Theres a hill on that stage which would probably be somewhat of a factor for a 9.5km prolougue. Would that explain it? Also, if it was a TT/PR ratio based on being longer than 7km, a rider with 77 TT+PR would be better than a rider with 80PR and 58 TT. (80x.8=64, 58x.2=11.6, 64+11.6=75.6)
The ratio is just a guess, but its probably something like that.
I'm not saying that Keough was perfect for that stage, but that PRL really stops having any sort of significant impact at around 10km... so if you want a PRL type stage for a race it really needs to be very short.
Oh my mistake, I must have misinterpreted your post a bit, I can see what you are saying now. I think the way its implemented in the game, a rider with 80PR and 75 TT for example, (which is a completely plausible skillset) would perhaps be one of the favourites, but 80/58 is not one you would really see on a real rider and in game a rider like that is not all that great.
I think moving forward we must be careful on what we call a prologue, and what actually is. As once it starts using a mix of TT and PR I'd say it isn't really a prologue at that point especially with the extreme stat matrix the MG can have at times.
Theres a hill on that stage which would probably be somewhat of a factor for a 9.5km prolougue. Would that explain it? Also, if it was a TT/PR ratio based on being longer than 7km, a rider with 77 TT+PR would be better than a rider with 80PR and 58 TT. (80x.8=64, 58x.2=11.6, 64+11.6=75.6)
The ratio is just a guess, but its probably something like that.
I'm not saying that Keough was perfect for that stage, but that PRL really stops having any sort of significant impact at around 10km... so if you want a PRL type stage for a race it really needs to be very short.
Oh my mistake, I must have misinterpreted your post a bit, I can see what you are saying now. I think the way its implemented in the game, a rider with 80PR and 75 TT for example, (which is a completely plausible skillset) would perhaps be one of the favourites, but 80/58 is not one you would really see on a real rider and in game a rider like that is not all that great.
I think moving forward we must be careful on what we call a prologue, and what actually is. As once it starts using a mix of TT and PR I'd say it isn't really a prologue at that point especially with the extreme stat matrix the MG can have at times.
To be fair, some sprinters have done well in prologues over the years without any TT skill irl, although maybe it's more rare now. Farrar had some pretty good results, and Cav wasn't too shabby either at times, besides both of them never doing anything in a full length TT. Our whole "track-sprint" dev model is based on an actual archetype.
It makes sense though that in order for as extreme of a difference as 58/80 to not matter much, we'd have to be talking an extremely short distance.
RIP Exxon Duke, David Veilleux, Double Feature, and Monster Energy
[url=https://pcmdaily.com/forum/viewthread.php?thread_id=55199&pid=1354841#post_1354841]baseballlover312 wrote:
To be fair, some sprinters have done well in prologues over the years without any TT skill irl, although maybe it's more rare now. Farrar had some pretty good results, and Cav wasn't too shabby either at times, besides both of them never doing anything in a full length TT. Our whole "track-sprint" dev model is based on an actual archetype.
It makes sense though that in order for as extreme of a difference as 58/80 to not matter much, we'd have to be talking an extremely short distance.
Tyler Farrar was the first guy that came to my mind too, he used to be very good in those opening prologues of many races, which were frequently under 5km. But I'd never consider him a TT specialist. Peter Sagan is another rider who when he was younger at least was very good at those prologues as well, and while he has obviously an overall good skillset he isn't really a TT specialist.
I don't really see what is the problem here though. That stage shows what valv said. Only short time - trials taken into account only the prologue stat. After that time - trial matters too.
In the stage that you mentioned, the prologue "specialists" have bad time - trial. Like really bad. Look at Bille who even though his time - trial is weaker than the prologue is still somewhat acceptable and he did decent enough.
When a stat that is taken in consideration as a main one is under 70 (even under 60 for your rider) well it's normal that that rider is taken out of contention for the first places. Keough still did reasonably well, only losing 20 seconds given his really weak TT stat. This proves that the prologue stat still mattered, but now that time - trial is taken into account, with such a bad stat it's normal that he won't be in the first places.
alexkr00 wrote:
I don't really see what is the problem here though. That stage shows what valv said. Only short time - trials taken into account only the prologue stat. After that time - trial matters too.
In the stage that you mentioned, the prologue "specialists" have bad time - trial. Like really bad. Look at Bille who even though his time - trial is weaker than the prologue is still somewhat acceptable and he did decent enough.
