Statistically Winning Strategy

[BrianR]

I spent a bit of time this afternoon looking at match results, and decided to have a little fun with Excel to see if I could tease anything out of the data. What I found surprised me, and I thought it would be good to share with everyone else.

What I found was that the number of teams that scored over 30 points and lost was very low. In fact, 98.515% of all losing teams had fewer than 30 points. This would seem to indicate that anyone who can score over 30 points (a 12" ramp) every match, without penalties, should be well on their way to finals, if not one of the top seeds, WITHOUT EVER SCORING A SINGLE TUBE.

The different regionals did produce some varied data, notably VCU where 99.898% of losing teams had less than 30 points, and St Louis where 99.92% of losing teams had less than 30 points, and on the opposite end, Pacific Northwest, where 92.53% of losing teams had less than 30 points. Regardless, out of ten or eleven matches, this means you would win ALL BUT ONE. That is a very impressive record, and will easily earn a spot in the finals.

I wouldn't advocate jumping up on a ramp right at the start of the match and saying "This is a statistically sound strategy", because you would give your opponents free reign to fill the rack. But I would say that getting up a ramp should be a priority, and I personally am planning on taking a full 30 seconds for this so that I make sure we have enough time to pick up these essential points.

What does everyone think about this?


Oh yeah, this data constitutes 270 qualification matches, and none of the finals matches, as those are clearly a different caliber, and need to treated as a seperate type of match so as to prevent skewing the data. I have attached my Excel spreadsheet if you want to look through it.

[David Brinza]

It is our belief that if you build a good rampbot, your alliance partners will surely come on board for "easy" points.

I also think that over the next few weeks, ringers will be hung more effectively and later into the match. Last second "landings" will become more frequent. I suspect that spoilers will even come into play. But, in the end, there's nothing like having a good ramp to roost upon after a tough match!

[EricH]

I dunno. Week 1 has historically been the weakest in terms of scoring. Week 2, teams find their playing style, hit their stride, and find out what the refs are going to call and not call, and then it gets better from there. Could you redo this after Week 2? I can think of at least one regional where there will be at least 4 top rack scorers, quite probably more. Call me skeptical, but it'll probably be Week 3 before things settle down to what it's actually be like.

[Jeffrafa]

I don't think having a solid ramp bot is quite as sound of a strategy as is suggested. It can be effective for qualifying matches as it stands now, but it was surprising to see just how infrequent lifting occurs. To get a combination of reliable ramps, capable climbing bots, and timing & coordination to complete the lift is slightly easier than waiting for planets to align.

I recall that in our ~8 matches we were only with solid lifting robots 4 times, 3 of which we got up for. Our 6wd drivebase is quite capable of most of the ramps at our regional, but many teams with casters, holonomic drive, or other systems had near impossible times with most of the lifter designs. I also know of some teams who never once were allied with a lifting robot during semifinals - so the 'random' pairings are also a significant factor.

Although an easy lift may be a sure fire way to success in current qualifying matches, I believe this will change as ring robots become more finely tuned, and as strategy shifts in finals.

Our team competed it the Pacific Northwest Regional where we made it to the semifinals before losing to the alliance that eventually won finals. Yes, a ramp bot can be useful for an easy 30 - 60 points - but in finals, its value comes down to a matter of time value. The biggest factor with lifters is not IF they can lift, but how quickly that can be done. If a robot can cut back at only 15 seconds left and reliably get the lift, then yes - its most likely worth it, but otherwise when enough rings are on the rack a lot can happen in the last 30 seconds.

From the last few matches of qualifying through all of the finals matches, our robot consistently scored 5 rings, even with decent defense on and off of us. Many of those rounds we also managed to make it back for a lift. At 5 rings, we had accumulated 32 points - just over the value of one robot up. In the remaining 30 seconds, had we not gone back for a ramp, we probably could have more than likely gotten a 6th ring, thus doubling our ring score to 64. Against a lifting team that needs time to lift, the defense drops drastically as the match nears its end. Typically even the best of lifters are headed back from their D roles by 45 seconds, and often robots - including ourselves - would return at 30 to guarantee a successful lift. This has a few drastic effects on the game. For one - scorers who choose to remain out there are free to do their thing with no defense. This means that last 30 seconds getting 2 tubes up by ourselves would not have been impossible - 128 points, far more than any ramp bots.

Now this is just our robot scoring. In these matches we didn't have the opposing alliance scoring on the same level, so nothing restricted us from continuing to extend our rows and multiply points - obviously this is situational. My point is that if our robot can get 5 rings up alone AND have 30 seconds remaining for a successful lift, than it makes little difference point wise if we go for one more tube, or get the 12". With another robot capable of guaranteeing even 3 rings to compliment our 5, we could've completed a row all the way around - completely untouchable, even against 2 lifted robots with a string of 7 tubes.

This game will change drastically when lifters begin finessing their skills and get matched up with other good lifters. Once alliances are capable of 7+ rings guaranteed, you'll start seeing strategic use of spoilers - which only started to appear at our regional.

