IRI ranking projections

Competition Off-Season Events
1

Here are the ranking projections for IRI according to my event simulator. This will be a trial run for my new pre-schedule projections. I’ll be updating this thread after we get a schedule and once or twice a day during IRI.

Team	average rank	1 seed	Top 4	Top 8	Top 12	Top 15
2056	10.0		15%	41%	58%	74%	79%
2590	10.7		15%	41%	56%	72%	75%
1619	14.6		8%	27%	41%	59%	64%
195	15.9		5%	20%	37%	51%	62%
118	16.0		6%	22%	38%	52%	60%
624	18.3		5%	19%	32%	47%	52%
67	18.3		5%	18%	31%	47%	52%
319	18.4		4%	16%	31%	45%	53%
225	18.9		3%	14%	30%	41%	52%
5406	20.2		3%	14%	26%	41%	48%
2791	21.0		3%	12%	25%	38%	46%
3707	22.3		2%	11%	23%	34%	42%
3357	22.4		2%	11%	23%	33%	42%
217	25.4		2%	8%	19%	28%	35%
4028	25.7		2%	8%	16%	27%	34%
3452	26.3		1%	7%	16%	25%	33%
234	27.3		1%	6%	14%	24%	31%
133	27.3		1%	7%	15%	23%	31%
359	27.9		1%	6%	14%	22%	29%
694	28.7		1%	6%	14%	22%	27%
865	29.7		1%	4%	12%	18%	25%
1741	30.1		1%	4%	11%	18%	25%
868	30.7		1%	4%	11%	17%	24%
1706	30.8		1%	5%	11%	17%	23%
3478	31.4		1%	4%	11%	16%	23%
2481	31.6		1%	5%	11%	18%	22%
3538	31.8		1%	4%	11%	16%	21%
340	32.3		1%	4%	10%	16%	21%
2337	32.9		1%	4%	10%	15%	20%
1533	33.2		1%	3%	8%	13%	19%
33	33.6		0%	3%	8%	13%	18%
1731	34.1		1%	3%	9%	14%	18%
494	34.7		1%	3%	8%	13%	18%
1747	34.9		0%	3%	8%	13%	17%
245	35.7		0%	2%	6%	11%	16%
3641	37.1		0%	2%	6%	9%	14%
2834	37.5		0%	2%	6%	9%	13%
1640	37.9		0%	2%	6%	10%	13%
1218	39.0		0%	2%	5%	8%	12%
4265	39.1		0%	1%	4%	8%	11%
302	39.8		0%	2%	5%	8%	11%
2614	39.8		0%	2%	5%	8%	11%
2771	41.0		0%	1%	4%	7%	10%
708	41.1		0%	1%	3%	6%	9%
1806	41.3		0%	2%	5%	7%	10%
4967	41.8		0%	1%	2%	5%	8%
1024	41.8		0%	1%	4%	6%	9%
1102	42.2		0%	1%	3%	6%	9%
2451	42.3		0%	1%	4%	6%	8%
20	42.9		0%	1%	3%	6%	8%
2168	44.0		0%	1%	3%	6%	8%
4587	44.1		0%	1%	2%	5%	7%
2013	44.2		0%	1%	3%	5%	7%
384	44.8		0%	1%	2%	5%	7%
829	45.1		0%	1%	2%	4%	6%
88	45.6		0%	1%	3%	5%	6%
4944	46.2		0%	1%	2%	4%	6%
469	46.9		0%	0%	1%	3%	4%
1710	47.1		0%	1%	2%	4%	5%
3847	48.2		0%	1%	2%	3%	5%
1720	48.6		0%	0%	1%	2%	4%
3940	48.6		0%	0%	1%	3%	4%
5254	48.8		0%	0%	2%	3%	4%
2655	50.5		0%	0%	1%	2%	3%
2826	52.4		0%	0%	1%	2%	2%
1746	52.7		0%	0%	1%	2%	3%
2363	54.0		0%	0%	1%	1%	2%
6800	54.3		0%	0%	1%	1%	2%
2468	55.3		0%	0%	0%	1%	1%
4499	56.0		0%	0%	0%	1%	1%

Looks like a very deep field, 2056 and 2590 are the only teams I have at over even odds to rank in the top 8, due in large part to their proficiency at getting the face the boss RP. It’ll be interesting to see how the predictions change after the schedule is released.