When a stat that is taken in consideration as a main one is under 70 (even under 60 for your rider) well it's normal that that rider is taken out of contention for the first places. Keough still did reasonably well, only losing 20 seconds given his really weak TT stat. This proves that the prologue stat still mattered, but now that time - trial is taken into account, with such a bad stat it's normal that he won't be in the first places.
I don't think it's a problem at all really. I was just mentioning that rider types with low TT and high prologue do exist in real life, and are represented by MG development tracks. The gradient makes sense as well. Just seems to be a question of at what distances that gradient is.
I think the discussion is not really to identify a problem, but just to figure out how exactly the stat works so managers can use that knowledge. Prologue has been a stat of confusion for me forever basically. I remember in PCM 11 reading it was active until 8 km and not afterwards. I have no idea where I got that from, but I based decisions on it for a few years after that. Obviously it was wrong. It never hurts to have more knowledge of this stuff out there, so managers are informed.
RIP Exxon Duke, David Veilleux, Double Feature, and Monster Energy
baseballlover312 wrote:
I think the discussion is not really to identify a problem, but just to figure out how exactly the stat works so managers can use that knowledge.
This isn't necessarily right, I don't know how in-depth the tester got, but from the PCM18 "Understanding rider attributes" thread:
EDIT: Also not sure if it scales discretely as above or continuously (e.g. with 17.5km as 50:50), assume it's the latter, but that gives a pretty good understanding
Edited by jandal7 on 28-04-2020 10:34
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
I was under the impression that Kentaurus was saying that prologue stats don't have an impact on a shorter time - trial (short but longer than a classic prologue), in the way valv suggested. But I think they do and even the example he used proves that.
From how I've been interpreting TT results this season informed by the info Jandal references above, I'm quite convinced if not exact is very representative. I have f.e. Boilly: TT 70, Pro 61 so have been paying close attention! It is why I advocated a couple of pages back that the OVR rating of riders needs to incorporate some balance of TT and Pro stat, not just TT stat as seems to be current model.
Re: The Tour of the Vineyards result being discussed specifically; I think the presence of riders like Brandt, Nibali and Eastman in the top 10 (as well as the likes of Borisavljevic and Cort Neilsen in the top 20) indicates that although it looks pretty minor on the profile, on the upward slope a combo of Pro and Mo has been applied as we have been experiencing in other Pro/TT events where the 'hill' is perhaps more apparent. The balance on this course actually seems to have been very good as the top 5 is still Pro type riders, some with v. poor Mo, whereas Nibabli and Brandt have made up for their TT/Pro = 73 stat somewhat on the 'hill' section.
Additional edit: And similar could be interpreted for today's Post Danmark Rundt stage 5, except the balance here is more weighted to TT than prologue on length (obviously), but Mo again a partial factor due to gradients on the course. Sterbo and Fraile benefitted from this, some of the top TT-only specialists still got close. Daily form still a big factor in TT events though, or how else do we explain Coppel's ride today!
As Roturn indicated in a previous post if I interpreted correctly, course design for ITT's intended for the true powerhouse TT riders may need to be pan flat, if we want to incorporate hill stat, then _________/ profile. Undulating profiles will pull in the Mo stat to a greater or lesser degree depending how long/steep slopes included. This would also make sense for Prologue/short ITT course design as well.
Edited by Scorchio on 28-04-2020 00:56
You still have to keep in mind the daily form. Altough it's very possible the hill had an impact, it could have nothing at all to do with it. Daily form could also be a a plausible explanation. Anything +2 would give a huge boost. Even a +2 could make those 73 rated riders top 10 candidates.
baseballlover312 wrote:
I think the discussion is not really to identify a problem, but just to figure out how exactly the stat works so managers can use that knowledge.
This isn't necessarily right, I don't know how in-depth the tester got, but from the PCM18 "Understanding rider attributes" thread:
EDIT: Also not sure if it scales discretely as above or continuously (e.g. with 17.5km as 50:50), but that gives a pretty good understanding
Just one small data point to support this from Stage 6 of the Olympia's Tour, which was an 8km ITT. Chavanne was there, who with 65 TT and 82 PRL is a good test case. If you assume the percentages given by jandal above scale continously, 8km would give you a 68:32 split in favor of PRL, which would give Chavanne a weighted value of 76,6. That corresponds pretty well with his 16th place at that stage. Nascimento, who has 77 TT and 61 TT (composite score of 66,1) finished 88th.
