The season is well underway already, so I suppose that everyone who wanted to post some kind of prediction has been able to do so. Once again, I collected all of them and merged them into what I now call "Aggregate Prediction" per division. Unlike last year, there actually is one for PT as well in 2023, where predictions have gone up from 1 to 5. It's still the lowest number of any division, which is a bit sad imho. And I thought some of the managers who just went up - at least half of which I consider being pretty active managers - would come up with some preview, but no.
On the other hand, 9 predictions for PCT and 12 for CT give us a really great base to work with
Before diving into the analysis, let's recall how I did this aggregate prediction:
- Find the median predicted rank of every team
- Find the average predicted rank of every team
- Find the range of prediction (i.e. difference between mininum and maximum predicted rank)
- Order first by median, then by average, then by range
The sorting by median values minimizes the effect of outliers. The average is taken as a tie-breaker in case of equal median values. And finally, we'll take the range as an "ultimate" tie-breaker, as I consider a more narrow range to be potentially more precise. However, that final criterion didn't need to be applied, as the first tie-breaker was sufficient in all cases.
In case of questions or ideas to further improve this aggregate prediction, don't hesitate to hit me up, please!
And no, without further ado, let's take a look at the
2023 PT Aggregate Prediction
Number of predictions taken into account: 5 You can find the predictions in the spoiler
Just like last year, if a manager submitted two predictions, I considered both of them, so he can also find out which one of his models is more accurate In the case of PT, MacC has two predictions, the first one without, the second one with a "TTT Factor".
At the top, we're expecting a really close race between Evonik - who only finished 13th last year - and Puma, who ended up 3rd. Evonik gets 4/5 podium predictions, whereas Puma are the only team to get 1st in two predictions. Evonik win that "German" duel (I know they prefer being called Latvians), because of a slightly better average value.
There's also a duel for 3rd place coming in according to the predictions, with King Power and EA Vesuvio both having a median value of 5 - and Polar not far off with 6. Fun fact: these three teams are the only ones getting 5/5 Top 10 predictions... King Power win the tie break vs. EA, and also are the only other team besides the Top 2 getting more than one podium prediction.
We then have 5 more teams likely battling for the other Top 10 spots, with Tinkoff (who even get a win prediction), Gazelle, Moser, Lidl and ELCO. For Gazelle and Moser, we have the highest range between best and lowest prediction - from podium to (second) last...
On the relegation spots, we almost exclusively have newly promoted teams - with one notable exception: ISA are predicted to finish dead last. Pretty rough for a team that finished inside the Top 10 last year, and only had one notable departure - but on one hand they're more dependent on Ewan than ever before, and on the other hand they've always positivley surprised people in the past, so they're by no means doomed yet.
Cedevita, Carlsberg, Amaysim and Minions take the remaining relegation spots - with Rabobank predicted to only just stay up, like last year. Hence, the predictors think that among the promoted teams, Binance and ZARA should have the best chances to stay up - with the former even having a shot at a Top 10 finish.
Compared to last year's final rankings, the biggest losers besides ISA could be MOL - despite signing last year's individual rankigns winner, Taylor Phinney. It remains to be seen if the American can indeed outweigh the loss of the likes of Tenorio, Kwiatkowski, Galta, Inkelaar and Theuns - the predictors rather expect them to drop quite a few spots.
Another pretty big loser compared to 2022 could be Aker - MOT; it looks like people think the Anderberg and Amezaway trainings are worth less points than the loss of the entire TT department.
As seen before, it's the other way round for Evonik, who could jump all the way up from 13th to winning the PT crown. No other team is expected to make as big of a jump up, but Polar, EA, Moser or Lidl could make some notable gains as well.
Well, we'll see how it all pans out - I'll hopefully be able to post 2-3 updates during the season, but don't take it for granted.
I was hoping someone would do this Very interesting to see these big ranges. kb's system seems to be responsible for most of them, looking at Evonik, Puma or Moser for example. But that doesn't mean that has to be wrong (I mean, Puma 17th will very likely be wrong, but you get the point ). An interesting thing to check in on as the season progresses.
Fabianski wrote:
Unlike last year, there actually is one for PT as well in 2023, where predictions have gone up from 1 to 5. It's still the lowest number of any division, which is a bit sad imho. And I thought some of the managers who just went up - at least half of which I consider being pretty active managers - would come up with some preview, but no.
I filled out Cunego's PT survey. I feel like I only understand PCT at this point, and I spent all the time I had on that preview. Which most managers didn't comment on.
