Last year I shared some predictions with the Skype group based purely upon stats. I won't pretend they were particularly good, but after tweaking the calculations slightly I thought I'd bring them back for this year and share them on the forum as well.
The teams are ranked based on an OVL value, which is calculated using the alternative OVLs I use for ranking riders in transfers. The top 20 riders (PT/PCT, top 17 in CT) on each team are weighted based on their expected contribution to their team's total based on last year's rankings. For example, if the 2nd highest scoring rider on a PT team contributed 15% of a team's points on average then the 2nd highest ranked rider on a PT team using my alternative OVL would have that value constitute 15% of their team's total.
This method also allows for the calculation of how much a team gained or lost compared to last season. It's a simple comparison of the OVL value their roster would have had last year with their roster this year, using the same formulas. Obviously promoting and relegating teams skew the results, but figured it would be interesting to show.
Also in the previews, ranks by terrain for individual riders are indicated by numbers in brackets: for example (1) means the best rider on a terrain by my formulas. Riders are either stage racers, puncheurs, TTers, cobblers or sprinters, though a few don't quite correspond to what would be expected.
While I won't do a proper in depth preview for each team, I'll run through each division preview and pick out some highlights - but please do ask questions if you have them! I'll post PCT and CT in the next few days, but first a look at PT:
First and foremost, there are three clear favourites in the PT according to my numbers: Puma, Evonik and Isostar. Perhaps it isn't too much of a surprise after their performance last season, but the same trio are expected to fight for the title. For Puma, Herklotz (2), Demare (2) and Ganna (4) are all top 5 on their terrain but it is their depth which gives them the edge over Evonik and their similar trio of Lopez (1), Eislers (3) and Taaramae (5) despite the latter's marked improvement from last season.
Isostar have an impressive double act of the top stage racer and sprinter: Phinney (1) and Gaviria (1) are joined by Lutsenko (4) as a superb top trio, but their weaker depth means the reigning champions have stayed still as their rivals improved. They still have a comfortable gap to Gazelle and Generali in 4th and 5th though, with both teams lacking star power. Madrazo (6) and Formolo (6) for the former and Dennis (5), Sicard (7), Debesay (7) and Oliveira (8) for the latter headline teams that will score well from depth but may find overall wins harder to come by. Generali have also weakened more than most compared to last season, suggesting their renewals strategy may have backfired.
In 6th and 7th we have two teams with little between them in Festina and Moser, though the latter has got substantially weaker with the loss of Bewley and Barguil. Festina have Lecuisinier (3) and Coquard (6) to score well for them, but the formulas are less keen on riders like Vlatos than their raw stats would indicate. Moser have Sagan (2) who would be top puncheur if not for MAL, but even with training Kudus (9) isn't quite at the superstar level yet if the numbers are right - in fact Hirt (12) isn't considered much weaker than him.
Close behind this duo follows the first promoted team in Farfetch, who were predicted to win the PCT comfortably using this approach last season - and did, to the shock of their manager. 8th seems a little high on paper, but with 11 riders in the top 20 on their respective terrains (though Paillot (6) the only in the top 10) maybe not. Similarly Grieg in 9th have a superb trio in Wisniowski (1), Pedersen (2) and Wurtz (3), but their depth isn't brilliant - and their focus on lower scoring terrains means another prediction which may not meet the eye test.
10th goes to Zwift narrowly ahead of Mapei. The former improved the least of the promoted teams due to the loss of Tenorio - though Cattaneo (1) is considered the best TTer for scoring potential, and Yates (7) should do well. The latter improved the most of the promoted teams, with Bewley (3), Ahlstrand (8) and Buchmann (10) all top ten on their terrains. Both are quite close to the massive cluster of teams that follows however!
Leading this cluster are King Power and Aegon. King Power are considered to have been weakened by their moves this season, with Mohoric (8) a downgrade on Lutsenko that isn't covered by Kasperkiewicz (13) and Gidich (16) maxing out (though categorisation may not help the latter). Aegon are considered to have improved by the most of any non-promoting team, with De Bie (3) still a top puncheur and Hepburn (9) considered to be an underrated signing.
Then come a trio of teams separated by a miniscule margin, but still clear of the drop. EA Vesuvio are very much a team in transition and that shows up here, though Wirtgen (2) and van der Lijke (7) should star. Huski Chocolate have a strong duo in Gerts (4) and Grosu (5), but Dombrowski (8) has stagnated as his rivals have improved. And Amaysim seem slightly worse off when replacing Bobridge with Summerhill (5), though Haig (10) and Morton (11) should perform well as a double act in the mountains.
That leaves four teams, and the model suggests that two of them will be close to the last few teams mentioned in the scrap for relegation. It expects that somehow Tinkoff will find a way to survive again by leaning on Yatsevich (7) and Senechal (10), just ahead of MOL. The newly promoted Hungarian team has two stars in Tenorio (4) and Theuns (6) but lacks quality depth, though in truth anyone from 12th - and maybe even 10th - down is likely to be at risk of the drop if luck goes against them.
