Wielerfox Rider Pricing

Wielerfox Rider Pricing

Data Source

compose exec -T db-dev psql --csv <<_EOF > riders.csv
select
    riders.dob,
    riders.short,
    riders.name,
    riders.country_long,
    competitions.cat,
    teams.cat as uci_cat,
    competitions.year,
    riders_seasons.pcs_total,
    riders_seasons.pcs_rank
    from riders
    join riders_seasons on riders_seasons.rider_short=riders.short
    join competitions on competitions.id=riders_seasons.competition_id
    join teams on riders_seasons.team_id=teams.id
    where competitions.year >= 2017 and competitions.year <= 2025
    order by competitions.year, riders.short

Basic stats

year	cat	uci_cat	n
2020	MEN	PRT	320
2020	MEN	WT	536
2020	WOMEN	WTW	58
2021	MEN	PRT	367
2021	MEN	WT	559
2021	WOMEN	WTW	77
2022	MEN	PRT	347
2022	MEN	WT	552
2022	WOMEN	WTW	115
2023	MEN	PRT	383
2023	MEN	WT	534
2023	WOMEN	WTW	176
2024	MEN	PRT	374
2024	MEN	WT	535
2024	WOMEN	WTW	213
2025	MEN	PRT	379
2025	MEN	WT	527
2025	WOMEN	PRW	100
2025	WOMEN	WTW	248

PCS Points dist

more or less as expected: - many with 0 or few points. - a little bit fatter in the middle - long tail -> few with a lot of points

convert this to costs: - scale 1 - 61 (most expensive, up for discussion, initial budget 360 for 25 riders)

year	max_pcs	min_pcs	mean_pcs	median_pcs	q25	q75	q90
2020	2431	1	151.35	58.5	25	160.75	406.6
2021	3328	1	214.79	92.0	32	255.00	560.6
2022	3413	1	232.83	101.0	40	286.75	577.8
2023	3602	1	251.43	118.0	50	298.00	648.4
2024	4588	1	253.84	122.5	44	321.00	647.9
2025	4021	1	214.58	99.5	39	262.75	543.1

interesting: women seem so have a wider dist, fewer dominant riders?

Pogacar Effect?

completely misinterpreted the pogacar effect. actually no impact on density (assumed the longer right tail would make more difference)

Pricing

How can people allocate a fixed budget on a team that has to be exactly n riders

boring strategies:

• superstar, fill rest with cheap
• superstar + some normal stars, fill up

better:

• stars + careful selection to overtake?
• chance for newcomers/undervalued riders?

in a way you will need to find riders that no one else has, otherwise you will not get ahead (everyone gets the same points) but if you don’t have the superstars other people have you will get left behind.

So: How do I discourage superstar collections?

The idea would be to have different “buckets”

• class A: very high prices for the few riders with most points
• class B: high prices for a few riders with many points
• class C: medium prices for persons of interest earning some points
• class D+E: basically 1 pointers, few points

price ranges, assuming max price of 60:

• A: 50-60 (Vingegaard, Pedersen, …)
• B: 35-50 (VdP, Pidcock, …, AdL)
• C: 5-35
• D: 1-5 (potentials, some chances)
• E: 1 (newbies, oldies, pure helpers, pack fill)
• A++: Inf: Pogacar

counts: - A: 1% -> 5m, 2-3f - B: 5% -> 25m, 10-15f - C: 20% -> 100m, 50f - D: 25% -> 120m, 60f - E: 50% -> 250m, 120f

Prices + budget should be balanced so that you will need to find good B+C and also look in D+E to fill up

A maximally interesting team would look like this?

• 10 (C+D)
• <=3 A
• <=5 B
• 10 E

I guess this is what the PdC people did:

• bucketize people, assign price ranges for buckets, price within
• restricted + double-restricted riders to balance strategies

mayb a pricing/budget balance could be found that does not require (semi)subjective rules, and decisions can blame “the algorithm”

Mapping variants

linear see some previous attempts

Gompertz based on points

Gompertz plots

Explore Parameters

Tables

Hist

Solution?

Stratification

Team Selection Strategies

Price Mix

We can use linear programming to maximise points by

Generate optimal team

Does not really work — too many for optimal solution.

## Error: no feasible solution found

short	cat	year	price	pcs_rank	pcs_total	normalized_points	rider_class

## # A tibble: 1 × 2
##    cost total
##   <dbl> <dbl>
## 1     0     0

Adjust Prices

Let’s transform the calculated prices to fit the range of pdc prices (which were priced from 1-36 (Pogacar is 46, special price, see!))

Wielerfox is 1-60, with Pogacar at 122, an even specialer price. It seems like we have about double price, so let’s scale down to half, while maintaining a minimum of 1. This shifts the lower end a little bit to the left.

Done!

write csv for import with: wielerfox prices import