Chief Delphi - paper: New Scouting Database from Team 2834

Chief Delphi (http://www.chiefdelphi.com/forums/index.php)

- Extra Discussion (http://www.chiefdelphi.com/forums/forumdisplay.php?f=68)

- - paper: New Scouting Database from Team 2834 (http://www.chiefdelphi.com/forums/showthread.php?t=70111)

Team 2834 Scouting Database (Week 3 Results and Week 4 Regionals Ready)

I have uploaded the Week 3 results. Please note that Peachtree Regional did not post the Standings. I calculated the standings based on match results. (You do that by using macro shortcut SHIFT-CRTL-R after you manually refresh the qualifying round data.) However if there are extra matches, it will still go into the standings because I would not know which match does not count. However it will not affect the OPR and CCWM data since all match data are used to calculate them.

I have also put in the links for Week 4 regionals and unless I made a mistake, they should all work. I need to write macros to create these links. It is very tedious to add the links manually and it is possible to make mistakes. Team 2834 will be going to Detroit District Competition. I will be busy this week since we have to work on the robot for 8 hours. I will not have time to change the database. However if someone finds an error in the spreadsheet, I will fix it as soon as I can.

I was looking at the results of the Cass Tech and I was amazed by the selection. It seems that the CCWM correlates extremely well with whiich teams were picked. Many teams that were ranked high were skipped. Except for one team, the teams that were skipped all had negative CCWM. It means either all Alliance Captains have access to this data (which I doubt that many teams actually use these data) or they can tell those teams were not that helpful to their partners in the qualifying round. Here are the numbers.

PHP Code:


		
			
Avg Score    OPR    OPR Rank    Avg WM    CCWM    CCWM Rank    Record    Regional Rank    Alliance    Finish

28.6    50.9    1    16.1    50.3    1    (11-0-1)    1 of 40    Alliance Captain  1    Regional Winner

24.6    38.9    3    8.6    27.2    3    (10-2-0)    2 of 40    # 1 pick    Regional Winner

23.4    33.0    4    7.3    17.0    4    (9-2-1)    3 of 40    Alliance Captain  2    Finalist

25.5    42.7    2    12.6    44.3    2    (8-3-1)    4 of 40    Alliance Captain  3    Semi Finalist

16.8    16.6    19    1.1    8.5    8    (7-3-2)    5 of 40    Alliance Captain  4    Semi Finalist

16.9    17.5    15    1.4    11.1    7    (8-4-0)    6 of 40    # 4 pick    Semi Finalist

18.8    17.0    17    2.0    2.7    18    (7-4-1)    7 of 40    Alliance Captain  5    Quarter Finalist

18.4    18.9    13    2.1    6.2    12    (7-4-1)    8 of 40    Alliance Captain  6    Quarter Finalist

16.4    14.2    24    0.3    0.1    21    (7-4-1)    9 of 40    Alliance Captain  7    Quarter Finalist

21.3    29.1    6    2.2    4.7    15    (7-4-1)    10 of 40    # 6 pick    Quarter Finalist

18.2    22.4    11    0.2    4.6    16    (7-4-1)    11 of 40    Alliance Captain  8    Quarter Finalist

21.2    25.9    8    4.1    8.2    9    (7-5-0)    12 of 40    # 7 or 10 pick    Quarter Finalist

17.3    14.7    22    -1.4    -7.8    29    (7-5-0)    13 of 40        

23.0    33.0    5    6.3    13.4    5    (7-5-0)    14 of 40    # 5 or 12 pick    Quarter Finalist

16.6    15.0    21    -0.3    -2.1    23    (7-5-0)    15 of 40    # 13 pick    Semi Finalist

17.3    11.0    31    -0.5    -11.5    30    (7-5-0)    16 of 40        

15.6    14.4    23    0.4    5.5    13    (7-5-0)    17 of 40    # 8 or 9 pick    Quarter Finalist

17.3    13.9    25    -0.5    -7.0    28    (6-5-1)    18 of 40        

17.3    16.3    20    0.0    0.5    20    (5-5-2)    19 of 40    # 12 or 5 pick    Quarter Finalist

15.9    6.4    35    -3.8    -19.2    37    (6-6-0)    20 of 40        

17.4    18.2    14    0.1    4.4    17    (6-6-0)    21 of 40    # 10 or 7 pick    Quarter Finalist

