[DonRotolo]

I've been wanting to describe 1676's system for a while, here's a first poke at that since I remember the details from 2 days ago.

We have at least 6 students scouting per match, each assigned to a team. They use laptops running a custom app feeding a database to keep track of everything the team does. Data accuracy is emphasized. We scout Thursday practice matches as well. This is our quantitative data, which we always offer to every team at the regional with no strings attached.

We have at least one (often 2) super-scout per match as well, they scout all 6 teams for qualitative aspects - what did they try at but fail (which wouldn't be caught by the quantitative folks), and subjective impressions. Including their behavior.

Pit scouting factual data,
Video of every match from way up high (including Thurs),
And one ring to rule them all - a coordinator to keep things running smoothly.

On Friday night, we use various combinations of what we know to rank teams into Yes/Maybe/No categories for First Pick, then we review teams for potential Second Picks.

For example, some of the factors we looked at this year (in no particular order):
Team ranking from FRC
Raw score from hung tubes
Number of logos
Number of minibots successfully deployed
Ubertubes hung
Subjective rankings of super scouts
Basic capabilities - what you do best
How consistent are you - was the minibot a fluke or can you do it almost every match.

Lastly, we look at our strategies and describe what ideal alliance partners would look like for each strategy. Last year, one such strategy was to have a good reliable hanger, a robot that could score in the home zone (even a pusher was OK) and someone who could score in autonomous from the middle. Strategy counts for a lot.

All this put teams into categories. Teams in the top category were watched on video, and a preliminary decision is made. I can safely say that for NJ, we concluded that if we were #1 we would pick 2016 as our first pick (which we did). We had 9 other teams ranked below them for the case where we were not #1, but our hope was to then be picked by 2016, since they were the highest team on our list.

I emphasize our list is based on performance, not friendship or lobbying. For example, team 25 is an awesome team, and they took us to winning NJ last year. This year, their performance was lower than 2016's, and were thus not our first pick. But, as part of the review process, we did note that their averages were depressed by early matches, and by the end of Friday they were driving a winning robot again. So, late performance on Friday counts more than early performance.

Anyway, we then re-hash the data looking for "sleepers", teams which are performing well but didn't appear on our top 10 lists. We found more than one this year.

We watch more match videos, clarifying comments we saw, verifying data, and getting questions to ask teams on Saturday morning. We end up with a list of 25 teams - mostly ranked, some to be ranked after clarifications and further review - which becomes our list.

Some tweaking may be done on Saturday.

So, for relatively new teams, or those inexperienced at scouting for potential alliances, some lessons:
Your actual FRC ranking doesn't matter too much*.
Friday's performance counts the most.
Later matches hold more sway than your first matches.
Consistency is important.
Certain unique capabilities can get you on the list.
Teams played with or against don't matter much, only your team's performance is reviewed.
Team behavior is a factor, not just drive team but anywhere we see someone from your team.

Hope this helps someone out there.


* True story: 1676 was ranked higher than 15 in one game, but never got picked for elims. Some kind teams explained why: we were inconsistent on Friday (but were excellent on Saturday) and did not meet most teams strategies. Had nothing to do with "they don't like us" and all to do with performance.