :+1: :slight_smile:

3 Likes

Didn’t the VEX via app break into more detail how points were scored? Perhaps it could be a metric to count the number of cubes scored and the number of cubes in towers that benefit one alliance more than another?

EDIT:
IMG_5795

Something like this, but keeps track of the orange/green/purple with each team, and then whatever color is more abundant is calculated to help one team and hurt another team in rankings? I agree that towers are going to sway OPR, DPR, and CCWM drastically, but I feel like there is some easy way to make a new counting system.

5 Likes

Thanks for pointing that out, I wasn’t aware of that. It looks like a new feature. It seems to work for Worlds, but when I tried it for my region’s state meet, I got an error stating “Score details are not available for this match at this time.” Hopefully this year it’ll work for smaller tournaments!

That’ll definitely help significantly. Hopefully the data from that makes its way onto vexdb; I could really see it being a useful reference for making statistics.

Some ideas I had in mind:

  • “Stacking Power Rating” - calculated the same way as OPR, but based on stack size instead of total score

  • “Average Score Ratio” - average ratio of the team’s score to the opponent’s score

  • “Average Tower Multiplier Ratio” - average of (the team’s total score / their number of stacked cubes) / (the opponent’s total score / their number of stacked cubes)

These should pretty reasonably estimate the strength of teams, regardless of the strategy they choose. The one issue I see is that the last two don’t account for the strength of the alliance partner in the same way OPR/DPR/etc. do.

I’ll probably mess around with this more once I get my hands on more data.

4 Likes

Of course, I forgot about this feature. Thanks!
Edit: unfortunately the event I found with updated results didn’t have any details about the matches.

1 Like

At least the last 2 years, Tournament Manager/VEX Via have gotten some big new worlds-exclusive features. This year, the scoring interface in TM mobile changed completely to allow individual tracking of each flag and cap, which is what enabled the fancier live-scoring on the worlds webcast. The Via screen pictured above was also exclusive to Worlds (though in theory that data should’ve been available in the pre-worlds build of TM), as was the “upcoming matches” display used in the queueing area of each division, and maybe some other features I’m forgetting right now.

I would love to see some of these features available at other events. That scoring breakdown in Via would be useful to lots of teams, the worlds scoring interface was MUCH easier to use for live scoring than the one it replaced, and I would totally use that queueing display at states if we had it available.

4 Likes

The intention was that this should have been available during the season, however what was ultimately decided to be put on screen for live scoring required different scorekeeping thus the changes to TM and TM Mobile. Also, the live scoring graphics were delayed, leading it to be used at Worlds only.

Things are a bit better this year - we did a thorough review with the GDC regarding what needs to be tracked to enable the Worlds-style live scoring and that was included in the initial release of TM for this year. The live score details are already on this year’s audience display. There may be some additional enhancements coming there, but for the most part what was at Worlds is already available in this year’s TM.

It was not developed as a “Worlds-exclusive” feature, rather it was developed as a new feature that debuted at Worlds that will now be part of TM going forward. The VEX via screen shown above will soon be available for this season for all events (it should become available this summer).

This has been available in TM for any event that wants to use it for some time now (2 years I think), so have at it for your state championship!

6 Likes

I should confess, I hadn’t actually tried the TM mobile interface for TT before writing the above post. I just played around with it a bit, and it looks great! The tower interface had some overlapping elements on my phone, but I imagine it would be fine on a tablet.

Overlapping Elements

IMG_1D87F8C54CD6-1

The new overlay audience display, including the live score details, looks great as well! Are there any plans to show those details on the non-overlay audience display? In my test, the “Full” audience display only showed the numerical score and the autonomous period status.

Awesome!

Also awesome, looking forward to using this!

3 Likes

@dhmmjoph I took a look at the overlapping elements that you showed and was able to correct the problem. This was an issue related to smaller screens as you guessed. We’ll be uploading new the fix after some additional testing. If you find any other issues please let us know.

5 Likes

Awesome, thanks for looking into it! Y’all do a great job with TM!

1 Like

Do the referees score using tablets, which directly upload data to the tournament manger? And there should be multiple referees, so what if their scores overlap? I tried downloading the tournament manger to determine how the system works. However it had to connect to an existing tournament.

Some events use paper scoresheets, the data from which is then entered into Tournament Manager on a computer. Other events use the TM mobile app running on one or more devices, which sends the match data to TM via WiFi.

The score recorded in TM mobile syncs in real-time between all the devices looking at that match. So if there are multiple refs scoring a particular match, they can see the changes each other are making as they are made. Then, after the match, all that’s necessary is for the refs to look over the score briefly to make sure it’s correct and for someone to hit the ‘submit’ button on their device.

If you want to fool around with Tournament Manager, the desktop version is a free download, and I put together a video series a while back covering the basic setup and use of TM, including scoring using TM Mobile.

1 Like