Most competitive regionals?

Now that regional lists are mostly shaken out, I think it’s time to make a prediction thread for the most competitive regionals of 2018!

Three are on my radar:

Silicon Valley: 254, 971, 973, 3132, as well as 192, 1868, and 604 who aren’t at the level of the first four, but had great seasons in 2017.

Arizona North: 254, 987, 842, 2122, 1011

Dallas: 148, 118, 2848, 1477

Interested to hear y’all’s thoughts!

Central New York will be pretty competitive for a week 1 event it seems.

I’ll focus on the Midwest.

Great Northern: 525, 876, 1986, 2883, 3130, 4539, and 5172.

Heartland: 1710, 1730, 1806, 2451, 2481, and 4499.

Central Illinois: 1736, 1756, 2169, 2338, 2481, 2826, and 4143.

Minnesota North Star: 1625, 2175, 2451, 2512, 2883, and 3130.

Midwest: 16, 111, 341, 1511, 1706, 2338, and 2451.

Seven Rivers: 1736, 1756, 1775, 1986, 2062, 2169, 2826, and 3928.

Just regionals or can districts be discussed as well?
In New England: North Shore and Hartford

Sacramento and Central Valley both seem pretty competitive both with 1323 and 1678. I agree that SVR will probably one of the most competitive events though.

Central New York looks pretty stacked, with a lot of big names from across the state along with a few newcomers to NY (27, 319, 4253, others).

For Ontario McMaster seems to have dethroned Waterloo as the most competitive event.

McMaster (Week 6): 188, 1241, 2056, 3683, 4039, and 5406

Waterloo (Week 4): 1241, 2056, 3683, 4678

MVR is pretty stacked.

No Week 1 event is going to be top-notch gameplay, but adjusting for that Palmetto should be interesting. I’m probably missing others between it being early and me being out of the south last season, but here’s who I’m seeing:

The usual local suspects: 343, 1319, 1876, 3490, 4451, 6366, etc.
The usual not-local suspects: 1369, 2614, 3140, 3824, 4265
The unusual suspects: 125, 900*, 263, 2383, 4013**, 4481

*I don’t care that they’ve attended 10 out of 14 Palmetto Regionals and that they’re only about 200 miles from Myrtle Beach, getting off the waitlists as a district team is never easy and the 900 of today is very different from the 900 that bemoaned the state of dancing at the earliest Palmetto Regionals.

**No, 4013 didn’t make playoffs anywhere last year. But the team that won the last two Palmetto RCAs now has a clock in their trophy case, and 4013 has won an RCA at three different regionals in the last four years and got “just” EI in the fourth. Those medals are shiny too. Shinier, if you count the printing on the ribbon.

I do expect to see some good matches on the ISR District #1 event, with teams like 1690, 2630, 1574 and more

Is the most competitive regional the one in which 2 of the three teams that win have the best chance at the championship, or the one where anyone has a chance to win because all teams are good?

I’d say the second, personally.

Most competitive imo means most alliances have a shot at winning. There are plenty of regionals where there are two or three teams that represent an easy prediction, but the most competitive are a toss up because of the volume of good to great teams. Alliances 1-8, given good scouting, have a shot at a competitive regional.

I’ve always defined “the most competitive event” by the event that will exhibit the highest level of play in the playoffs. Usually this means there’s more than a few great teams at an event so the number one alliance doesn’t just run away with things.

Minnesota events are much more competitive this year than I think I’ve ever seen them. 1625 and 2451 at North Star, 233 at 10,000 Lakes, 141 and 706 at Northern Lights, 525 and 1986 at Great Northern on top of all the great MN/ND teams that usually show up. I’m excited to see how things play out!

As has been mentioned, there are numerous ways to define “competitive.” I have attached a sheet which has the start of season Elo ratings for all teams for all events. You can use this data to realize your own definition of “competitive.” I have found the average Elo ratings and standard deviations of Elo ratings for the top 4, 8, 24, and all teams at each event.

