According to our scouting data, here’s the most efficient strategies at Montreal as measured in points scored per second, I find these results fascinating and unexpected:
1.06 pts/s [3 Assis] L.G.
0.70 pts/s [2 Assis] L.G.
0.67 pts/s [3 Assis] L.G. + Truss
0.55 pts/s [2 Assis] L.G. + Truss + Catch
0.45 pts/s [3 Assis] H.G. + Truss
0.39 pts/s [2 Assis] H.G. + Truss
0.28 pts/s [2 Assis] H.G.
0.27 pts/s [2 Assis] L.G. + Truss
0.24 pts/s [1 Assis] H.G. + Truss
0.16 pts/s [1 Assis] H.G.
Pretty cool data. The low goal seems quite efficient.
It is unfortunate this data can’t really be applied to anything but the Montreal regional qualifiers. The alliances that produced this data were randomized, and their strategies for each match may not have been clear.
How was the data gathered?
Android app where you punch in events as they happen, “pickup”, “drop”, “truss”, etc and it timestaps and logs everything. It’s actually on the play store here
https://play.google.com/store/apps/details?id=ca.loyola.robotics.knightwatch
Super fascinating. Is there any brilliant OPR/Stats/FMS Twitter expert that can figure out a way to compile these kinds of statistics from the FMS twitter data?
I’m surprised 2 assist+truss+low goal didn’t come up higher.
To be useful for comparison, we need some more information on “population size” for each of those stats. That is, there may be only a handful of examples for 3A/LG and they are skewing the numbers. Does the app provide that data?
Neat idea…
When does the time start? When the pedestal is lit or when the ball first enters the field?
This is actually really interesting data that would be cool to catalogue and compare against other regionals.
There would be outliers of course - cycles where they change strategy midway (ie miss a few HGs and go LG) but with a large enough sample size and maybe exclude it to elimination matches, we might get fascinating insights like ‘the truss is overrated’ as suggested by your data.
To answer my own question, I gave it a try. Looks like you could export your data and we could review the text file, but it doesn’t save those stats. No way to get population size within the app…
Thanks for the feedback and ideas!
My thought-process in designing this was: “I play a lot of MMOs, in many MMOs DPS is the only stat that matters…why is FRC not scouted that way?” In order to measure PPS (pts/s) I need detailed timing information and that needs some kind of tech support. This was my first stab at the the whole package (learning Android programming along the way). I’ve convinced myself that the concept is good, but there are some things I want to iron out of the implementation. Next year’s scouting system will be better!
You can see population size if you look up an individual team (and that needs the match data copy&paste’d from the official regional site so it can tell which teams were in which match). I decided not to clutter up the pages I was gonna be looking at often.
@FRCFMS only gives us Match Final Score, Auton Points, Tele Points, Foul points. We get no sense of what those tele points consist of (truss/catch/assist).
Team standings give us Total Assist, Auto, Truss/Catch, and Goal/Foul points. But these are summed up across all matches.
To get CLOSE to compiling strategies, we would need to pull the Team Standings after each match to get the updated details that FMS doesn’t provide us to infer how much of the Tele Points FMS told us about were Truss/Catch points and how many were assists. Thus, we, unfortunately, cannot retroactively compute them to my understanding.
One thing to think about is that this game isn’t 2 alliances playing on the same field together like many games in previous years. There is enough active defense where this game is, very clearly, 2 alliances playing against each other. Each ones’ strategy directly effects the outcome of the other. Therefore, it’s hard to tell which strategies are executed by teams who can effectively evade defense, and those that cannot. That team who can weave through a field is naturally going to be able to perform any offensive strategy better than the team who gets stuck at every robot. Just something to consider when looking at this data.