I've never been a fan of the mathematical predictions. Maybe I just don't understand them. Although the aggregate which puts me 2nd last is probably pretty accurate
A team being predicted 1st with an average position of 4.8 shows how much uncertainty is near the front and highlights the lack of a proper title frontrunner.
cunego59 wrote:
Very interesting to see these big ranges. kb's system seems to be responsible for most of them, looking at Evonik, Puma or Moser for example.
Indeed, kb's prediction is pretty different to most others. His system values depth way more than top-level riders, and I think that at least for PT with some really, really strong leaders this isn't very accurate. For example, it would mean that the 11th best climber should still score 80% of the top rider's points - last year, only Herklotz came close to Phinney, with all others already over 25% off. Even worse for the cobbles, where only Wisniowski came close to Pedersen - and Summerhill as #3 was almost 40% off already... kb's system doesn't reflect those huge differences at the top level - but it's still an interesting prediction, looking forward to see how accurate it will be!
redordead wrote:
I filled out Cunego's PT survey. I feel like I only understand PCT at this point, and I spent all the time I had on that preview. Which most managers didn't comment on.
I've never been a fan of the mathematical predictions. Maybe I just don't understand them. Although the aggregate which puts me 2nd last is probably pretty accurate
Oops, I thought I had already commented on your PCT preview after getting back from vacation, but apparently I just had the intention to... Will do. And I fully understand the frustration of putting a ton of work into something and getting only a couple of reactions.
Maybe next year you'll understand PT well enough to do a preview as well, after you hopefully stay up
Personally, I do like both kinds of predictions, but both obviously have their flaws. But as I haven't been able to come up with a decent mathematical approach yet, I rather did "qualitative" predictions, where it's also easier to consider "top rider bonuses" or the calendar. The aggregate one is obviously mathematical, though.
And it's pretty easy to understand I think, with median and average being straightforward methods. Median just means the "middle" value of all predictions; for example, predictions for Cedevita were 9-22-21-19-17, so the middle value was 19. I then sorted all teams by their median prediction, taking the average as a tie-breaker (which was 17.6 for you, but with no importance given that no other team had 19 as a median value).
Median basically means that half the predictors see you better, and the other half see you lower than this value. In case there's an even number of predictions (which is only the case for CT), you take the average of the two middle values. The more predictions there are, the clearer the picture should be - I suppose that with 5 predictions the aggregate won't be very accurate, though.
And I hope you'll finish higher than 21st, of course
knockout wrote:
A team being predicted 1st with an average position of 4.8 shows how much uncertainty is near the front and highlights the lack of a proper title frontrunner.
With such a low number of predictions, one outlier has a big impact on the average - which is why yours is "only" 4.8 - and this is the reason why I rather sort by median prediction. If you look at the details, your team was predicted 1-2-2-3 - and then one significant outlier with 16th, only just safety, and worse than last year. As cunego said, this also is the reason why the range of predictions is huge for most teams, as most have one big outlier. But you have a median prediction of 2nd place, just like Puma - with you being 0.2 better in the average prediction.
With two teams getting a median value of 2, and the next value being 5, I do think there's quite a good consensus on the Top 2 teams - but not on a potential title frontrunner indeed.
I'm not sure I'm happy to be called out that often for the outlier results of my prediction(s), but to be fair it only supports my initial claim that I have no clue whatsoever regarding the PT And if my predictions just serve as a kind of control group I'm totally fine with it because it strengthens the statistical significance of the other (and probably more sophisticated) predictions, something at least Fab will be happy to figure out
Apart from that, thanks to everyone who came up with a prediction, I've read all of them with interest, though I don't feel experienced enough yet to really commentate on them (especially for PT and PCT, as I hardly understand the CT).
Unlike in previous years, PCT actually doesn't have the highest number of predictions this time - but 9 is still a pretty solid base to build up upon.
Like in PT, prediction ranges for most teams are rather high, with more than half of them being above 10. And for those with the really small ranges, it's not a good sign - we'll come to that later. And yeah, once again it's kb's model that most often has either the MIN or the MAX prediction of a team - which isn't necessarily a bad thing, only the course of the season will show us who made accurate predictions.
However, for the predicted winners of the division - Los Pollos - it's Croatia who's responsible for that range of 11 spots. He's the only one not having them on the podium - and not even in the Top 10, actually. But with 5/9 predictions having them first, the median value is 1, and the average of 2.67 also is clearly the strongest of the class.
In terms of median, it's also pretty clear that Fastned should finish 2nd - although two predictions outside the Top 5 mean that their average prediction is worse than Lierse's, who have the same median of 4 as Assa Abloy and might be in a close fight for 3rd against them. One thing to note is that Lierse's predictions range from 2nd to 6th - which is the 3rd smallest range of all teams. Time for promotion then?