Scorchio wrote: Additional edit: And similar could be interpreted for today's Post Danmark Rundt stage 5, except the balance here is more weighted to TT than prologue on length (obviously), but Mo again a partial factor due to gradients on the course. Sterbo and Fraile benefitted from this, some of the top TT-only specialists still got close. Daily form still a big factor in TT events though, or how else do we explain Coppel's ride today!
Wasn't sure about MO at first, but I do agree it does seem to have had some influence. If you filter all 77+ time trialists and sort them by mountain stat, 6 of the 7 best climbers (which means 69 mountain or more; it's a drop to 66 after that) were in the Top 10, but only one (stage winner Fraile) had 79 TT, the others 77 or 78, so they were punching above their weight. At the same time, that was also true for third place on the stage, Haugaard, who only has 57 MO, so daily form definetely played a big part as well. Durbridge and Kittel with 58 and 56 MO each were also able to perform somewhat well (4th and 9th). Dillier with 51 MO completely bottomed out though and finished 31st. Hill stats did not seem to have a similarly strong influence.
Stage profile for those who didn't follow the stage:
TT wise there definitely will be a mix of long, short, very long, medium TTs. That is pretty sure.
The thing to fix is that e.g. above Denmark`s ITT, which was supposed to be flat, is even flatter as those small bumps use the mountain stat and not the hill stat, which would at least make more sense normally.
Hill wise there also won`t only be classics. Pure hill stage races can work and can be next to hill/mountain stage races.
So as I see it, the sprinter issue is obviously the big issue. Obviously teams can adjust to the issue and focus on less strong sprinters that still reward pretty decent points (Groenewegen, Kennaugh, Howard at around 350k wage) or improve the sprint train as the examples above (Podium Ambition) shows.
Right now (a bit in the past ranking wise) sprinters like Swift, Grosu, Ewan, who are on ~600k wage score pretty similar around 1,2 points/1000 € wage.
Similar wage but less scoring can have to do with being in the Tour de France, which is obviously not raced yet and so race days would need to taken into calculations mid season as well, which they aren`t.
Pretty much low scoring seem to be the hilly sprinters. From overpowered to regular it seems, so here the formulas might need to be reduced again to make them a bit cheaper. That said, calculations need to be done again a bit later when having more race days.
That said, the fix of stage design for hill stages will already help them clearly as they use the hill stat more often then and not the mountain stat, which makes them underperform in many races this year I feel. So might not need a formula update but simply stage design.
I think what works best overalö would be to change the formulas for the wage/rd a bit so that sprinters wage are changed a bit with the top sprinters earning less, medium sprinters maybe a bit more or same (as it was only just highered already) and lower sprinters earn similar as now.
Sprinters that drop out of this a bit in the positive like Kemboi and Guardini are pretty decent riders for low wage (less than half of category Groenewegen). Though the total points are much lower as well and Kemboi a) was only maxed now and b) had this weird awesome Giro as main source of points. So basically I guess they might still work fine at the bottom of the formula.
roturn wrote:
Pretty much low scoring seem to be the hilly sprinters. From overpowered to regular it seems, so here the formulas might need to be reduced again to make them a bit cheaper. That said, calculations need to be done again a bit later when having more race days. That said, the fix of stage design for hill stages will already help them clearly as they use the hill stat more often then and not the mountain stat, which makes them underperform in many races this year I feel. So might not need a formula update but simply stage design.
counter point: The stages with changes of the stage profiles might strengthen "normal" sprinters on the stages as well that are supposed for hilly sprinters so that they dont have too many stages where they are as heavily favoured as in PCM15 so some sort of adjustment certainly is needed. (or i might not have understood the effect of those adjustments)
roturn wrote:
Pretty much low scoring seem to be the hilly sprinters. From overpowered to regular it seems, so here the formulas might need to be reduced again to make them a bit cheaper. That said, calculations need to be done again a bit later when having more race days. That said, the fix of stage design for hill stages will already help them clearly as they use the hill stat more often then and not the mountain stat, which makes them underperform in many races this year I feel. So might not need a formula update but simply stage design.
counter point: The stages with changes of the stage profiles might strengthen "normal" sprinters on the stages as well that are supposed for hilly sprinters so that they dont have too many stages where they are as heavily favoured as in PCM15 so some sort of adjustment certainly is needed. (or i might not have understood the effect of those adjustments)
Totally unbiased ofc.
