On Scouting: Picklisting and Strategy

One of the biggest misconceptions about scouting is that it’s just for the purpose of picklisting- so if your team isn’t going to be in a picking position, they don’t need to scout or picklist.
First of all- scouting has many purposes beyond picklisting- individual match strategy, eliminations strategy, team bonding, etc.
Second of all- every team at competition should have a picklist. Not just if you’re in the Top 8. Sometimes you might be thrust into the top 8 by in-picking and the effects of Saturday morning matches, sometimes you might be a first round selection (even if you can’t score points), because a team thought you could or some other weird reason, and you need to help select a third or fourth machine. Maybe you’re a third round robot, but you have to determine whether it makes sense to try to fix your broken partner or replace them with a backup robot.
Every team needs to scout, every team needs to picklist. In my last post on scouting, I discussed how to go about doing that, even if your team has low resources or is a rookie.

Now I’m going to discuss picklisting.
Picklisting is a tricky art, because it can vary wildly depending on your seed, the other alliances being formed, what your robot is and needs, etc.
Some teams have very formal picklisting procedures and they do the same process at every competition.
Others vary their methods based on the competition and other factors. Sometimes picklists look very formal, with a series of lists based on certain needs and factors. Other times they’re smaller and more focused (and maybe a little goofy), like this. All of these can be successful if you know what you’re doing.

To begin picklisting, you need to understand your goal. Most of the time the goal is “Win the event”, but sometimes it’s more complicated than that. Maybe you want to win the event, but getting a wildcard slot or just being more successful than the previous year is more important. Maybe you want to win your division, but you aren’t sure that the best alliance to win your division isn’t the same alliance that will win on Einstein. Identify your goal.

The second step is understanding the different possible approaches to attaining that goal. For the purposes of this post, I’m going to assume the goal is to win the event you’re at, and make a few remarks about subgoals.

The usual methods to winning an event, as I explain them during strategy and picklisting discussions are:

  • Captain a high seeded alliance- usually the #1 or #2 seed allows you to pick the best possible 1st robot and dictate for other teams how alliance selection is going to go
  • Be picked first by a high seeded captain- be the most desirable pick for the top seeds
  • Captain or be picked first on a lower seeded alliance- I group these two together because they have similar requirements
  • Be the third robot on an alliance- this is certainly a viable option, but it’s difficult to be in control of your own destiny with this approach. Often desirable 2nd round picks are picked too early in the draft to have a good shot of winning an event.

It’s possible to pursue multiple of these approaches, but which ones you actually might be able to do depends on your current seed, what your robot is and isn’t capable of doing, and what your goals are. If one of your subgoals is to grab a wild card spot if you don’t win, it might be more important for you to captain your own alliance than to try to win as a first round pick. Don’t be ashamed to pursue the fourth option (and just because you pursue the fourth option doesn’t mean you shouldn’t picklist). There are ways to pursue the fourth option actively instead of passively hoping to be picked by a good alliance.

My next post is going to be about aiming to be picked and how you can actively pursue that option, regardless of how good your robot actually is. The rest of this post is going to discuss the other three options.

Now is when the picklisting process can diverge wildly. A few rules of thumb across all possibilities to consider:

  • At an event with 24 spots for eliminations, have at least 24 robots on your list, preferably 3-4 more as well. If you’re the #1 captain, 22 other robots are going to be made unavailable before you get to pick your last pick. You need to have contingencies in case your dream pick gets scooped up early.
  • Put all of the top teams on your list. Even if they seem like an unattainable option to pick, you never know when you might end up seeded above them. It also can help you determine who at the event to reject, and what alliances you might be afraid of. Include your own team on that list (if you think your robot deserves to go on that list).
  • Change your list as new information becomes available. If your #1 pick for all of Friday breaks Saturday morning, they might not be the best pick anymore. Similarly, if there’s a very inconsistent team on Friday that has a spectacular match on Saturday, or demonstrates a skill that your alliance badly needs for the first time, take note, move them around on your picklist, etc. So many regional wins and World Championship wins have occurred because that captain was paying attention Saturday morning while other captains were not. The third robot can make or break some alliances.

With those rules in mind, I’m going to give a few radically different scenarios and how picklisting should have been or was affected by different factors.

I’ll begin with the 2014 Tech Valley Regional. Team 20 was one of the best robots at the event, despite some major inconsistencies Friday morning, and we were going into Saturday likely getting to be the second seeded team. 1126 was locked into the #1 seed, barring something crazy occurring. After reviewing our scouting data, we didn’t feel that 1126 was the best partner for us when there were so many good trussing robots at the event, and they had been mostly finishing, which we wanted to be doing. We came up with a list of 5 or 6 robots we would pick above them.
This list included 340, 379, 3044, 3990, as well as one or two others. We confidently agreed that we could reject 1126 knowing we’d be able to select one of these other machines.

