They got the press release for the new qualification system out before Muskoka…would have been real shit of them to not have their shit together, run this race, then announce a new qualifying system on Tuesday.
Looking at Jönköping, I would say it will be hard to the classes at 1.0 grade or closet to it to take a slot.
3rd at 18-24 - Time of 4:10 on a long T1 course. He was 9 overal. Got 11 AG:s in front of him in 35-39 and 45-49 and got punisched to a place of 66 in the grade adjustment.
Dno, I think this needs to be abit more tuned
2 Likes
@dsring maybe can scrape and show us this?
Or @x-chain?
1 Like
I’ve got a little something cooking. Stay tuned!
3 Likes
I agree, it looks a bit too exessive at times.
Easiest way.
- Go to 70.3 results page…
example (Results | IRONMAN 70.3 New York) - In Chrome open dev tools. network tab
- change AG category (like filtering by M40-45, doesnt matter which)
- in dev tools search for wtc_eventid
Hope the helps…
p.s. step 3 is important. You won’t find wtc_eventid on unfiltered results screen
3 Likes
I’m generally a fan of this new system, but I have to also agree. I don’t have any data to back this up, but I think it’s likely that the range of times within the top 20% of say M 25-29 is much tighter than M 60-64. So, if you’re in the top 5% of M 60-64, your relative time is going to look much better against an average with a larger range that’s averaging in times that are significantly slower than the top end of the age group.
Don’t get me wrong, there are some really fast times looking at the Muskoka results for some of the older age groups that I think should be ranked at/near the top. Just seems a little off if first overall at 4:11 is getting 8th in the adjusted standings. Same with Jonkoping…first overall at 3:55 is good for 11th adjusted.
3 Likes
Just so I’m aware, no one can actually have a “worse” scored than than actual time right? 1.0 co-efficiency just means they get no decrease, but a 3:55 can’t suddenly become a 4:06 adjusted can it (if it’s from the 1.0 AG)? You can only have an improved peformance time right and that’s based on the individual AG co-effiicent.
1 Like
Some more number crunching done.
This is based on the 33 most recent IM races in the formidable work of CoachCox, from Sweden 2024 onward. I have taken the number of slots for Kona (female), and multiplied by 1.4 to get a number of slots for each race that reflects about how many they seem to be getting in the coming cycle. These slots, I allocate both by the old model and the new model. No rolldown. Next, I compare and summarize. Easy as that!
It’s a busy table, so I’ll start with an explanation of the columns.
First four are the aggregate of all the races
- old_qual: # qualified from all 33 races by the old model
- new_qual: ditto, new model
- delta: new_qual - old_qual
- % delta: delta / old_qual
Next three counts how many of the races that end up with fewer slots for that age group, the same or more. Doesn’t matter if the difference is -5, 1 or 7 in a particular race. If it’s not 0, that race counts as 1 with loss or gain.
- races_w_loss: # of races with fewer slots with the new model.
- races_no_change: # of races with no change
- races_w_gain: # of races with more slots
Last five: largest loss or gain in any single race
- max_loss: I think you get it…
- max_pct_loss: max_loss / slots in old model
- mean_delta: mean of #lost and #gained in all 33 races
- max_gain
- max_pct_gain
(Saying that the agegroup as such gets a number of slots in the new model is incorrect, but you know what I mean.)
