Best Ways to Scout

[BenjiSG]

Originally Posted by Citrus Dad

I'll make 2 points. The first is more basic--you can use Bluetooth and cellphone data to transmit which avoids WiFi. See our scouting whitepaper. And we use tablets which can get through an entire day of scouting-you only need to charge at night. Plus the system is fairly reliable--at least as much as paper when those can get scattered. The final one is that it is much easier to communicate with the drive team in real time with an electronic system. (Scouting is about much more than draft lists.)

My second point is about philosophy. The point of FRC is not competition and winning; it's about education and training (which is fun in a sports setting). Having your students create an electronic scouting system is another educational opportunity. We have many students programming apps, servers and other devices. They learn about math and statistics. This has been one of the best educational opportunities for our team.

At most events I've been to, cell reception has often been spotty at best. We also investigated Bluetooth, but were put off by price and hassle.

Our core scouting team is relatively small- generally, there are three people who actually dedicated and the rest are simply members we get to scout matches while in the stands at the competition. There's definitely value towards devising your own system and learning all of the necessary skills involved, but since our dedicated scouting leads all have other demanding jobs which are also requiring of their time, paper scouting seems to give us the best efficiency with the least effort required. It may sound lazy, but it's really just a solution to make sure that we can show up with effective scouting without tearing people from other projects.

A larger, even more dedicated scouting team could likely complete such a project and get through all of the workarounds, but it honestly seems to be for a different goal from simply scouting in the "best" way possible- that is, the way which can accurately inform selections while being relatively low maintenance. Scouting for the learning experience is excellent for a team with members dedicated to putting vast amounts of time into the intricacies of processing data and utilizing statistics, but paper scouting gets the job done and gets it done well.

[ngreen]

I have two answers with the first reflecting the goals of FIRST, that is to really test your abilities and to learn new things. I encourage those building apps, networking, automating processes, doing statistical calculations, etc.

And second, ideally for my team, I feel like three things needed to happen within scouting which is to utilize the FIRST FMS stats fully (develop automated processes and visualizations to pick out not only the best teams, but overlooked teams for potential 2nd picks), develop qualitative intelligence of the teams (through pit scouting, watching matches, drive team interactions, etc), and form bonds with the other teams.

You can do some pretty amazing things with the FIRST ranking data, a copy of Tableau, and some Excel magic (and also R). Our quantitative scouting ended up being scraping ranking data (which includes auto, defenses, goals, and scale points), calculating an OPR-like value for each of those, and visualizing in Tableau. It is very simple to filter and sort in Tableau if you like a more exploratory way to identify teams, or you can pre-build your analysis.

We looked at the data Friday night (after 8 matches) and Saturday (after 11). In retrospect, I would have liked to share information earlier between the quantitative and qualitative scouts. We did this at the end of the day but could have done it another time and had an extra iteration before the scouting meeting. By sharing information, I mean that from the data you can start to see which teams would be best to watch more closely (particularly for things that aren't captured well in the data -- like breakdowns, driver skill, etc), and that from observations you may find some teams that don't pass the eye test that isn't captured in the data. For example, I watched a robot that rated 14th by OPR-like stats in one of their final matches seem slow and indecisive.

Every team is capable of scouting in this way. Certainly, teams with more data have an advantage, although I think slight as sometime it doesn't make the picture clearer than aggregate data and the eye test. Next step to use the free data fully is to pull all the match data and analyze it. A team's ability to capture the tower was an additional thing we looked for in the best teams (with low goals being important, but not counting as much in goal score).

I regret this year not having as many students working with the data for scouting, given the educational opportunity. We did switch systems following our first regional where we were paper-based. I'd like all the scouting team students next year to know how to extract the data from the website, be able to do basic tasks in Excel, be able to build their own dashboards/plots in Tableau, and more but it is a start.