Feel free to critique/criticize all you want. If you see any consistent biases please let me know, my model has certainly had biases before. Just understand that if you wait until after the event to point out a flaw, I’m not going to give your opinion much weight since you’ve likely been deceived by hindsight bias.

2

It make me excited just the fact that basically no team has a extremely high chance to get into the top 8 it just shows how fierce the competition will hopefully be.

3

The biggest variable in predictions will be the student drivers.

From experience, I know there are at least a few teams that use off season events (even IRI) as practice and try-outs for next year, especially when they have graduating drive teams.

4

Totally agree - I know we are bringing an entirely new drive team. All of our drive team members from last year graduated. Time to throw them into the toughest competition and see what happens.

5

Yeah, I don’t presently have any kind of correction for this in my model, which probably causes the predictions to be more overconfident than they would otherwise be. Without knowing exactly which teams are swapping drivers, the best I can do would be to revert all skill metrics toward the mean some amount. I’ll probably look into this eventually.

6

Our ranking is preposterous!!!
Its way to high.

7

From a cursory look at these projections, I assumed that the sub-components were already heavily regressed to the mean. I think that’s safe way of doing things in general and pretty much standard among many of the major sports projection systems. (DVOA projections, ZiPS, etc.)

8

this 2791 prediction is almost as spicy as us seeding #1 on Darwin

9

If you disappoint me again I might have to throw a special 2791 penalty into my model.

10

No, I haven’t regressed anything toward the mean yet, which means that the Elo ratings used here are each team’s end of season rating and the calculated contributions used here are each team’s max calculated contribution in that category from the season. I regress Elos toward the mean 20% (in addition to factoring in the team’s Elo from two seasons ago) between seasons and calculated contributions toward the mean 10% between seasons though. I’m pretty sure my off-season predictions could be improved by regressing a bit toward the mean, but I don’t know off hand what would be an optimal amount, and the optimal value likely even varies across off-season events, depending on how close we are to the end of last season and how “seriously” teams take the event.

11

I have a question about how the program/you discover the average rank?

12

Are you asking how the probability distributions generally are created? Or just how do I get an average after I have calculated probabilities for all ranks?

For the latter, all I need to do is use the definition of mean for a discrete probability distribution, which is:

Where in our case, x is a rank and P(x) is the probability of getting rank x.

For example, say a team has a 70% chance of seeding first, 20% chance of seeding second, and a 10% chance of seeding third. Their average rank will then be (70%)*1+(20%)*2+(10%)*3=0.7+0.4+0.3=1.4

13

Updated projections with the new team list (no 2363). No really huge changes, I just wanted to make it easier to compare with future projections.