Unfortunately, things did not go as planned. Our match with 340 did not go well, and we dropped into the 5th seed at the end of qualifications. Additionally, 3044 had begun missing a lot of autonomous shots, and 3990 was having catapult issues. Suddenly our list of 5-6 robots we could pick from the 2 seed was a list of 2-3 robots we might be able to pick, and we were the 5th seed. It was now far too risky to hope we could pick one of those few robots we had above 1126, or that one of the teams that had mechanical failures would be fixed.

We had to re-adjust our picklist to make sense of the new developments. We accepted 1126’s invitation, knowing we’d be playing a different style of play than previously discussed. With one of our other picks (perhaps 379 or 340), we would have played a more aggressive, quick style of play, but with 1126, we knew we’d be aggressive on defense and slow our opponents down just as much as we’d play offense.

Alliance selections were an interesting affair, and the last selection came around and we were ready to select Team 4203, Robokronos, an awesome, very professional team with a consistent autonomous shot, a catching mechanism to make inbounding very easy, and a tremendous desire to improve. One of the teams that was previously on our list Friday night was Team 229, Division by Zero. I had removed them from the picklist just minutes earlier when their robot had a mechanical failure on the field, and their field representative told me that their machine was broken and couldn’t collect a ball.

One of our members was discussing the mechanical failure with one of their members during alliance selections, and was told that the broken piece of their robot was a blown Versaplanetary gearbox, and they had no spares. We had a few spare Versaplanetary gearboxes, and right before our turn to select, I was signaled that we should pick 229 instead of 4203. I was puzzled, and discussed the situation with the 1126 field representative. 4203 was the first backup robot available, and if the 229 pick went horribly wrong and they broke again, we could have had 4203 as our backup. We selected 229, and they played amazing on the field, being one of the most important members of the 2014 TVR Championship alliance on the field.

By constantly changing our picklists based on information available to us, we were able to come up with a regional win at a hotly contended event.

My second example is from the 2013 River Rage Off-Season Competition.

The River Rage off-season competition incorporates a randomized first pick system, which changes up strategy and scouting quite a bit, with picking proceeding 1-8 in the second round. 20 was the first seed at this event, and received Team 151 as our randomized first pick. After first round random selections, there were two teams that were viable picks for our second selection. Team 190 at this event was scoring 50 points each match, with a consistent 30 point climb and 20 point dump in the pyramid goal. Team 2648 was the other pick, who during the season was scoring significantly more, but had been having mechanical issues throughout the event, causing them to score less on average than 190 by quite a bit during qualifications.

The obvious pick at this point is Team 190, however there is more to consider. The second seed was Team 3467. They had a phenomenal machine, scoring over 50 points on average shooting frisbees, with an additional 30 points from a climb inside of the pyramid.

If 3467 ended up paired with Team 190, they couldn’t both climb on the inside of the pyramid, and 3467 would have had to pick another team who scored significantly less than 190 did on average.

With this in context, the best possible pick would have been 2648, as it would have prevented the 3467-2648  alliance from forming. In the end, we picked Team 190 and lost in the finals to the alliance of 3467-1517-2648.

The final case study is picking from a lower seed. Team 20 was the #8 seed at the 2015 Tech Valley Regional, and moved up to the position of 5th Captain during alliance selections.

The first and second seeded alliances both had the ability to produce 4 capped stacks from the human player station and receive cans from the step.

At the time, 20’s machine was capable of 1.5 capped stacks in our best match, and none of the robots available to pick were capable of producing more than 1.5 stacks consistently. There were also no teams capable of producing any capped stacks from the landfill, so only two robots could produce capped stacks at a time.

With this knowledge, we needed some kind of risky play in order to have a shot of winning the event. Going into Saturday morning, we were looking for a few different things that could possibly boost the output of an alliance we made.

  • Landfill Stackers
  • Robots whose output we could boost with the addition of a ramp
  • Capping robots
  • Robots that could get cans off of the step

Throughout Friday morning, we found a few teams that showed some potential in just a few matches. Team 3624 in their last match made a single stack from the human player station, then capped it. Additionally, Team 5254 had demonstrated the ability to make one capped stack from the feeder station or from the landfill, but we felt they could improve their performance with a ramp. Additionally, they had the ability to grab cans from the step during autonomous.

We managed to pick both 5254 and 3624, and developed a strategy that attached 5254’s new ramp to 20’s robot, with 5254 making as many stacks as possible and 3624 capping uncapped stacks produced by 20.

Despite a poor performance in our first quarterfinal, our alliance clicked in our second quarterfinal match, with our picks doing incredibly well. 5254 made two capped stacks on their own, and 3624 capped two of 20’s stacks, making 4 capped stacks. Our alliance ended up losing in finals due to our inability to get cans from the middle of the field with 340 beating 5254 to the cans every time, but making it to the finals from the 8th seed was a victory in and of itself.

This was only possible with extremely vigilant scouting on Saturday morning and the willingness to take risks from a low seed.

Too often, low seeded alliances try to pick the most consistent robots available. The problem with this is that, from a low seed, you need to take risks if you want a shot at beating the stronger, higher seeds.
From a lower seed, robots with higher standard deviations become more valuable, and risks need to be taken .

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s