Xero are looking like pretty clear promotion contenders as well, ending up in the Top 5 in all but two predictions (including mine, which therefore should be ignored I guess ). As expected, kb's model doesn't like this very, very top-heavy team and even has them relegating - with predictions ranging from winning the division to relegating, it looks like it really depends on how well very few riders perform, with one failure potentially meaning a drop of several spots.
Sony should be the final team having some realistic chances of promoting, as they do so in 4/9 predictions, with their median 6th place meaning they'd narrowly miss out. But with a prediction range of only 5, they should be locks for a Top 10 spot.
Then we have a jump in the median value from 6 to 8, with Jura being next, followed by three teams with a 9th place median - UBS, Popo4Ever and Tryg. But while Jura and UBS have two predictions resulting in promotion, P4E has none - and Tryg have one. The two latter teams however will clearly get a head-start over the two Swiss ones after TDU.
The fight for the Top 10 spots should also include Indosat, DuckDuckGo, Lampre, Bralirwa and Kraftwerk, with median values from 10 to 13. And although Kraftwerk's median prediction is 13th, that's only enough for 15th place in the aggregate prediction - showing just how stacked the division is.
It's probably exaggerated to say that the fight against relegation starts with Colombini - but if one of their two top-leaders fails, it could still be the case. Zalgiris and McCormick are next, and then we have another small gap to Zwift.
And it's probably safe to say that the Americans don't look too safe. They "only" relegate in 3/9 predictions, but end up between 17th and 19th in 5 others - things have to go pretty well for them or they'll indeed be in danger.
Sauber then have 5/9 relegating predictions, Trans have 7/9, Spark 8/9 - and it's a 9/9 for both Genii and Glanbia. The last two teams are those I previously mentioned - those with the smallest range of predictions. From 21st to 24th for Glanbia, and even 22nd to 24th for Genii - the predictors don't seem to fancy these two teams' chances too much.
It's always difficult to compare PCT predictions to the previous year's rankings, because half of the division are new teams. What's interesting to see is that only two of the relegating PT teams seem to have retained enough quality to be clear promotion favorites with LPH and Fastned. UBS at least look like outside candidates, and whereas DuckDuckGo get a midtable prediction, Zwift might even be at risk of back-to-back relegations. Quite a wide range...
Of the promoting teams, Jura and Tryg should have the best chances of staying up - or even getting a Top 10 finish. Colombini and McCormick might be able to stay up, whereas chances for Glanbia, Genii and Spark don't look too high. We'll see...
Among the established teams, Lierse, Assa Abloy and Xero are expected to make the biggest progress compared to 2022, with all of them being promotion contenders. Philips could move up a couple of spots as well - and so should Indosat, from the region at risk to the Top 10 neighbourhood. Popo4Ever are also expected to be back inside the Top 10 after a really unlucky year - if TDU means that they now get all their luck they didn't have last year, they could end up even higher...
A couple of teams are even expected to slightly regress, with Lampre and Kraftwerk both missing out on the Top 10, Bralirwa dropping a few spots as well - and Zalgiris, Sauber and Trans being dangerously close to the relegation spots or even in real danger.
We'll find out if that aggregate view is more or less correct - or if an individual predictor stands far above the rest. But first, we'll take a look at CT soon, with the record number of 12 predictions included - stay tuned!
Thanks for doing this. It does seem like 3 mini divisions with Sony and up fighting for the automatic promotions spots Colombini and down fighting relegation and everyone else to varying degrees in between.
Hope you are right about Lierse, feels like Ollfardh is due.
Comforting that only 1 prediction has us relegating but it is certainly well within the bounds of possibility. Hopefully it doesn't come down to us versus our fellow Americans at Zwift. Although I guess better that than us both going down. The US riders in the peloton might need to plan on tightening their belts next year if things go really wrong and all 3 heavily US focused teams go down.
To be honest, finishing 6th would be worse than 8th
This reminds me of our first ever CT season where Jandal and me were predicted to be battling for the last promotion spot(s). Hopefully both of us can emulate that season at worst
The pressure is on! It feels a lot of teams can promote though, there's always a few overperformers and 1-2 overlooked teams, so with Bonifazio still in the team we might be missing out again though. Thanks for doing this!
12 predictions is a lot of data to work with - I love it But will it make the aggregate prediction more accurate? We'll find out later in the year - for now, let's look at what these predictions actually say!
There are two likely title contenders according to the predictions - with Tafjord getting even better numbers than Gjensidige. Tafjord have 12/12 podium predictions, thereof half of them having them as division winners. Six 1st place predictions and the 7th one being a 2nd place prediction means their median value is 1.5 - the average of below 2 shows they're the main favorites for the CT crown.
