A ramble through some of the things I found most interesting about the team selections for this year’s FSA-DS. With apologies, this is focused purely on the men’s side – time doesn’t permit a second deep dive yet. If you want to know about the drafts, Squeems has you covered. If you want to know about the ed’s draft… ah, I don’t want to talk about it. In a five-certainty draft with twenty two players, picking 6 and 39 was the worst possible slot, as far as I’m concerned.
The Importance of Being Earnest About Your Restricted Picks
Tricky at the top
So, after some chat yesterday, I went back and found the quote that got me thinking. This is from MOUA:
There is so much value in the lower ranks that my guess is most of us following this site will end up with very similar variations among 1-8 pointers. It’s the top five picks that will decide the contest: whether you picked Primrose, whether or not he has a decent season, which of the 26-pointers hits paydirt. I fear Remco won’t be much of a factor, a big Covid question mark over Sagz, and plenty of winning or losing choices available in those critical spots 3-6 on your lists.
Within the confines of what I can manage – I’m no kind of database magician, and I didn’t have all weekend to work on this – I’ve had a look at a “widely picked” team, and at the valuations among the cheaper riders. What I can’t do is look at “those following this site” as a subset of players – I’ve just had to take everyone who added a team.
First of all, yes, it does appear that there is a little more commonality among the most popular riders this year, but not to a particularly significant extent. You can see my workings here, if you like that sort of thing. You’ll see that in 2021 the top ten picks were made by an average of 40.3% of players each, compared to 39.5% last year. For the top 25, those numbers are 31.2% and 29.9%. As near to the same as makes no difference. What is more surprising is that the three biggest picks were huuuuge this year – 55.7% against 47.2%. Kuss, Kamna and McNulty are setting a new standard for popularity.
We do see more difference when it comes to the price of the most popular riders. Last year, two of the top three were 22-pointers (Evenpoel and van Aert), whereas this year they weigh in at one, two and three points. In 2021 an average top ten rider cost 7.20 and an average top 25 rider 5.96. In 2020, that was 8.00 and 7.20.
Now, you can argue that all riders are cheaper this year (true, basically) or that arbitrarily picking the top few riders is cherry-picking (sort of true, but I’m using fairly obvious cut-offs). You can also argue that, with a few outliers changing things, it doesn’t prove a great deal. That might well be right. I do think, though, that MOUA’s point is right. Go back to the 2016-2018 era of FSA-DS and you saw lots of riders priced higher, and finding hidden gems could really separate you from the pack. With more riders cheaper, over-performance from your 1-4 point riders is more commonly achieved, and it is getting your big riders right that is harder and more important than ever.
The Elites in Combination
Let’s look at the data another way. Assume you want to have a team with three restricted riders, one of whom is double restricted (i.e., to spend as much as possible on “elite” riders), your most popular trio would be Wout van Aert, Egan Bernal and March Hirshi (on, respectively, 287, 174 and 188 teams). Again, you can follow along with my workings if you’re that way inclined. You could also have a team with Evenpoel, Kristoff and Trentin (24, 9 and 9) but let’s leave that aside.
The most popular trio is, if my manipulation is right, shared by just nine teams. Last year, the most popular elite trio (Mathieu van der Poel, backed with Tadej Pogacar and Remco Evenpoel) was shared by 44 teams, I believe. Now, we know that because of Remco’s injury that trio wouldn’t have won you the league. However, you’d still have needed to pick well for the cheaper riders just to make the top ten of that elite rider subset. Get your big riders right this year, and you won’t have to worry so much, because your company will be a lot less numerous.
Another thing that is impressive about the 2021 returns is the spread of the riders picked. I’ve added a pie chart (oooh! aaah!) to show the egality, but among the double restricted riders, Remco Evenpoel is picked on 24 teams, by far the fewest, whilst only 97 teams elected not to pick at least one of the elite riders. With seven groups nicely distributed (the six double restricted, and no double restricted pick) you’d want to be in the right group to maximise your chances of success.
Team Hive Mind
The natural progression of this idea is to build the most popular team of all. I found a team which follows all the rules (including using 150 points, and all the restricted rules) and where each rider is picked by an average of 32.2% of teams. I’m going to talk to Urusla and superTed and see if there’s a way to get this team added to the database, but if not, it is easy enough to update periodically and we can all follow along and see how we do against par (and, indeed, how many riders we have in common).
If you could be bothered, I suspect you could slightly increase that average by playing around with the optimisation (I left out a few lower-priced riders to bring in Gaviria and get the points to work – replacing Gav with a more popular ten pointer, for instance, might bump up the popularity) but as an exercise in demonstrating the key riders, this is quite helpful.
I also think that team looks… pretty good? There are a few picks I wouldn’t have made but it isn’t too bad. On a tiny sample size, that might be right. Last year’s hive mind team (I left out Fabio Jakobsen as my team was too expensive, and bulked up the top 25 with two two pointers and a one pointer, after leaving out Jakobsen and some additional restricted riders) scored 12,041 points, good for 88th place.
It isn’t easy to pull data together that directly addresses MOUA’s hypothesis that the regulars on this site will converge on the same cheap riders, but a combination of a peculiar 2020 and yet another wrinkle in Ursula’s diabolical scoring system will, I think, place additional importance of getting the right combination of elite riders. Although for clarity I’ve focused on the restricted riders, I think that anyone hitting on their spend on riders over 10 points can expect a very successful season.
That was somewhat true last year, but the winner was able to make up for disappointing returns from Higuita and Ciccone, whilst the second overcame the drag of Pinot and third place managed to thrive despite picking both Remco and Gaudu. A fuller season and a different structure means I’d be astonished if that repeats itself.
Yes, the lower-priced gems can still give you a massive boost, and that’s still where the fun lies in picking a great team. But you’d better hope you gave serious thought to your elite riders… and you’d better get lucky as the season progresses.
A few more quick thoughts from the reveal.
Always interesting to look at the fallers from last year’s top selections. Remco’s drop was precipitous (441 teams to 24) but readily explained, and Vlasov (321 to 56) had the hammer dropped by Ursula, and rightly so. Looking for a “post hype sleeper” as US fantasy writers would say? How about Alberto Dainese, nicely set up at DSM for a big second year in the world tour, still a promising sprinter, and now representing just 71 teams, down from 346. Wish I’d thought of that before I finalised my team.
I mentioned in one of the threads that Andreas Leknessund getting onto 218 teams was the biggest surprise in the top 25 for me. That’s not to knock him as a rider, I just didn’t think he’d have that kind of prominence in people’s minds. He’s another rider who’ll be hoping the Olympics go ahead as planned, and hoping that he outperforms Dainese as the popular DSM/Sunweb one-pointer of the year.
My biggest surprise of all, however, was the relative unpopularity of the 14- and 16-point riders. I thought lots of people would be struggling to spend all their points, simply because I was. My approach was to replace Pog/Ala (I never really finalised which one I wanted) with Roglic and spend twenty points on Demare. But I had multiple iterations with a few expensive non-restricted riders. In the end I only took Schachmann (on 139 teams) but the lack of love for Hindley (46), Lampaert (54), Naesen (39) and Timmeh (38) was particularly surprising to me. Obviously, you guys liked the 5-12 range far more than I did.
Speaking of which, if it isn’t too soon for regrets, who would you guys swap teams with? Right now, it is still to soon for me to want to give up my squad, but if you made me, I’d be happy to swap with Queen Lemond. Long Live the Queen looks a lot like my team did about three days before submission, and has a bunch of riders who I really wanted.
So, that’s me. What did you guys learn?