| old qual | new qual | delta | % delta | races w loss | races no change | races w gain | max loss | max pct loss | mean delta | max gain | max pct gain | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| agegroup | ||||||||||||
| F18-24 | 32 | 34 | 2 | 6.2 | 0 | 29 | 2 | 0 | 0 | 0.1 | 1 | 100 |
| F25-29 | 42 | 51 | 9 | 21.4 | 3 | 22 | 8 | -1 | -50 | 0.3 | 3 | 300 |
| F30-34 | 52 | 62 | 10 | 19.2 | 9 | 13 | 11 | -1 | -50 | 0.3 | 4 | 200 |
| F35-39 | 51 | 73 | 22 | 43.1 | 4 | 15 | 14 | -1 | -50 | 0.7 | 6 | 300 |
| F40-44 | 60 | 76 | 16 | 26.7 | 5 | 17 | 11 | -1 | -50 | 0.5 | 4 | 150 |
| F45-49 | 58 | 74 | 16 | 27.6 | 6 | 16 | 11 | -1 | -50 | 0.5 | 4 | 200 |
| F50-54 | 52 | 65 | 13 | 25.0 | 5 | 20 | 8 | -1 | -50 | 0.4 | 6 | 200 |
| F55-59 | 37 | 63 | 26 | 70.3 | 1 | 22 | 10 | -1 | -50 | 0.8 | 6 | 300 |
| F60-64 | 34 | 51 | 17 | 50.0 | 0 | 26 | 7 | 0 | 0 | 0.5 | 6 | 600 |
| F65-69 | 26 | 29 | 3 | 11.5 | 0 | 24 | 2 | 0 | 0 | 0.1 | 2 | 200 |
| F70-74 | 11 | 11 | 0 | 0.0 | 0 | 11 | 0 | 0 | 0 | 0.0 | 0 | 0 |
| F75-79 | 5 | 5 | 0 | 0.0 | 0 | 5 | 0 | 0 | 0 | 0.0 | 0 | 0 |
| M18-24 | 76 | 64 | -12 | -15.8 | 17 | 11 | 5 | -2 | -66 | -0.4 | 2 | 100 |
| M25-29 | 138 | 98 | -40 | -29.0 | 20 | 8 | 5 | -4 | -80 | -1.2 | 3 | 100 |
| M30-34 | 170 | 129 | -41 | -24.1 | 21 | 5 | 7 | -7 | -80 | -1.2 | 3 | 150 |
| M35-39 | 174 | 128 | -46 | -26.4 | 22 | 5 | 6 | -6 | -80 | -1.4 | 2 | 50 |
| M40-44 | 189 | 149 | -40 | -21.2 | 20 | 8 | 5 | -5 | -80 | -1.2 | 5 | 71 |
| M45-49 | 177 | 130 | -47 | -26.6 | 23 | 6 | 4 | -6 | -85 | -1.4 | 5 | 133 |
| M50-54 | 170 | 190 | 20 | 11.8 | 11 | 6 | 16 | -5 | -75 | 0.6 | 6 | 150 |
| M55-59 | 114 | 130 | 16 | 14.0 | 14 | 6 | 13 | -4 | -80 | 0.5 | 7 | 166 |
| M60-64 | 69 | 96 | 27 | 39.1 | 5 | 11 | 17 | -1 | -50 | 0.8 | 4 | 400 |
| M65-69 | 35 | 60 | 25 | 71.4 | 0 | 19 | 14 | 0 | 0 | 0.8 | 4 | 400 |
| M70-74 | 30 | 34 | 4 | 13.3 | 0 | 26 | 4 | 0 | 0 | 0.1 | 1 | 100 |
| M75-79 | 11 | 11 | 0 | 0.0 | 0 | 11 | 0 | 0 | 0 | 0.0 | 0 | 0 |
| M80-84 | 2 | 2 | 0 | 0.0 | 0 | 2 | 0 | 0 | 0 | 0.0 | 0 | 0 |
So find your (or your friends) age group and see what to expect!
My take-away: Bigger impact than I expected, both in single races and in the aggregate. Also more variability in the changes between races. Same age group can end up with losing 4 of 5 slots in one race, and adding 3 to 3 in another. This confirms that luck had very much to do with it in the old model.
6 Likes
That’s correct. Just theorizing that the coefficients potentially are more generous to age groups where there’s a bigger range of times within the top 20%. Also fully acknowledging that there have been only a few races with this in place and it’s likely we’ll see a good amount of variation in how this actually plays out.
