VexScout Website

Berthoud Robotics has been working tirelessly this summer to benefit the entire Vex community by creating an easy to use and comprehensive scouting website (www.vexscout.com). We have collected information from every tournament this past year and tournaments from TossUp and stored it in our database, which is available to the public. We will continue to add useful features to this helpful resource.
Email me at 1069B@BerthoudRobotics.org if you have questions or concerns. Feedback is always greatly appreciated.

www.VexScout.com

I really like the site, could be a useful tool for high school teams.

I’ll say it right now that your “Power Ranking” is going to cause controversy, considering the team that most consider to have had the best Sack Attack robot of the season is ranked 61st. I don’t think it’s possible to rank teams directly against each other like that, but good luck if you want to keep trying with it :p.

I noticed when I search my team last season (2941A) it’s listed as having 14 ties? This definitely isn’t correct. Maybe a tournament that didn’t receive robotevents data or something? Perhaps you could filter our tournaments where no data is uploaded (assuming this is the issue, no idea if it is or not)? Not too sure why that’s happened.

I like what data you have made available, and the layout of the site is really easy to use. Keep at it, I can easily see this helping a lot of teams with scouting, especially at worlds.

Well maybe this is only you? I quite like this scouting system! (In no way connected that it ranked my team 2nd :smiley: )

The Power Ranking is going to be controversy, one formula can’t please everybody. I probably should elaborate on how it works…
I have the average win percent (W%) of every team in the database.
The winning team’s additional score can only be positive and likewise the loosing team can only loose points. The formula for calculating the gain/loss for every match is: (sum(opponentsW%) - allianceW%) * sqrt(tournament size) for the winners and the negative for the loosing teams.
Unfortunately this has a heavy weight of the number of matches a team plays.

Keep in mind the Power Ranking system isn’t the only system out there and is **intended **to be used to filter the good teams from the bad, not used to determine the best robot. We are currently working on an ELO ranking system which might have better results.

Thanks for the feedback!
Trevor

Who?

edit
Your Trevor. I was so confused.

It seems that counting more competitions as a better is flawed.

Telemascope, unfortunately a tournament you (2941A) attended did not report the scores for each match. So the all match you participated in along with all the other teams from that tournament had a score of 0-0.

Here is the RobotEvents url for that tournament:
http://www.robotevents.com/robot-competitions/vex-robotics-competition/auckland-scrimmage-september-2.html

I’ll start creating a blacklist of tournaments for the second revision so this is no longer an issue.

Keep in mind you can sort teams by their win percentage if you prefer.

Why not just divide power ranking by the total number of matches played? Would remove that weighting and shouldn’t cause any other issues?

Yeah, I figured. Sounds like a good solution, pretty easy to monitor as well because you can just sort by the teams who have the most number of ties and check if they are legitimate or not. :slight_smile:

Amazing work ! Thanks for sharing

I think this is a good start.

You can also add the top robot & prog. skills scores weighted ,
maybe this weighting %60 Match Results + %20 Robot Skills + %20 Prog Skills

I tried to search some teams from our region, like 9581M and 8581R, and no results came up. I know these teams competed this past season with those numbers, at tournaments that are on Robotevents, but they don’t seem to exist in your database.

To help me find the update the database, do you know what tournament(s) specifically these teams have competed in?

). We have collected information from every tournament this past year and tournaments from TossUp and stored it in our database, which is available to the public. We will continue to add useful features to this helpful resource.
Email me at 1069B@BerthoudRobotics.org if you have questions or concerns. Feedback is always greatly appreciated.

www.VexScout.com

I’ve noticed several posts about missing competitions from your data and see that as well for my teams (323A/B/Z). One area that I ran into problems with last year when compiling my power rankings were competitions that had multiple divisions. Before an Excel download was available I was using screen scraping to pull the results from the Robot Events web site and it was only picking up the first division. Depending upon how you are getting the data you may be experiencing something similar.

Jay

I know that those two teams were both at the Wayne State competition that was in early January (not certain on the exact date, I think it was 1/12/13).

I re-generated the database this weekend, any previously missing teams should exist now. :slight_smile:

The numerical order is very strange

639
64
6401

It is probably just the order in which the spider received the information, I might make some of the grids random order by default (not power ranking)

New Power Ranking algorithm? Or just differences in what tournaments are considered? Of course, “power rankings” are always going to be influenced by the strength of a team’s region etc. but what you have at the moment is looking pretty good.

I noticed some of the win percentages don’t appear to be quite accurate?
For example 12A has 105 wins out of 124 matches played. This winrate should be 84.7% but is being displayed as 80.58%? Not too sure what’s going on there, seems to be happening for almost all of the win percentages I checked though.

Still really liking the user interface, I can see myself using this a lot (even though I’m not competing in High School any more).

Keep it up! :slight_smile:

More than likely this is because the numbers are actually strings when sorted.

Jay

The formula is the same but I totally restructured the calculations and data flow. The reason behind this is so that the power ranking could be used for a specific tournament, separate from the entire population.

I don’t know what was going wrong with the percentages, I think it has to do with null values in the calculations; I will investigate this further.

Thanks for the feedback, its always appreciated!

Trevor

Very classic problem :slight_smile:

The solution to this would be to use the Alphanum Sorting Algorithm. I’ve used it with success in Java in the past but there are PHP and JavaScript versions as well (I assume you’re using one of those 2 - I didn’t actually look at the site.)

Looks pretty easy to port even if you’re using some wacko language.