19.8    26.1    7    2.7    12.2    6    (6-6-0)    22 of 40    # 3 or 14 pick    Semi Finalist

16.8    11.6    30    0.8    -4.3    25    (5-5-2)    23 of 40    # 9 or 8 pick    Quarter Finalist

17.8    24.2    10    -0.1    8.1    10    (6-6-0)    24 of 40    # 11 pick    Quarter Finalist

16.6    13.2    27    -1.8    -6.5    27    (6-6-0)    25 of 40        

19.7    22.0    12    2.3    5.1    14    (5-7-0)    26 of 40    # 14 or 3 pick    Semi Finalist

18.4    24.6    9    0.6    6.6    11    (5-7-0)    27 of 40    # 2 or 15 pick    Finalist

16.3    17.1    16    -1.8    0.1    22    (5-7-0)    28 of 40    # 16 pick    Regional Winner

13.9    2.4    39    -2.4    -11.6    31    (5-7-0)    29 of 40        

16.4    13.4    26    -1.1    -3.0    24    (4-7-1)    30 of 40

Ed Law

Team 2834 Scouting Database (Week 4 results)

The database has been updated with Week 4 results. There are now results of 1301 teams who have played.

Also the missing data from previous weeks have all been posted and is reflected in the database.

I will add the links for Week 5 regionals next week when I have time. To add links before the events start is very time consuming unless I have a macro to do it in the future. Since the number of downloads is very low each week, I will only do the regionals on request only. PM me if you need it for a particular regional and I will email it to you. Otherwise I will post them after the regional starts on Friday. Thanks.

Ed Law

Re: paper: New Scouting Database from Team 2834

I was using this at our event this past weekend and had a couple people as about it, thanks again for writing it.

Re: paper: New Scouting Database from Team 2834

Quote:

Originally Posted by Phyrxes (Post 838927)

I was using this at our event this past weekend and had a couple people as about it, thanks again for writing it.

I am glad it was useful for you. The latest one is version 4c. All previous versions can be deleted. This one contains all of the latest data and the small bugs that I can find. I will be posting version 5 some time on Friday for Week 5 regionals.

Ed

Team 2834 Scouting Database (Week 5)

Hi,

Week 5 events have started. A new file is posted with all the links. Some regionals are not posting properly. That is why the spreadsheet may not be working. Give it some time until every team has played once and hopefully they would have corrected it by then.

If you have any problems with any of the links, please let me know.

Ed

Team 2834 Scouting Database (Week 5 Results)

Week 5 results are out. Rankings for Troy District Event are missing. I did include the Troy event based on rankings my program calculates. However the ranking may be slightly off because my RS numbers do not match the actual RS numbers which takes into account penalties that I do not have info of. Rankings for Connecticut Regional are also missing from the FIRST website but fortunately I have them stored on my spreadsheet as I update the rankings throughout the day.

Ed

Re: paper: New Scouting Database from Team 2834

(should I mention this here?)
Hi, first: I love the database! This isn't just an amazing tool for FRC info, it's a great example of how databases in general should look and feel.

Bug report: I can see the picture when the document loads, but it then disappears when the "enable automatic refresh?" pop-up appears, and is replaced with FALSE. The cell stays like that regardless of if I click Enable or Disable.

Re: paper: New Scouting Database from Team 2834

Quote:

Originally Posted by MrWibbles (Post 844631)

Yes, this is the place to report bugs. However that is not a bug. Before I get to that, I would recommend that you select disable automatic refresh.

The picture you see is the picture of our robot that was there when the Excel file was last saved. However when you open it up on your computer, it disappears because you do not have that picture file on your computer. Please read the instruction tab for details.

Basically you have to provide the picture files, name them as xxxx.jpg where xxxx is the team number. For example team 2834 would be 2834.jpg and team 33 would be 33.jpg etc.

Ed

Team 2834 Scouting Database (week 6)

Hi,

Week 6 is ready and available to use. We will be at the Michigan State Championship. If you use this spreadsheet and have any questions on how to use it etc., please stop by our pit. Unfortunately EMU will not be providing internet access so we have to manually update the scores.

Ed

Re: paper: New Scouting Database from Team 2834

Thanks Ed! This is the first week I have a real use for this beyond curiosity/novelty!!

