2023 Newton Division Scouting Data

[Henrique_Schmit]

With the championship coming to an end, and us having some time to go back home and get organized, I would like to share some of 1156’s scouting data on the Newton Division. We collect this data from the qualifying matches by taking note on what each team scored in every match. This certainly introduces human error into our data, but it also allows for some interesting insights.

A couple of notes on this data. Some texts places contain texts in Portuguese, if you need any translations, let me know. Also, all objects scored have 5/3 points added to their manual values regardless of location to account for the contribution of those objects to links.

The main information we used to understand our field during the event was the ranking by average individual score:

This covers the top 40 teams and is split by auto/teleop/charge station and by grid level on teleop. Here it is interesting to see that, while 1538 and 4522 were outliers and were playing super well, we also had a very deep field, with the top 39 teams in the division averaging 50 points or higher alone.

Another interesting view is the dispersion of average amount of scored objects on high and mid levels on the grid:

Here 1538 really appears as an outlier by averaging a lot of objects on both mid and high. 1023 also featured more on this graph by being the best mid level scorer on our division. Kudos to them who did a similar strategy to ours of making concessions on the high level to be a really specialized robot, they really perfected this strategy.

This year we also started looking at the score rankings for the last 3 qualifications we scouted for each robot. This really helps in finding teams on an upward trajectory:

Some teams that impressed us in this metric by climbing over 15 places above their full qualification rankings were: 1757, 4069, 11, 5990, 9140 and 1629. This helped to put 1757 in our radar and when we watched their matches we could see the potential they had even though they ranked 32nd.

I ran our match simulation tool for all elimination matches. I will attach a zip file with screenshots of all expected results here, but the most impressive was definitely the finals:

While averages will almost always be lower than real match values if no robots stop working, alliance 3 managed to beat their average by 40 points twice. Throughout the eliminations 4329, 1746, 3184 and 5804 overdelivered on our expectations and I am really curious on how they managed to synergize so well.

We also have individual pages with more detailed data for each team. If your team was part of the Newton division, feel free to send a direct message and we can get your team’s individual information. Also feel free to ask for any other insights and we will do our best to provide them.

I hope everyone had as much fun as we did on this division!
Newton Elimination Predictions.zip (6.6 MB)

[Kellen_Hill]

We just kept winning somehow. We didn’t stop to ask too many questions

As the drive coach for 1746, I can provide some perspective here but am always interested in an outside perspective and I think you are asking interesting questions.

In reality, I think there were a number of factors at play here, luck being a big one. Consistency of our alliance compared to other alliances was another substantial factor. We managed to win matches 8 & 11 by a combined total of 6 points. One of those two matches being a tie that had to go to the 3rd tie breaker of total balance points.

As for our two matches in finals with 200+ points, one match involved 10 points of penalties and the other gave us an additional 20 so in reality that puts our over performance in a more realistic ballpark.

Luck and external factors aside, I think we ended up on an alliance that was well shaped to not get in each other’s way. While we occasionally bumped into each other here and there, we each stuck to our roles pretty well. We all had our own substation to load from since we always played two double substation bots (4329/5804 and 3184) and us handling single substation. Traffic at that end was usually pretty smooth. As for the grid side of the field, we had some adjustments to make based on lineups. Whenever it was 4329/1746/3184 on the field, 3184 and us would take care of the high links and then move down while 4329 focused on mid links before moving to low. This involved a bit more coordination due to the overlapping links. This was simplified a bit when 5804 played since everyone pretty much just covered the grid in front of their station.

Our alliance’s scores bumped up a bit in finals as 4329 decided to bench themselves and instead play 5804. Our scouting numbers had 5804 putting up a couple of additional game pieces and it showed in finals. The switch was seamless enough since 5804 and 4329 were running the same auton mode and didn’t change the teleop plan too much. I wish I could say there was some special element to our routine that really boosted our performance but it really was coming up with a basic plan and just executing it again and again. Again, our consistency was a big deal.

As for scouting numbers, I think it can be hard to compare quals matches and elims matches in this game. A lot of the time, teams are bypassing the highest scoring game piece for a piece that might instead impact their chances of earning the link RP. This might slow teams down a bit and cap their points output. We also had quite a few quals matches where we had to play with two other robots who also specialized at the single substation so our performance was slowed for the benefit of the whole alliance in some matches. If you isolated all of our quals matches where we got to be the only single substation bot, I’m sure you would see an associated point bump as a result. We got that point bump in every elims match since the single substation was not being shared.

Again, happy to hear outside thoughts and happy to answer any questions. Newton was obviously quite a blast for us and an experience our team won’t soon forget.