Then in 19th and 20th we have two teams the model seems to hate. Like MOL, Xero have improved considerably after promotion but always faced an uphill battle - and some interesting transfer decisions didn't help them. Areruya (5) and Houle (9) are expected to be surprise stars, and should keep them ahead of ISA - Hexacta. Despite Degenkolb (4), the South American team are predicted to finish last even after improving their roster, though last season the model their predicted relegation and they stayed up, so maybe it'll be wrong once again?
Edited by jph27 on 12-10-2021 22:56
The PCT offers a markedly different picture to the PT. Not only is there a clear title favourite, but the average degree of improvement within the division is much greater (0.39 vs 0.09):
Aker are the comfortable favourites according to the model, comparing well in terms of the calculations to the Farfetch team that dominated in 2020 (which came in at 74.86). Kwiatkowski (2) is one of the division's stars, while Galta (7) and Lammertink (9) are supported well by another 5 riders considered to be top 20 on their terrains.
In second we have Red Bull Zalgiris, who the model consider to be the most improved non-promoting team - and by quite a large margin! Keizer (1) and Brandle (5) give them an incredible TT duo, and are supported by impressive depth. Polar take 3rd having shown a lesser degree of improvement, with Latour (5) and Bernal (6) highlighted as ones to watch.
4th and 5th sees our first tie, at least to 2 decimal places. Assa Abloy impressed last season and in Monsalve (2) have a top tier leader supported by ample depth, with Gonzalez Salas (10) and Breen (11) the pick of the bunch. Voyagin - Bird are ranked as the second best relegated team, though are more top heavy than many of their rivals - Kinoshita (1), Sh.Kureoda (8), Shikai (9) and Liu (9) all possess strong scoring potential.
A gap then opens up to 6th and cycleYorkshire, the relegated team that has been weakened the least compared to last season. Fraile (3), Durbridge (7) and Oliveira (8) are a great trio of TTers, but the British team are lacking a little firepower on other terrains. They still have a reasonable advantage over the fourth relegated team in UBS though, with Ponzi (7) and Bohli (10) picked out as the names to watch.
After the trio of relegated teams, we have our first promoted team in Carlsberg. Trentin (1) is considered the best cobbler in the division by the model, with Guldhammer (6) also getting a lot of love - though seen as a puncheur rather than a stage racer. Narrowly behind them follow Popo4Ever, as the decline of Pluchkin (1) fails to be offset by an improving Padun (22), though expectations will be high for a better season from Ackermann (7).
In 10th we find Lierse, who appear to have got marginally weaker by moving on from Gaviria. Oomen (4), Vanbilsen (9) and Senni (11) should keep them in the top 10 mix, but their weaker depth compared to similar teams may hurt them. They find themselves narrowly ahead of one of last year's surprise packages in Kraftwerk, a very depth focused team with Arndt (9) and Zabel (10) the key stars. This mini-pack is completed by Duolingo in 12th, who will have to hope that van der Sande (4) continues to perform in line with the model, while Gesink (19) declining definitely hurts their chances.
In 13th and 14th we see the second biggest gainer of the non-promoting teams, and the least improved promoting team: Binance and Philips. The former has a leading duo that the model likes in Fernandes (6) and Kennaugh (6), plus quality depth, while the latter may not be able to rely as much on Kelderman (11) but could have a surprise star in David (4). The Bobridge (4) led Project: Africa squad find themselves narrowly behind, locked in a tie with a Cedevita team seen as suffering from a lack of depth - Saber (3) is a surprise pick for a good cobbles scorer, while Beltran (5) is a known quantity at this point.
The teams continue to come in pairs as we reach 17th and 18th. In Wellens (3) and Skujins (3), Tryg arguably have the best leadership duo in the division, but they lack the support riders to make the most of their remaining racedays. Minions are another team hurt by a declining leader in Uran (18), which limits the benefit from adding EBH (8) and an improved Zepunkte (13) - the latter no friend of the formulas. Indosat in 19th find themselves somewhat further back, and appear to be the team most weakened compared to last season. Howard (3) and Dunbar (8) are a quality leadership duo, but poor depth is expected to hurt them.
Such are the margins in the PT according to the model that only as we reach 20th does our attention turn to the relegation battle. Sauber have Van Garderen (10) and not a lot else bar Ranaweera (25), but their depth is impressive and they'll hope to be relatively safe. On the other hand, Carrefour find themselves slightly closer to the drop zone despite the presence of Van Stayen (2) and Daniel (7), as their supporting riders are less well liked by the formulas.
Yet they remain clear of the team in 22nd, with Los Pollos Hermanos expected to relegate despite taking the crown of most improved team in the PCT this season. Zmorka (2), Manninen (5) and Spilak (10) may be the best leadership trio in the division, but their poor overall depth and generally lower ratings for sprinters and TTers leave them in a slightly surprising 22nd. Trailing narrowly in 23rd is Jura - Fiat, with the Swiss team having much the opposite problem - Quintana (13) and Masnada (16) headline some fine depth, but they lack top end firepower, with Moscon (22) and Reinhardt (22) unlikely to repeat their CT heroics.