Re: paper: New Scouting Database from Team 2834

so does anyone know of good scouting software? in need for state tourny this weekend

Re: paper: New Scouting Database from Team 2834

I have read a bit about this method of generating OPR and I was curious about your input for the Seattle Regional. My team, 488 ranks quite low on OPR and average score, but multiple teams of scouts confirmed we were in the top few for average scoring per match including human and robot scoring. I would say 3 of the top 5 scorers are ranked below 10 for opr. There are teams ranked in the top 10 that we did not record scoring a ball during the entire event. Is this biased based on strength of schedule or something like that? I would love to use this tool in ATL, but having my own experience at a regional, I need to know how far to trust this data.

Thanks!

Team 2834 Scouting Database (Week 6 Results)

Week 6 results have been updated. Troy standings is still not posted. If anyone has the ranking, please let me know.

I will update the spreadsheet again once the Championship divisions have been decided.

Team 2834 will be going to Atlanta. If anyone has any questions on how to use the spreadsheet or just want to talk about scouting, please stop by our pit.

Ed

Re: paper: New Scouting Database from Team 2834

Quote:

Originally Posted by Fred Sayre (Post 845379)

Sorry it took so long to get back to you. Our team was competing last weekend at the Michigan State Championship. We qualified for Atlanta and I was busy yesterday and today securing money from various sources and arranging bus transportation and hotel at the last minute.

I looked at the Seattle Regional. Your team, 488, is ranked 22 out of 64 teams. Your OPR of 19.7 is ranked 23rd which is quite respectable. Another thing I look at is comparing the average score with OPR. The green color means the OPR is higher than the average score. It means your team is contributing more points to the matches than your alliance partners on average. I don't know if there were any penalties against your alliance for your matches. If there is, it will lower your OPR score.

Team 2660 is ranked 4th with an impressive record of 6 wins and 1 loss. However its OPR is only 12.2 which ranked 40th and they have a negative CCWM. You can PM me and tell me how they actually perform. They picked 2 very good partners though.

Team 1983's record of 3 wins and 4 losses ranked as 33rd. However their OPR is 39.4 and ranked number 1. Their CCWM is 9.4 which is quite high also. With the average score of 19.5 and average CCWM of -0.4, I can tell that this team had a tough schedule or some very weak partners because they scored most of the points and their partners dragged them down. The data seemed to show that they were picked number 1 which means other teams recognize them as a good team with some bad luck. I think the numbers are quite accurate in telling the story.

I would be careful with the data when there are 64 teams and each team only plays 7 matches. There is not enough of interaction between all the teams. I would put a +/- 5 on the OPR data.

That is the problem when there are many teams and some of them are not competitive. At the Michigan State Championship, most of the 64 teams are highly competitive so the data is a lot more accurate.

I will be at Atlanta. If you want to talk a little more about OPR and CCWM, please stop by our pit and ask for me.

Ed

Re: paper: New Scouting Database from Team 2834

I received some questions through PM and I decided to answer them here so others can benefit from it also.

First of all, these OPR and CCWM should not replace watching matches and gathering match statistics if your team has the human resources to do it reliably.

OPR does not predict what a team (robot and human player) can score. It is the calculated contribution by that team on average to all the matches they were involved in to their alliance partners. A team that has high OPR score means that every time they are on the field, good things happen to that alliance meaning high score. Some of the possibilities are:

1) their robot score a lot of points
2) their human player score a lot of points
3) their presence allow their alliance partners to score a lot of points which they don't normally do as well.
4) they have on average stronger partners and weaker opponents by the luck of the draw than other teams.

A low OPR is just the opposite.

CCWM is the calculated contribution to the winning margins of the matches the team was involved in. About half the teams will have negative CCWMs and if you add up all the CCWMs of all the teams in the event, it will add up to zero. A negative CCWM means the team is a liability to their partners. They may be able to score points, but they get scored on more than they score on others and it may be because of penalties also. A team with negative CCWM should not be picked as alliance partners.

A team that has high CCWM means that every time they are on the field, good things happen to that alliance and in this case it means winning by a big margin. Some of the possibilities are:

1) they score more points on others than others score on them
2) their presence allow their alliance to score more points on others than others score on them. This could be from playing defense or help pin an opposing robot so their alliance partner can score more.
3) they do not incur much penalties.
4) they have on average stronger partners and weaker opponents by the luck of the draw than other teams.

In another scenario, a team with high OPR and negative CCWM could mean that they are a liability to their partners. They probably get scored on a lot more than they score on others but their presence helped their partners to score more than they normally do. To me, it raise a red flag about their game strategy. I hope this helps.

Ed Law