Given every coefficient is <=1.0, maths is (BD’s) friend.
I wonder how the ‘generosity of coefficients’ will be assessed.
Given that it’s 20% (so not a set number) and in the sharpest of pointy-end races: the last 5 IMWC held at Kona, I think we need to see some maths to support that theory.
There have been exactly Zero “races with this in place” for full distance and only two (Jonkoping and Muskoka) 70.3s with separate W and M. We’ll have to wait for Kalmar and Gopenhagen in August for the first full distance play (and those races are offering starts in Kona (2025) for the women as well, btw).
I would like to see this for a year before the Kona/Nice Split. Say 2019. I feel that’s a more comparable year than using data from 2024.
IIRC, they’re just using the last 5 years where there was a single day Kona. In other words, 2015-2019.
Weather is too much a factor in figuring out times to even use 2022
You recall correctly: this is what was/is said:
“We first create the “Kona Standard”, which is the average finish time of the top 20% of Kona finishers (per age group for each gender) over a rolling 5-year (editions) period. We use the top 20% of finishers and a rolling 5-year period (single-day Kona editions) to minimize the impact of any outliers.”
I was referring to the analysis above, not the KS
If they are going to add, on as average 130 more female slots to Kona, fair enough. I’m surprised to see it comes at the expense of the particular male cohorts it does while at the same time adding to some of the older male cohorts. It’s just a function of the formula, but it’s a little quirky the older M AGs got a slot boost when they were fairly well represented already.
The only hope IM has from a business perspective with this strategy is if the system is so opaque they don’t turn off their frequent customers who might otherwise realize their odds got a lot worse.
I think there is a bias against men in this dataset. These races were all Nice qualifying men’s races and I know that many fast, regular Kona-qualifying men have been sitting out of Nice qualifying events.
As suggested by Tribike53, it would be interesting to see this analysis on all of the 2019 Kona qualifying events, although racing demographics have also changed since then.
1 Like
I think what you are actually seeing is the depth of field are worlds is relatively uniform and dense at the front of all age group, but the depth of field in the 40-59 at least at Muskoka reflects some of the top athletes in Ontario. It’s the only MDot race in Ontario, and 1/3 of Canada lives in Ontario and another 1/5th of Canada lives nearby in Quebec, so we end up with several of the top 40-59s nationally at this race. I know many of the names and they have been pointy end for multiple years and perform as they should. A lot of these guys were 4 hrs flat to 4:10 in their prime. My best times were 4:14-4:22 range all the way to 45 years old. I am no where in that category age adjusted now (I was 4:36 age adjusted), but damn right some of the guys are certainly 4:0x in their prime anyway and their performances on the day reflect that, and if they had access to today’s tech 20 years ago, they would likely be sub 4 in line with their age adjustments.
So would the 4:11 guy who ended up 8th age adjusted get beaten by the 1-7th guys. The second age adusted guy was Mike Greenberg, who was an old training partner. Mike had finished top ten at Olympic tri nationals back in the say (around 1995) before he set off on his finance career (as things roll). I don’t recall Mike racing half IM back then, but if he did back then on today’s tech, he would definitely be a 3:59 athlete.
I can’t comment about the athletes in Sweden, but I know approximately half the athletes in the age adjust top 50 at Muskoka and these guys all did those types of times in their youth, so eyeballing it, Ironman is getting this right.
2 Likes
Yes, things have been “different” since 2019, and perhaps they’ve been different in different ways for different age groups. If they’ve been different in the same way for all age groups, the 2025 data would still be mostly representative, but we can’t trust that.
Another pre of 2019 is that the actual slot numbers can be used to compare old/new, as they weren’t split M/F.
Third reason to examine 2019 is that I very politely declined a slot on roll-down in Copenhagen! I was only there to do recon for my long term plan to qualify in 2020. Little did I know!
Anyone have a list of qualifying races for Kona 2019?
Looking at a single race will tell us little about how the two models differ because of great variability between races.