That being said, the aforementioned teams are given a much better chance of survival than the three below them. Podium Ambition are another team with three solid leaders but very poor depth: Ewan (1), Cataford (4) and Altur (6) will need to be perfect to keep them up this season. Bralirwa added two quality leader in Stallaert (2) and Silvestre (8) but miss another big scorer to keep them safe, while Eurosport are expected to be cast adrift due to the model being less keen on Blythe (5) than many analyses, and the rest of their roster being seen as relatively weak.
Edited by jph27 on 12-10-2021 22:55
Interesting that you think my team actually improved - which is contrary to what SotD and - as far as i can gauge - most people seem to think.
2nd is a nice prediction, though it must be noted that Isostar is very close behind me while the methodology seems to be working against them since their team looks a lot more top heavy than the average team. This leaves riders such as Lutsenko or Gaviria with a too low impact on their jphOVR. Am i thinking wrong?
No shock your transfer value overrates us - though when I can trust you to be the predictor most high on my team every year and you only put us one spot above last... yeah it says it all Have to say I'm really surprised how much your formula has us as gained on last year - guess it's how ridiculously overrated you have maxed Areruya but I'm not entirely sure we've improved at all having lost Debesay and Schreurs, definitely not by such a big margin!
Enjoy the overview, obviously we probably think quite similar things about riders and stats so fun to see maybe the formula which might end up closest to how I see things subjectively, since I always snipe about formulas
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
knockout wrote:
Interesting that you think my team actually improved - which is contrary to what SotD and - as far as i can gauge - most people seem to think.
2nd is a nice prediction, though it must be noted that Isostar is very close behind me while the methodology seems to be working against them since their team looks a lot more top heavy than the average team. This leaves riders such as Lutsenko or Gaviria with a too low impact on their jphOVR. Am i thinking wrong?
The improvements to Eislers and Lopez (from maxing and training), plus the addition of Taaramae and Polanc more than cover off the losses, largely due to the weighting. The formulas prefer Polanc to Blythe and Eislers to Van Stayen now he's maxed, while Taaramae is seen as an upgrade over Skujins.
It's an interesting point regarding Isostar, and one I'd want to account for if developing the model further. Ideally I'd weight for the variety of different past performances so that having a top heavy team would get calculated slightly differently (probably by presenting team OVLs as a range as a result). But that requires more technical skill than I have to automate the process, and will take a long time in Excel. So maybe next year
For now, I can show the evolution of the scores for the top 3 which should add context. The green highlight shows the top ranked team after the first however many riders, the yellow the second ranked team (not accounting for the rest of PT, though shouldn't be much different I imagine). Isostar's fall from the lead is mainly due to the weakness of the lower end of their roster, whereas Puma have greater depth that kicks in outside the top 8/9 riders - ditto Evonik after 13/14 riders. The gap between Isostar and the others is greatest after the top 4 riders though (Per as their 4th), which suggests that a more flexible weighting system would definitely favour them.
Another prediction placing me in the top 8! Will not complain. Again, I don't expect to be that high up, but would be quite satisfied if if means I'm at least out of the relegation fight. I regret to say that Guldhammer won't be getting many puncheur race days though, so that might do me in.
Thanks for the prediction, great read and could quantitiative elements.
RIP Exxon Duke, David Veilleux, Double Feature, and Monster Energy
Ahhh, finally an accurate prediction, took a while to get here. I do however feel that Popo are greatly undervalued under your model, I got Aker and Popo as locks for promotion
I'd be happy if the prediction of staying up came true, I've really got to rely on some good depth scoring behind my leaders, but I have built my team with that in mind so hopefully it works out ok for us.
I can't really agree with where my team is ranked. I don't agree that Beltran is a known quantity given that he scored 500 pts less in 2020 than in 2019. Also you clearly didn't watch Saber in the cobble races last year
So given that my team has no depth I assume your version of the OVL overrates Beltran and Saber even more than roturn's OVL
Interesting how high you rate Senni, I definitely consider Novak to be the better pointscorer. Also a bit surprised you only rank Vanbilsen as 9th. But yeah, 10th is about where we should be. Bonifazio could make it 5 places higher or lower, but we'll see.
AbhishekLFC wrote:
A bit surprised and intrigued to find David number four in your cobbles list
He's got the "jandal has tried to sign him" clause which gives a 8x multiplier in jph's formulas
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
Thanks for the preview. I hope you're wrong, but when you say my leadership trio is one of the best trio's but I lack depth you're obviously not lying.
Yep, hard to argue with a Bottom 5 prediction for T3A. Correct we have zero depth and need our trio of leaders to be as close to perfect as possible.
I like you recognise AC as a Top5 TT'er, and hopefully we've planned him well to suit his strong all-round skills.
It is interesting to see how much depth plays into the prediction, and a boost if you have hilly/stage racers. Like with one or two other predictions some of those teams will take away points from each other, and i think the 2 or 3 of the bottom 5 will surprise as they are targeted different and less congested areas.
![imgur.com/KlJUqDx.jpg[/img[img]]https://imgur.com/yUygrQ.jpg](https://imgur.com/KlJUqDx.jpg[/img[img]]https://imgur.com/yUygrQ.jpg)