Gjensidige aren't that far off, though, getting 4 win predictions, and 6 more podium predictions - with 4th and 7th being the only non-podium ones. If everything goes according to expectations, these two teams will easily promote and reinforce the Scandinavian fraction in PCT next year.
Podium Ambition probably wouldn't be unhappy if their journey was as lonesome as the predictions suggest - clearly behind the Top 2, clearly ahead of the rest. Which means 3rd place and easy promotion - but can they really do it? The average prediction of 5.17 suggests that not everyone views them as a favorite for promotion - and while they even get one win prediction, 5 participants don't see them promoting. But in terms of median, their advantage over the rest is still noteable.
And who should be best of the rest? Well, that's rather hard to say. Looking at the median, 6 teams are within 1.5 points (with the average difference being only slightly larger), so basically any of them could end up on an automatic promotion spot. The newbies from JEWA TIROL get the best values, slightly ahead of Crabbe, who are looking to bounce back after last year's relegation. Stela-Vita, Strava, Würth and Ekoi are really close, though. It looks like a great fight for 4th to 9th!
So, who takes the final Top 10 spot? Air New Zealand seem to have some margin over the second half of the table - but not an overly big one. Caja Rural, Hilcona, Bordeaux, Manada, Simba, Mercatone and the Turtles aren't that far away from each other - with Simba (2x) and the Turtles (1x) even getting some promotion predictions! So we'll surely have a close race for the Top 15 spots as well.
And then there are the bottom 3 teams. Especially Saeco with predictions ranging from 16th to 20th look like a lock for the backmarker fraction - but Babymetal with 9/12 20th place predictions also look more likely to end up very low, despite that one outlier seeing them finishing 11th. It's a similar story for Adidas, who get 10/12 predictions from 17th to 20th. These three new teams are looking to have a rough season ahead - but looking at the number of developing riders in the Babymetal and Saeco rosters suggests they didn't choose an instant-win strategy anyway.
Like in PCT, we can see that last year's relegating teams aren't necessarily promotion contenders the year after, some of them rather going for some kind of rebuild. Gjensidige should go up if they didn't get their planning completely wrong. Crabbe and Strava are serious promotion contenders according to the aggregate predictions - whereas Manada will have to fight hard to even finish inside the Top 15.
Among the new teams, JEWA is looking strongest, thanks to their strong sprinter department plus Berhane. They've been able to confirm this so far (to my big surprise in what concerns Degenkolb, to be honest) - we'll see if they can continue on their high. Stela-Vita should be a strong contender for the title of "Best New Team", though.
The other 7 teams are mainly fighting for (or against) the bottom spots, with only Hilcona beating some established teams according to the aggregate prediction.
When speaking of established teams, Tafjord and Podium Ambition should be the strongest of them. Whereas promotion is long-awaited by Tafjord, Podium Ambition more or less deliberately dropped down to the very end of the rankings with their shift of focus, which could now start to come to fruition.
Würth then is the next non-new non-relegating team on the list, making a pretty big jump compared to last year's rankings as well. Ekoi might also move up a bit, whereas especially Caja Rural would be disappointed to miss out on the Top 10, after missing out on disband promotion by one spot last year. Air New Zealand and Bordeaux could both finish slighty higher than last year, without making gains like Podium Ambition or Würth, though.
And that's it for now. As said in the PT prediction, I hope to post 2-3 intermediate updates during the year, and then obviously look at the final standings and at who are our experts and who has some room for improvement in their predictions.
I'll give all of the people who predicted us to promote or even challenge for the title the benefit of the doubt and say you were over-optimistic instead of just stupid, since you were making those judgments before we saw how awfully Houle is being treated by PCM now
24/02/21 - kandesbunzler said “I don't drink famous people."
15/08/22 - SotD said "Your [jandal's] humour is overrated"
11/06/24 - knockout said "Winning is fine I guess. Truth be told this felt completely unimportant." [ICL] Santos-Euskadi | [PT] Xero Racing
Yes, I know it's late. Really late. The 2024 season is already about to start, so what's the point of posting the outcomes of the 2023 predictions?