I'm not saying lifters are worthless. Effective lifts can dominate, especially in qualifying if used correctly. As has been noted - 30 points, the value of one lifted robot, was enough to win over 9 out of 10 qualifying matches in week 1. I suspect a similar trend may carry through week 2, but that as the weeks progress the score needed to consistently pull off a win will gradually rise. Much like previous years - low scores are notorious to first week regionals, but by the fifth week scores of zero will be very uncommon.

- Jeff

[Wayne C.]

Originally Posted by David Brinza

It is our belief that if you build a good rampbot, your alliance partners will surely come on board for "easy" points.

Agreed- but define good.

My definition-
1. sturdy
2. wide enough to allow a sloppy fast entrance
3. durable enough to stay retracted in regular play and not accidentally deploy in mid field
4. well trained team who knows when to deploy and when not to

Observation- if your team is putting up the 7th or 8th in a ring a ramp bot deflecting a spoiler beats the 60 points

WC

[Daniel_LaFleur]

Originally Posted by Wayne C.

Agreed- but define good.

My definition-
1. sturdy
2. wide enough to allow a sloppy fast entrance
3. durable enough to stay retracted in regular play and not accidentally deploy in mid field
4. well trained team who knows when to deploy and when not to

Observation- if your team is putting up the 7th or 8th in a ring a ramp bot deflecting a spoiler beats the 60 points

WC

Agreed.

I would also add that rampbots have to be able to play some tough defense (and that makes your point #3 even more important).

Rampbots also cannot (usually) win by themselves. They are an alliance 'bot and require their alliance partners to be able to score on the rack and climb the ramp (not easy considering some of the ramp angles out there).

[Wayne Doenges]

I dunno. Week 1 has historically been the weakest in terms of scoring. Week 2, teams find their playing style, hit their stride, and find out what the refs are going to call and not call, and then it gets better from there. Could you redo this after Week 2? I can think of at least one regional where there will be at least 4 top rack scorers, quite probably more. Call me skeptical, but it'll probably be Week 3 before things settle down to what it's actually be like.

This might be true if all the teams from all the 1st regionals went to all the other regionals. And it will be true at the Championship,
We are only going to the Boilermaker (that's all we can afford). I'll bet you that allot of the other teams, at Boilermaker, will be their first one also (not proper english, bu hey it's early).
I think lift bots will still be a big factor.
My team (1501) do NOT have ramps, we have wings, and if you can climb 2.25" than we will lift you with ease.

[ALIBI]

I was surprised at the number of teams that did not design to climb ramps. There are what appear to be several good liftobot designs that have not yet competed (i.e. 1501) that may make getting bonus points easier. The team I mentor desiged a robot that could climb last years ramp (aprox. 30 degrees) and deploy a ramp(s) with an angle of about 22 degrees. Ramp(s) is something special that will be reveiled at Boilermaker that I have not seen anywhere yet. To win in the end you will need the bonus points by either being able to get on a ramp/liftobot or be a ramp/liftobot that is user friendly. And, as the weeks progress more and more ringers will be scored making way for the spoilers and even more of a last ditch run for the ramp/liftobot bouns points.

[Ellery]

From the few matches I watched via webcast I agree with Alibi's summary that alot of teams didn't take into account the steep angles it needs to climb and the clearance needed. Also alot of traction issues were seen as well. But I still believe tha rampbots will definitely remain a big player as Wayne C pointed out.

[BrianR]

I'll be glad to redo this after week two.

I agree that the scores will probably rise, but it will be interesting to see how much better teams actually do in the second week.

[VanMan]

The alliance that got into the finals had a ramp bot that would just deploy at the very beginning and do nothing else, which was good except their alliance mates didn't come back in time. The reason we won is because we stopped lots of tubes from being scored with our crazy defense and then our alliance partners drove up us. We had meet with our alliance partners before where they tried to drive up our ramps and neither could so we got rid of our chucks which once their robots where on our they would not come off and 540 got rid of their back bumpers. Without chucks it just makes it so we have to hit the raise button at the perfect time. Our ramps never failed except the one match in elimination where the battery was basically dead.

[BGiraud]

Originally Posted by Wayne C.

Agreed- but define good.

My definition-
1. sturdy
2. wide enough to allow a sloppy fast entrance
3. durable enough to stay retracted in regular play and not accidentally deploy in mid field
4. well trained team who knows when to deploy and when not to

Observation- if your team is putting up the 7th or 8th in a ring a ramp bot deflecting a spoiler beats the 60 points

WC

I agree with Wayne.

We designed for defense first, ramp second, auto. third, and no ringers.

1. Sturdy - Drive train is key!!! You must be able to support your team during the match. Strong, fast and maneuverable.
Result for us: Four cims x 8 wheels. Stripped gears in both right side gear boxes in match 5 and fixed by match 7. Besides that, the feedback was that we dominated as a defender. We were picked up in the first round by the 8 seed for that skill.

