Rookie scouting coordinator requesting input

[gblake]

Wildcat,

There is a way to test-drive both your match scouting process, and your analysis of the data you collect during the matches.

Use the 5th Gear simulator with live humans players and scouts to run a set of qual matches; and use your scouting methods as those matches unfold. Then use your analysis methods and see if you are happy with the result.

If the results look "off" you can review (using the replay feature) the pertinent matches to see if you can figure out why the scouts came up with inaccurate assessments, or to identify how to modify your metrics.

At the end you can decide if the scouting results identify the "best" potential allies for your team?

See this post (5th Gear Scouting-Aid Suggestion).

Also, by searching for the key words "5th Gear Scouting" you will find a couple of recent threads on the subject of scouting. At least one of them has formats and metrics you can compare to the ones you are planning to use.

Blake

[Wildcat]

thanks for the info, it looks a lot simpler than what i originally had planned for a simulation... definitely gonna have to try this out

[gblake]

Originally Posted by Wildcat

thanks for the info, it looks a lot simpler than what i originally had planned for a simulation... definitely gonna have to try this out

Glad to hear it - Let us know what happens - by PM or in the forum

[Ben Martin]

In my 5 years of doing scouting, the two biggest items I have found useful are 1) how many points/game pieces each robot is worth and 2) pictures. An objective measure of driver ability is also useful (last year, how easily you got scored on was a limited measure of this). Last year, summing the average number of game pieces each team could score accurately predicted about 80% of the matches at the regionals we attended. This year, points may be used similarly (we'll see after week 1).

[Ian Curtis]

Originally Posted by BMartin 234

In my 5 years of doing scouting, the two biggest items I have found useful are 1) how many points/game pieces each robot is worth and 2) pictures. An objective measure of driver ability is also useful (last year, how easily you got scored on was a limited measure of this). Last year, summing the average number of game pieces each team could score accurately predicted about 80% of the matches at the regionals we attended. This year, points may be used similarly (we'll see after week 1).

I have to agree with this. I ran a scouting crew for 5 years (, that scares even me!) and every year we got better.

As currently written, there's no way you'll be able to concisely summarize the information you have available. IMHO, what you need is numbers, and you don't need very many of them.

In my experience, the more information you ask for from scouts, the less you can trust it. In 2007 we asked for probably a dozen numbers per robot per match... and we ended up using only 2. The scouts didn't even report most of the other numbers, there was too much going on in the match. So in 2008 and 2009 we asked for way fewer numbers. In 2008 we asked for: # lines in auton, # of lines crossed in teleop, balls hurdled, balls placed. In 2009 we asked for: balls scored (approx), balls scored on robot (approx), and Empty Cells delivered. Ask for any more and the scouts won't get good numbers, and they'll get really mad at you really fast.

That said, we also had a pair of experienced kids who kept subjective notes. It's important that you trust the judgement of the note keepers though, a parent you've just pulled in off the street probably won't be able to give you opinions you trust.

In pit scouting you've got a lot more room to play, but whatever you do, it needs to end up in an easy to digest format. And as others have said, pictures are really important. You've probably got that one kid who can keep all the robots at your event straight, but not everyone can do that.

PM me if you're interested, and I can see if I can hook you up with copies of our old stuff to look over.

[Wildcat]

well i gave the presentation, got plenty of positive feedback on the powerpoint, 5th gear refused to work, and final update of match scouting sheet

[Fennig247]

Have you maybe thought about going electronically?
I know I converted my team and now I make our teams scouting program. You know, save the trees!! Go Green!!

[Jon Stratis]

Over the past few years, we've realized that the best information you can get from scouting is straight, qualitative numbers. Things like:

- How many balls scored
- How many balls passed into another zone
- How many shots/passes blocked

Gather those numbers for each zone (defense, midfield and offense). Then if you want to know the best defensive player out there, you look at the defense column - blocks + passes. For the midfield, look at the midfield column - blocks + scores + passes (or drop the passes if you want to pull back the offensive person to be a second midfielder). For offensive, look in the offensive column - scores.

Have good, objective numbers makes alliance picking extremely easy - and it makes selling yourself extremely easy too. It's hard to justify "well, our shots are more accurate than anyone else's" without numbers - being able to say "during our matches on Friday, we made 34/37 attempted shots on goal. We had the most attempted shots, and the highest success rate of anyone at the competition" will make your team look a lot better.

[tim-tim]

I like the general path that you are going with.

I would have to reiterate the ideas of KISS, and simply asking them to right down the cold hard facts of what they see, and not try to interpret it. Lets face it, numbers do not lie, unless you twist them or put an interpretation behind them.

This is a link http://www.chiefdelphi.com/media/papers/2334? to our team's database. Being a former driver and coach, I can tell you that with information very similar to this, and knowing what each team is really capable of, you will be fine. We had a timeline of what we were to do and check every 15-20 sec during the match and change accordingly. Most of the time, we were right on from scouting (Thanks Steve - author of the database).

I was a scout my first 2 of 4 years on the team, and saw the reports go from labor intensive and lots of writing to something like this. The problem is, if the kids are writing the whole time, they are not watching the match and missing key events. I would recommend keeping it simple and looking at the numbers. The only thing left for interpretation usually is: Would we want them on our alliance?

[gblake]

Originally Posted by Wildcat

... 5th gear refused to work ...

Can you give us more information about the problem? Our freeware installer has been giving some people a hard time, and another pernicious bug got tracked down this weekend; but many people have used it successfully. I'm disappointed that you had a problem you couldn't get past.

Blake

[Wildcat]

Originally Posted by gblake

Can you give us more information about the problem? Our freeware installer has been giving some people a hard time, and another pernicious bug got tracked down this weekend; but many people have used it successfully. I'm disappointed that you had a problem you couldn't get past.

Blake

i couldnt really tell much about why it wouldnt work, i since i found out about it a few days before presenting to my team, i wasn't really that concerned if it didnt work, just thought it wouldhave been a nice addition.

And to all who offered their input, thank you. after my first offical competition i think ive got the hang of it, aside from the near physical fights between my scouting team and the rest of the team everything went pretty good and has a positive outlook ahead

[kirtar]

Originally Posted by eagle33199

Over the past few years, we've realized that the best information you can get from scouting is straight, qualitative numbers. Things like:

I hope you meant quantitative. In any case, we did numbers last year, and it was a nightmare to collect it accurately. At Boilermaker, there was a 6 team scouting alliance in which each team contributed 3 people to scout a single team on the field. Not very efficient (I could have done a single team by myself).

This year because there is such a small spread, I'm only looking at a few basic numbers such as:
Success rate at elevation/suspension
Shooting % in autonomous
Time it mechanically takes to hang (once in place, mechanism deployment only)