Well, maybe it gives you some hints of which 2024 predictors you can trust, and which ones you can't. And also, I just didn't want to leave this one unfinished, despite not givin any intermediate updates. So, let's go straight to the final results:
Team
Final
Predicted
Difference
Aker - MOT
1
11
-10
EA Vesuvio
2
4
-2
Evonik - ELKO
3
1
+2
Team Puma - SAP
4
2
+2
Polar
5
5
0
Gazelle
6
7
-1
Tinkoff - La Datcha Team
7
6
+1
Cedevita
8
21
-13
King Power
9
3
+6
MOL Cycling Team
10
14
-4
cycleYorkshire
11
13
-2
Lidl Cycling
12
9
+3
Carlsberg - Danske Bank
13
20
-7
Moser - Sygic
14
8
+6
Grieg-Maersk
15
16
-1
Rabobank
16
17
-1
ZARA - Irizar
17
15
+2
ELCO - ABEA
18
10
+8
Amaysim Australia.com
19
19
0
ISA - Hexacta
20
22
-2
Minions
21
18
+3
Binance
22
12
+10
Whereas 4/5 of the Top 5 teams were correctly picked, the community definitely didn't see Aker in there - let alone winning the title. Credits to MacC, though, who was the only one posting a prediction that had them in the Top 5! But generally speaking, Aker is the shared 2nd biggest "outlier", so their fantastic season was pretty surprising for most.
EA Vesuvio were correctly predicted to finish 2nd by kandesbunzler, with the aggregate prediction having them in 4th. Not awful. Similar for Evonik, who were seen as the top title contenders, finishing slightly lower but still on the podium. MacC was again the only one seeing this coming - however, he had two predictions, having them in 1st place in the other one ^^
The Pumas also rather surprisingly finished outside the podium spot (which I got right, blame me, cio), but like Polar still got a Top 5. The latter being one out of just two correctly predicted teams by the aggregate.
Gazelle and Tinkoff swapped their spots compared to the prediction - but the really, really big surprise were Cedevita, finishing 13 spots higher than predicted! Credits once again to kb, who had them in 9th, just 1 spot off the final outcome. No-one else even had Cedevita in the Top 15 - apologies for underestimating your fantastic manager skills, red!
We round off the Top 10 with King Power and MOL - the former performing far below expectations (in particular Sosa I guess), the latter quite a bit higher. But well, you should never underestimate Phinney, even after a decline...
The only other team finishing more than 5 spots above expectations are Carlsberg, who instead of relegating just easily stayed up and even got a Top 15. This success is obviously closely linked to a fantastic season by Per - we'll see if he can replicate that.
While there were 3 teams with a -5 or better, we had 4 teams with +5 or worse. King Power were the first one, but Moser did just as bad. The lack of depth, plus another GT that Kudus couldn't quite win, plus Gaviria showing some mediocre performances have them down in 14th instead of a safe Top 10.
It's even far worse for ELCO - ABEA, though, who after a strong 2022 season convinced most predictors to have them in the Top 10! Only MacC in his two previews saw the dramatic fall coming (17th and 19th) - 18th place means a rather surprising relegation. Coquard was good, but not as great as in 2022, and both Carapaz and Koretzky were bad. Just not quite enough to stay up, but we'll hopefully have them back up here in 2025.
Amaysim were predicted to narrowly miss out on safety, and sadly for them the aggregate prediction was right on them. They couldn't replicate the dominant 2022 performances, with McCarthy in particular being mostly disappointing. Haig being rock solid wasn't quite enough to save them, which 3/5 predictions saw coming (all of them having them in 18th).
ISA managed to not get the red lantern in the end, but didn't have a realistic chance of staying up in the end. Ewan would have needed to be a lot better than he was, and Vakoc's season wasn't really good, either.
The Minions on the other hand lost 3 spots compared to the prediction - which also saw them going down, though. Again, just not enough scoring by their leaders, Kinoshita, Cattaneo or Ahlstrand.
But our biggest losers finish dead last. Biggest losers because they end up 10 spots lower than the aggregate prediction - I hope finishing last wasn't the decisive factor for Binance to sadly disband after a horrible year. Getting declining leaders was always a risk, and getting both Sagan and Kwiatkowski maybe even more so - but all in all they just didn't perform at all. By the way, cunego's prediction saw their downfall coming, having them in 20th place...
In terms of numbers, the aggregate prediction is exactly as good as last year's (where Croatia was the only PT predictor), meaning a deviation of 3.9 spots per team on average.
Which is pretty good - but one individual prediction was slightly better. I don't want to boast my PT knowledge, as I have none, but for some weird reason it's actually my own prediction that came out on top, having an average deviation of 3.5. I guess with 15 teams predicted within 3 spots of their final rank, including 6 exact predictions, I got pretty lucky indeed.
In 2nd place, being exactly as precise as the aggregate prediciton, is cunego's prediction, based on the survey he did, hence including many other managers' work. Good job!
MacC also did a pretty good job, with his two predictions being 4.4 (normal) and 4.6 (with TTT factor) spots off the outcome on average.