Team	average rank	1	Top 4	Top 8	Top 12	Top 15
2056	10.0		15%	42%	58%	74%	78%
2590	10.8		15%	41%	55%	72%	75%
1619	14.6		8%	26%	41%	58%	64%
195	15.8		5%	20%	37%	51%	62%
118	16.2		6%	22%	38%	52%	59%
624	18.1		6%	20%	32%	48%	53%
319	18.3		4%	17%	31%	45%	54%
225	18.6		3%	14%	30%	41%	52%
67	18.7		5%	18%	31%	46%	52%
5406	20.3		3%	14%	26%	41%	47%
2791	21.3		3%	12%	24%	37%	44%
3357	22.2		2%	11%	23%	33%	42%
3707	22.4		2%	11%	23%	33%	42%
217	24.6		2%	9%	20%	29%	36%
4028	25.6		1%	8%	16%	28%	35%
3452	26.1		1%	6%	16%	24%	32%
133	26.8		1%	7%	16%	23%	31%
234	27.5		1%	6%	13%	23%	30%
359	27.8		1%	6%	14%	22%	30%
694	28.4		1%	6%	15%	22%	28%
865	29.6		1%	4%	11%	18%	25%
1741	30.3		1%	4%	11%	17%	24%
868	30.7		1%	4%	11%	17%	23%
1706	30.7		1%	4%	11%	17%	23%
3478	31.2		1%	4%	11%	17%	23%
3538	31.4		1%	4%	11%	17%	22%
2481	31.6		1%	4%	11%	17%	22%
340	31.8		1%	4%	10%	16%	21%
2337	32.4		1%	4%	10%	16%	21%
33	33.2		0%	3%	8%	13%	19%
1533	33.4		0%	3%	7%	12%	18%
1731	34.0		1%	3%	9%	14%	18%
494	34.3		1%	3%	8%	13%	18%
245	35.0		1%	3%	7%	13%	17%
1747	35.0		1%	3%	8%	13%	17%
3641	36.9		0%	2%	5%	9%	13%
2834	37.0		0%	2%	6%	10%	14%
1640	37.6		0%	2%	6%	10%	14%
1218	38.6		0%	1%	5%	8%	12%
4265	38.7		0%	2%	5%	8%	12%
302	39.3		0%	2%	5%	8%	12%
2614	39.6		0%	2%	5%	8%	12%
2771	40.5		0%	1%	4%	7%	10%
708	40.6		0%	1%	3%	6%	10%
1806	40.7		0%	2%	5%	8%	10%
4967	41.4		0%	1%	3%	5%	8%
1024	41.7		0%	1%	4%	6%	9%
1102	41.8		0%	1%	4%	6%	9%
2451	42.0		0%	1%	4%	6%	9%
20	42.5		0%	1%	3%	6%	8%
2168	43.2		0%	1%	3%	6%	8%
2013	43.6		0%	1%	3%	5%	7%
4587	44.0		0%	1%	3%	5%	7%
384	44.3		0%	1%	3%	5%	7%
829	44.9		0%	1%	2%	4%	6%
88	45.4		0%	1%	3%	5%	6%
4944	45.8		0%	1%	2%	4%	6%
469	46.7		0%	0%	1%	2%	4%
1710	46.8		0%	1%	2%	3%	5%
3847	47.4		0%	1%	2%	4%	5%
1720	48.2		0%	0%	1%	2%	4%
5254	48.7		0%	0%	1%	2%	4%
3940	48.7		0%	0%	1%	2%	4%
2655	50.5		0%	0%	1%	2%	3%
1746	52.2		0%	0%	1%	2%	3%
2826	52.2		0%	0%	1%	2%	2%
6800	53.9		0%	0%	1%	1%	2%
2468	54.7		0%	0%	0%	1%	1%
4499	55.4		0%	0%	0%	1%	1%

I’m surprised by how confident people seem to be that 2056 will seed first. They may well be the favorites, but I’d take the rest of the field over them to seed first in a heartbeat. Indeed, every team except 2468 managed to seed first in at least one of my simulations.

14

Prove him wrong, Team Appreciate.

15

I wonder if there’s a way that you can take into account things other than pure stats…this works super well for in season events, but offseason is hard because robots/drivers can change spontaneously.

tldr please make an AI that can consider everything ever :wink:

Looking forward to comparing this to how it actually shakes out!

16

2363 is still attending IRI. Not sure why they fell off the TBA list.

Some of these predictions are spicy, to say the least. I find it interesting that 2056 and 2590 have near equal chances of seeding despite 2056 going 17 for 17 on 90 point endgames on Tesla, while 2590 went 12 for 16…

17

I just ran 100000 simulations and 2468 seeded first in 26 of them, so there is hope. Feeling like doc strange over here:

18

Meh, AIs are overhyped: https://www.chiefdelphi.com/forums/showthread.php?t=165122

19

I’m expecting that the ranking point spread at the top will be very narrow such that one good or bad match by any one of a dozen teams will result in large swings in their position, even toward the end of quals. Certainly this was the case with Turing. I think there are a large number of teams that have the capability of getting both of the bonus RPs in every match so their RP average will be based on the W-L ranking points. That will depend to a large degree on the luck of the schedule and other intangibles.

20

It’s all part of our strategy. I everyone thinks we aren’t attending then they won’t see us coming!