God, I hate tables of data, but this site has reduced me to that. Mea culpa.
What is a seed worth?
I have been attempting to answer this question for a number of years. The challenge is that the data is not necessarily stable. Advancement points are pretty predictable, placement points are less so, and bonus points are very unpredictable. But we wont let that stop us, will we? For this iteration I am using the data that @Cali_Nittany so generously provided for us. I built these models off of 2017 - 2024 (the 16 and 33 seed eras).
What is the probability of AAing?
In spite of how offensive to the eyes the above table is, I am going to push forward with one more god-awful table.
One of the things that bothers me when reading Intermat or Flo predictions about AA's or scoring is that they accept the seeds or their ranks as predictions, even though in Flo's case they also speak out of the other side of their mouth and say they are not predictions. But by treating them as binary (1-8 = AA and placement points, 9-33=no AA, and no placement points), they are also using them as de facto predictions.
But, the reality is that a #7 or #8 seed only AA's about half the time. And a #9 or #10 seed has a decent chance to AA. There have even been #1 seeds who have not made the podium. As a result, I like to take a probabilistic approach to these situations.
Oof, the editing tools are so primitive. I could not get the second table to center align unless I did it a cell at a time. Brutal. Sorry.
Teaser: The PSU numbers are better.
What is a seed worth?
I have been attempting to answer this question for a number of years. The challenge is that the data is not necessarily stable. Advancement points are pretty predictable, placement points are less so, and bonus points are very unpredictable. But we wont let that stop us, will we? For this iteration I am using the data that @Cali_Nittany so generously provided for us. I built these models off of 2017 - 2024 (the 16 and 33 seed eras).
Seed | Expected Points |
1 | 19.7 |
2 | 16.1 |
3 | 13.4 |
4 | 11.4 |
5 | 9.8 |
6 | 8.4 |
7 | 7.3 |
8 | 6.4 |
9 | 5.6 |
10 | 4.9 |
11 | 4.3 |
12 | 3.7 |
13 | 3.3 |
14 | 2.9 |
15 | 2.6 |
16 | 2.3 |
17 | 2.1 |
18 | 1.9 |
19 | 1.7 |
20 | 1.5 |
21 | 1.4 |
22 | 1.3 |
23 | 1.2 |
24 | 1.1 |
25 | 1.1 |
26 | 1.1 |
27 | 1.0 |
28 | 1.0 |
29 | 1.0 |
30 | 1.1 |
31 | 1.2 |
32 | 1.5 |
33 | 2.1 |
What is the probability of AAing?
In spite of how offensive to the eyes the above table is, I am going to push forward with one more god-awful table.
One of the things that bothers me when reading Intermat or Flo predictions about AA's or scoring is that they accept the seeds or their ranks as predictions, even though in Flo's case they also speak out of the other side of their mouth and say they are not predictions. But by treating them as binary (1-8 = AA and placement points, 9-33=no AA, and no placement points), they are also using them as de facto predictions.
But, the reality is that a #7 or #8 seed only AA's about half the time. And a #9 or #10 seed has a decent chance to AA. There have even been #1 seeds who have not made the podium. As a result, I like to take a probabilistic approach to these situations.
Seed | Probability of AA |
1 | 98.3% |
2 | 90.9% |
3 | 83.3% |
4 | 75.6% |
5 | 68.1% |
6 | 60.6% |
7 | 53.4% |
8 | 46.5% |
9 | 40.0% |
10 | 33.9% |
11 | 28.3% |
12 | 23.2% |
13 | 18.6% |
14 | 14.6% |
15 | 11.1% |
16 | 8.2% |
17 | 5.9% |
18 | 4.1% |
19 | 2.8% |
20 | 2.0% |
21 | 1.6% |
22 | 1.6% |
23 | 2.0% |
24 | 2.5% |
25 | 3.5% |
26 | 4.5% |
27 | 3.5% |
28 | 3.0% |
29 | 1.5% |
30 | 1.3% |
31 | 1.3% |
32 | 2.5% |
33 | 2.5% |
Oof, the editing tools are so primitive. I could not get the second table to center align unless I did it a cell at a time. Brutal. Sorry.
Teaser: The PSU numbers are better.