On the other hand, kandesbunzler with a 7.5 spots deviation may need to fine-tune his model for PT. While he definitely did have some good predictions, he also had 7 teams with 10 or more spots off prediction, while all others had 4 at most. Not to blame him, I'm grateful for everyone doing a preview - that's just the raw numbers which sometimes are really harsh.
PCT and CT will come, hopefully before the season starts (maybe one of them still tonight) - and yes, I have planned to do the same thing again for 2024. Iirc, seancoll also mentioned he'd do some "meta-prediction", looking forward to what he'll come up with
There were 9 predictions for PCT, so quite something to work with. But how good was the outcome? Let's take a look:
Team
Final
Predicted
Difference
Los Pollos Hermanos
1
1
0
Jura GIANTS
2
7
-5
Xero Racing
3
5
-2
Fastned
4
2
+2
Assa Abloy
5
4
+1
Indosat Ooredoo
6
11
-5
Bralirwa - Stevens
7
14
-7
Sony - Force India
8
6
+2
Lierse SK - Pizza Ullo PCTeam
9
3
+6
Team UBS - Tissot
10
8
+2
Lampre - Pinarello
11
13
-2
Team Popo4Ever p/b Morshynska
12
9
+3
McCormick Pro Cycling
13
18
-5
DuckDuckGo - Everesting
14
12
+2
Tryg - Gobik
15
10
+5
Sauber Petronas Racing
16
20
-4
Kraftwerk Man Machine
17
15
+2
DK Zalgiris
18
17
+1
Colombini Cycling
19
16
+3
Zwift Pro Cycling
20
19
+1
Glanbia
21
24
-3
Trans Looney Tunes
22
21
+1
Genii Hyundai N Cycling
23
23
0
Spark-BNZ Racing
24
22
+2
What instantly sticks out is the lack of big numbers. Zero 10+ differences, with the highest ones being 7 up and 6 down! It looks like PCT was easier to predict than PT, despite having two additional teams...
Unlike in PT, the winner was absolutely not surprising here. 5/9 predictions had Los Pollos on top, and 8/9 had them in the Top 3. They had good leaders everywhere, and they delivered. Not much more to say, just a really, really dominant season.
Whereas - like in PT - the aggregate prediction has 4/5 Top 5 teams right, Jura is the outlier. And finishing 5 spots higher than predicted is actually the shared 2nd highest value in that direction (with two other teams doing the same). So maybe a bit of a surprise, but not *that* surprising. They made their depth - in particular TTT-wise - work, and with Theuns had their strongest leader ever. Easy promotion - which was only expected by 2 participants (not including myself, by the way).
Xero, Fastned and Assa all were expected to promote, and they did. With a range of +/- 2, not much else to say. They just did what they were expected to do.
Which isn't the case for the next two teams - in a positive fashion, though. Indosat match Jura's -5 - probably mostly thanks to Yates having a stellar season (which was expected anyway for Dunbar) and Abdul Halil doing really well given his overall skillset.
Bralirwa were the most surprising team in the end with a -7 outcome - and for most of the year it even looked like they could end up in the promotion range! Once again, it wasn't to be, but still a much stronger than expected (by the aggregate prediction) for them.
Sony, Lierse and UBS were expected to get a Top 10 finish, and they did. Sony and UBS slightly lower than predicted - but Lierse actually are the biggest losers compared to the prediction. Instead of ending up on the podium. And this despite Bonifazio actually doing OK for once - but Novak, De Plus and maybe even Gamper were a bit below expectations. And Oomen was great, but not as dominant as he probably was expected to be. So it will be another year of trying for Lierse.
Lampre almost did make the Top 10, whereas Popo4Ever - mostly due to an underperforming Padun - unexpectedly missed out. McCormick were another positive surprise with -5, which could have been even better with a scoring-optimized planning for McNulty. But he gave us a great show in ToA for sure. Otherwise, UU probably just planned his riders really well, and in particular the puncheurs did a good job (well, Bagioli and Narvaez).
DuckDuckGo finished slightly below expectations, after it looked even much worse for a while. Tryg were the 2nd most disappointing team, most of which comes down to Démare being good, but definitely not great. On the other hand, Sauber never really were in relegation danger for once, rather unexpectedly. Sure, 16th isn't overly comfortable, but it's enough. Their amazing uphill depth made it work once again, making up for Tom David's pretty awful season.
Kraftwerk, Zalgiris and Colombini all were predicted to stay slightly above the relegation zone - which they did, even though it was a bit closer than expected for the latter.