2. Wide enough for a sloppy and fast entrance - This was painfully evident when we had only 1.5" on each side of our bumpers to make it onto the RoboWizards ramp. Over 2 matches we made like 8 runs at it and only made it up once. They informed us that they plan to upgrade this at NYC.
Result for us: We had a 40" wide expanding ramp at a 16 degree angle, poor luck leaves it mostly untested in competition. Too many teams couldn't or wouldn't even try. The two ramps we triet to get up were too narrow for us with a 24" track + bumpers.

3. durable enough to stay retracted in regular play and not accidentally deploy in mid field- It is devastating when this happens.
Result for us: The excessive contact while playing strong defense caused our ramp to deploy in the field more than once. We under estimated the forces generated. Even after a solid latch was installed (we had none before)and tested, our team was so shell-shocked they wanted to tie the ramp down during elimination matches to prevent another accident.

4. well trained team who knows when to deploy and when not to - This is easier in the qual rounds and critical in the eliminations.

I feel that a good rampbot is essential for a couple of reasons.
1. If they can play good defense, they can neutralize a top scoring team for the time that they are on them. During this time their partners need to place ringers in specific locations that will prevent the top team from running up the score (ie. positions 1 and 5 on the middle row) while the ramp is deploying.
2. The rampbot has the option of trying to stop a spoiler or preventing more ringers.
3. In a close game if the other alliance has a ramp and you do not, then you have a big problem.

I found out that our 8th seed alliance captain made a mistake and selected two defending rampbots instead of a ramp and a ringer. Big Mistake. Don't try that strategy unless you can score a lot by your self.

The biggest downside of being a rampbot is that you are really at the mercy of your alliance partners. If they can't score well or can't ramp you, then you are really out of luck.

[AHS1599]

I think that the alliances that make it to the top will all have to be able to score a high number of ringers, be able to block the other team and ramp both teammates above the 12" level to win. At the end of the VCU regional this was true and will only become more normal now that everyone has had an opportunity to learn from the first competitions... The spoilers will become a decisive tool in future matches IMHO...

[dtengineering]

Originally Posted by Daniel_LaFleur

Agreed.

I would also add that rampbots have to be able to play some tough defense (and that makes your point #3 even more important).

Rampbots need to be careful when playing tough D. At least once at PNW I saw team 488 (one of the great rampbots at the tournament) get caught downfield. They had gone to the other side of the rack and played great D... but when it came time to go home and deploy their ramps the tables were turned. It was a great strategy by their opponents, and one that... done wisely... will limit the defense abilities of a ramp bot.

Another strategy to mess up the ramp bots included knocking the ringers all over the home zone, so that when the ramps came down they landed on tubes, forcing the bots to take time to clean them out. As entropy would suggest it was easier to make a mess of the home zone than clean it.

As for the robots that deployed early, I saw at least one case where once the ramp bot had committed to deploying at a particular location, one of the opposing team members just rolled down and played D in front of them.

The best play, however... perhaps one of the most exciting moments I've seen in FIRST... came in the final match when one robot was tipped while attempting to climb the ramp.... the robot lay there, helpless.... and victory seemed assured for the other alliance.... then slowly the arm deployed and the tipped robot lifted itself back up and climbed onto the rampbot and was elevated as the buzzer sounded for a 30 point bonus... and a regional championship.

Does anyone have any video of the final match at PNW?? The crowd was almost as loud as it is at the GTR, I think.

Jason

[Fred Sayre]

Originally Posted by dtengineering

Rampbots need to be careful when playing tough D. At least once at PNW I saw team 488 (one of the great rampbots at the tournament) get caught downfield. They had gone to the other side of the rack and played great D... but when it came time to go home and deploy their ramps the tables were turned. It was a great strategy by their opponents, and one that... done wisely... will limit the defense abilities of a ramp bot.

Another strategy to mess up the ramp bots included knocking the ringers all over the home zone, so that when the ramps came down they landed on tubes, forcing the bots to take time to clean them out. As entropy would suggest it was easier to make a mess of the home zone than clean it.

As for the robots that deployed early, I saw at least one case where once the ramp bot had committed to deploying at a particular location, one of the opposing team members just rolled down and played D in front of them.

The best play, however... perhaps one of the most exciting moments I've seen in FIRST... came in the final match when one robot was tipped while attempting to climb the ramp.... the robot lay there, helpless.... and victory seemed assured for the other alliance.... then slowly the arm deployed and the tipped robot lifted itself back up and climbed onto the rampbot and was elevated as the buzzer sounded for a 30 point bonus... and a regional championship.

Does anyone have any video of the final match at PNW?? The crowd was almost as loud as it is at the GTR, I think.

Jason

We recorded the raw footage from the whole competition, so I believe we should have all of the matches. We are slowly getting organized to cut it up and format the video. We will post links here, or at least let interested people know when we have them ready.