Personally, in week 1 I am most interested in high-scoring playoff matches, since these matches are the best testing grounds for advanced strategies that we may see later in the season. So my metric of choice is average top 24 Elo, and I’ll be keeping an eye on Great Northern, Palmetto, and Central New York.

However, if you are interested in seeing playoff matches where there is not a really clear favorite, your best bet is probably to look for events with a low top 24 standard deviation. In week 1, Miami Valley, Northern Virginia, and Monterrey all stand out.

If you want an event where we have no idea who will win beforehand, I recommend watching one of the new foreign regionals like Istanbul or Shanghai. These events have a very low standard deviation of Elo ratings for all teams, which means all teams are relatively close to the same skill level.

most competitive events.xlsx (1.14 MB)

Sounds like a good time to revive the BBQ project from a long time ago.

Another potentially useful metric that can be used is the normalization of OPR distribution. Looking at the average weight from the top 24 teams, here are the 16 highest ranked events.

Seven Rivers Regional
ONT District McMaster University Event
South Florida Regional 
Midwest Regional
Buckeye Regional
Minnesota North Star Regional
Silicon Valley Regional
Central Valley Regional
MAR District Montgomery Event
Wisconsin Regional
Idaho Regional
Iowa Regional
IN District Tippecanoe Event
NE District Hartford Event
Greater Pittsburgh Regional
FIM District Milford Event

I also ran the same code, but built the models using data from 2005-2010 and it yielded some interesting results. I attached both CSVs for fun.

2018-Event-Difficulty.csv (7.38 KB)
2018-Event-Difficulty-2005-2010.csv (7.36 KB)

Using this metric then West Valley is alwats a perfect fit. By far the closest matches you’ll see in elims, and everyone really has a fair shot at winning if their picks pan out.

I would also say another way to describe an event as ‘competitive’ is by asking “How unpredictable is the outcome?”

When you increase the amount of FRC teams participating which are deemed “high-performing” via some combination of objective and subjective metrics, the difficulty by which you can guarantee a certain outcome also increases.

On an even more tangential note and from the perspective of any team looking at their odds of winning an event, there’s a spectrum of cases which range from most to least favorable in terms of guaranteeing a certain outcome of a competition:

  • Case 1:
    You are the most dominant* team at your event. You believe you will seed 1st and get the best shot at winning eliminations.
  • *Case 2: **
    You and one other FRC team are the two most dominant
    teams at an event. There’s some unfavorability in the other team seeding first and having the option of not choosing you, but you firmly believe no other option is truly likely and you will therefore be with them for elims
  • Case 3:
    There are two dominant* FRC teams at an event in addition to yourself, and so, you may very well end up on the wrong side of the field against these two teams during eliminations.
  • Case 4:
    There are three or more FRC teams which are considered dominant* at the event you are attending. Even sets of good* FRC teams in addition to your own means they could all end up pairing up with each other and you are left on your own, and odd sets of dominant* FRC teams means you may end up on one of these alliances with a moderately-difficult, but not impossible bracket.

*The use of ‘dominant’ and ‘good’ is pretty subjective but there’s definitely a certain category of teams most agree upon as highly likely to outperform, and that’s what I mean here.

This is most likely something that doesn’t have to outright be said, but I think the competitiveness of an event is definitely correlated to the # of dominant* FRC teams and whether or not that number is odd or even due to the nature of alliance selections.

In the PNW it’s probably one of these four events:

Wilsonville (Week 2): 955, 997, 1425, 1540, 2471, 2990, 4488
Auburn (Week 3): 948, 1778, 1983, 2046, 2557, 2907, 3238, 3574, 4131, 4469, 5803
Glacier Peak (Week 4): 492, 1983, 2522, 2930, 3663, 4450, 4488, 4911
Auburn Mountainview (Week 5): 360, 492, 1318, 1425, 1778, 2046, 2557, 2811, 2907, 3574, 4131