Zwift ended up just one spot lower than predicted, which usually isn't a big deal - when it's the spot deciding between safety and relegation, it is a big deal, though. With 8/9 predictions having them between 17th and 21st, relegation isn't overly surprising, though - although it's obviously still the worst case scenario for this incredibly talented team.
The other four relegating teams were correctly predicted by the aggregate - and by 7/9 predictors, with the others having Trans in the safety zone (and one of them even Spark). Glanbia didn't finish dead last, which is positive, but they needed a -5, not just a -3 to stay up.
So, where do we end up on average? Well, I'll tell you that the numbers are pretty insane: The aggregate prediction has an average deviation of just 2.8 spots per team! We had 4.7 last year, so that's a massive improvement! We had even better numbers for CT last year - but with 18 teams, not 24... And surely this can't be beaten by an individual prediction?!?
Sure, it can. Not by one, not by two, but by three individual predictions! And we have two ex-aequo winners: knockout's lazy prediction and cunego's survey (posted in the same thread) both end up with an average deviation of only 2.5 spots! Kudos and big congrats to these two!
Only slightly less accurate was redordead's 2023 PCT Prediction, ending up with a 2.7 average. Well done by all of them, beating the common wisdom!
But actually, 8/9 predictions were really good (all of them more accurate than last year's top prediction), with MacC's 2nd prediction (including TTT factor) being 2.9 spots off on average (the one without was 3.7 away - the 2nd worst for PCT!). Ulrich Ulriksen and myself were 3.1 spots off, and Croatia14 was 3.2 spots away on average.
Just like in PT, kandesbunzler's model didn't work too well - but with 5.6 still clearly better than one division up.
Generally speaking, 5/9 predictors ended up without a 10+ outlier (just like the aggregate prediction). cunego got 6 teams right, but also had 6 5+ outliers - whereas knockout only had 4 of the latter, allowing him to match cunego's overall score.
All in all, a fantastic job by our predictors - up to you to tell me why the 2023 PCT class was so much more predictable than the 2022 one...
Love these writeups, always cool to see how well your predictions are doing in hindsight.
Very accurate pct ones across the board. While doing the preview, i thought that there were some clearly dividable tiers which might have helped with having fewer outliers.
Thanks for the wrap up. My impressions was people thought PCT was harder to predict, but I guess this doesn't support that, although it could be a one season anomaly. If it is true then I guess PCT managers should be bracing ourselves for the 2024 aggregate prediction. But Aker gives all the PT teams hope.
It looks like the lower the division, the more attractive it is for doing predictions. 5 for PT, 9 for PCT - and 12 for CT. Here's the outcome of the aggregate prediction - completely excluding Mercatone Uno from both predictions and rankings, by the way:
Team
Final
Predicted
Difference
Crabbe-CC Chevigny
1
5
-4
JEWA TIROL
2
4
-2
Stela-Vita
3
6
-3
Podium Ambition
4
3
+1
Tafjord Kraft
5
1
+4
Ekoi - Le Creuset
6
9
-3
Strava
7
7
0
Manada Coyote
8
14
-6
Gjensidige Pro Cycling Team
9
2
+7
SEE Turtles
10
16
-6
Bordeaux Métropole - Euskotren
11
13
-2
Team Würth MODYF
12
8
+4
Simba Cement - Tanga Fresh
13
15
-2
Hilcona Racing Team
14
12
+2
Saeco
15
17
-2
Babymetal Cycling
16
19
-3
Caja Rural Cycling Team
17
11
+6
Adidas - Eurocash
18
18
0
Air New Zealand-Alfa Romeo
19
10
+9
As in the other two divisions, 4/5 Top 5 teams were correctly predicted, although the order might be slightly surprising.
While it's true that Crabbe didn't have a lot of depth, Chiarello more than made up for this. In a pretty weak GC racer field, he really dominated most of his races and scored 51% of his team's points. He could've scored 500 points less and Crabbe would still be Top 5, so it was a convincing season by the whole team nonetheless.
JEWA Tirol's recipe of success consisted in getting two declining but still very strong leaders, a climber and a sprinter - and both delivered far better than I expected, to be honest (I'll admit I had them in 15th, expecting nothing from Berhane and Degenkolb).
Stela-Vita was another team the majority predicted to finish rather high up - Top 10 in all but 2 predictions - and where I fully underestimated the big Xhamberaj factor. 2nd in individual rankings was a fantastic performance and the #1 reason for the Albanians to promote in their debut season, just like JEWA.
Podium Ambition are the first team ending up slightly below expectations - but they'll be more than happy with 4th place, as things didn't look like automatic promotion at all after a rocky start. And Tafjord, going into the season as easy top favourites, struggled quite a lot and only just secured automatic promotion in the end.
Ekoi had a slightly better than expected year as well, which in the end resulted in a disband promotion. A guy like APP is just worth a ton in CT, and - speaking from experience - can also be an overachiever in PCT. Add an insane Gall performance in ToA, and that's where you get.
Strava finished exactly where they were expeced to - as one out of just two CT teams. Great cobble results, mainly by Gaday, some good performances in the hills - but not that much more. Still, 7th would have been enough for a disband promotion - sadly, they did themselves leave the game, too.
Which leads us to Manada as the final team earning a disband promotion spot - and here we can say that this is pretty surprising! Finishing 6 spots higher than predicted is the best outcome of a CT team, and Marin's great sprint performances throughout the year definitely are the main reason for that. KHJ and Lemus Davila also ended up in the Top 20 - whereas the team mostly wasn't spectacular, they were consistent scorers and (unexpectedly) secured another year in PCT!
From one of the biggest winners to probably the major losers of the year - Gjensidige. Narrowly behind Tafjord in the predictions - while even getting 4 first-place predictions - they just didn't perform all year long. Unexpected TTT issues, maybe some subpar planning - in the end, it's just 9th place and no promotion. Next try in 2024.
The Turtles did just as well as Manada in terms of final result vs. prediction, ending up 6 spots higher than expected. Once more, TTT depth really paid off in CT - and apart from sprinting, there was not much on top of that. A Top 10 was surely the maximum they could reach in 2023, we'll see what they can do this year...
We then have Bordeaux finishing a tad higher than the aggregate prediction suggested (gaining one spot on the final day of racing), whereas Würth had a rather mediocre season, ending up 4 spots lower than expected.
Simba, Hilcona and Saeco all finished more or less in the expected region, within 2 spots of the prediction, in one or the other direction. Babymetal did even 3 places better - predicted last, ending up 16th. So they did a pretty good job given their roster.
On the other hand, whereas Adidas were expected to finish really low (18th), the other two teams down there were massive disappointments. On one hand, we have Caja Rural, who finished 8th last year, missing out on a disband promotion spot by 3 points. This year, instead of improving, they were only predicted to finish in 11th - and did a whopping 6 spots lower. 17th place. Not good.
Even worse for Air New Zealand, taking the final spot in the rankings (while having a 10th place prediction). Their downfall wasn't as hefty as Caja's, from 14th to 19th - but something really went wrong. And that "something" is mainly called Nur Aiman Zariff, who on paper was one of CT's best sprinters - and scored just 37 points. It wouldn't be fair to blame just him - he'd have had to score 700 poins to get ANZ to the Top 10, which is unrealistic - but let's hope that Jajajasper found the right levers this offseason.
Even though there were just 19 teams in CT (not counting Mercatone Uno, as mentioned before), the aggregate prediction was slightly worse than in PCT, with an average deviation of 3.5 places per team. Still better than the PT one, though.
But one predictor showed us that individuals sometimes can do better than the group (although not as good as for PCT). Well, it's still the "group" that was better, in fact, as the winner is Ezeefreak's CT Community Survey, more precisely the overall prediction, being 3.2 spots off per team (with the summed-up terrain prediction coming 2nd to last with 4.7).
Two more predictors got really close to the aggregate's outcome, with Croatia14 averaging 3.6 spots of deviation per team, and redordead3.7.
Next are MacC, being 3.9 spots off with the TTT factor and 4.2 without. cunego's Survey was exactly 4 spots away per team, followed by Ulrich Ulriksen's contribution in the CT Community Survey, having 4.2 with the first and 4.3 with the second version.
Besides Ezee's 2nd guess, which was already mentioned, we have Laurens and myself left, both at 4.6.
And finally, kandesbunzler makes it 3/3 red lanterns - however, 5.3 is his best result accross the divisions.
That's it for the 2023 Aggregate Predictions.
It's not sure I'll ever do one again, as it has become clear by now that seancoll is doing in 2024 exactly what I previously did. And there's obviously no point in posting the same data twice.
I hope some of you enjoyed it!
@knockout
That definitely helped. In particular the (likely) relegating teams were far easier to predict than the year before, but as you said, there were multiple rather clearly defined tiers.
@Ulrich
We'll see if it was just this year. One possibility could also be that there were mostly PCT and CT managers participating (no PT manager posted a PT prediction last year), so these managers probably just have a better knowledge of how PCT and CT work, with PT being pretty different in several aspects.
In the case of PT, MacC has two predictions, the first one without, the second one with a "TTT Factor".
). An interesting thing to check in on as the season progresses.