Obviously needs testing, yes.
But my point was that the stages in general aren`t supposed to become easier! They are supposed to use the hill statt instead of the mountain stat.
This should then favour Van Stayen, Demare, Kump, Van der Lijke etc.
Right now they are dropped same as the pure sprinters as mountain makes them suffer the same.
And yes, when hill stat counts, pure sprinters obviously on a good daily form can still challenge them, e.g. Swift in Paris Nice 1-2 years ago and/or in RL Hushovd going crazy on some pretty hilly finishes.
But indeed must be taken care to get a decent balance. This year was supposed to be quite kind to hill sprinters with GTs having pretty difficult flat stages, that said, the H/MO ratio then might have gone against this plan.
Comparing TTs as done in the closing TT in Denmark remember that pure TT specialists will always have a disadvantage there due to race fatigue. That explains while the specialists on tougher terrain perform better, cause they go into the ITT with more energy left.
That has been the same in 2015 already obviously...
Yup. Some interesting stuff for sure. The last one won by Evonik was probably somewhat realistic though. There wasn't many sprinters left and they tormented the GC fast men with their tactics
Been catching up on the posts here so one thought and some more testing:
roturn wrote:
Sprinters that drop out of this a bit in the positive like Kemboi and Guardini are pretty decent riders for low wage (less than half of category Groenewegen). Though the total points are much lower as well and Kemboi a) was only maxed now and b) had this weird awesome Giro as main source of points. So basically I guess they might still work fine at the bottom of the formula.
I am not sure how weird Kemboi was - if you sort all the 79+ sprinters by FL+RS top two are Kemboi and Holloway both of whom have been good. I think most of the rest at the top have done well considering they tend to be lower SP. I think this is a good thing about 18 that it requires more complete sprinters. Particularly the low end guys shouldn't be good (how can you be 64 FL and 80 SP?). I would vote any FA sprinters with less than 135 combined FL + RS be retired or have their SP + AC lowered. Otherwise they tempt managers into signing them cheap and then they clutter sprints before collapsing before they reach the line.
Full list sorted by RS+FL
Spoiler
FL+RS
SP
AC
Holloway
155
80
82
Kemboi
155
79
80
Bewley
154
83
83
Guarnieri
152
82
81
Kragh Andersen
151
80
77
Gaviria
149
79
82
Manninen
149
80
80
Houle
149
79
78
Lander
149
79
77
Silvestre
149
80
81
Rowe
148
79
80
Kennaugh
148
82
81
Petit
148
80
78
Van der Sande
148
79
79
Te Brake
147
79
76
Degenkolb
147
84
81
Ahlstrand
147
83
82
Kump
147
81
80
Van Asbroeck
147
80
78
Coquard
147
84
83
Saber
147
81
79
Van Stayen
147
84
78
Granjel Cabrera
146
80
77
Keough
146
79
79
Enger
146
81
78
Colbrelli
146
79
75
Howard
146
82
79
Kristoff
146
83
79
Ewan
146
83
81
Coutinho
146
79
80
Krieger
145
79
79
Von Hoff
145
79
77
Demare
145
82
81
Samolenkov
145
80
79
Aberasturi
145
79
78
Stallaert
145
79
76
Boivin
144
79
80
Boeckmans
144
81
83
Swift
144
84
82
Zabel
144
81
79
Perera
144
79
80
Bertilsson
144
79
77
Van der Lijke
144
81
79
Roelandts
144
79
77
Pelucchi
144
79
77
Groenewegen
144
82
81
Scully
144
79
80
Stauff
143
80
81
Haller
143
80
79
Vanderbiest
143
82
77
Bouhanni
143
80
77
Afewerki
143
79
80
Grosu
143
83
80
Goss
143
81
77
Lutsyshyn
143
79
80
Van Poppel
143
81
78
Ciolek
142
82
78
Matthews
142
79
79
Stepniak
142
79
79
Cavendish
142
82
80
Itami
142
80
82
Guardini
141
80
82
Havik
141
79
76
Moser
141
80
79
Mezgec
141
79
77
Soupe
141
79
77
Maksimov
141
81
84
Serebriakov
141
80
77
Hayakawa
141
79
79
Vesely
141
81
80
Willwohl
140
79
78
Kupfernagel
140
80
80
Appollonio
140
81
78
Peeters
140
79
77
Keukeleire
140
80
79
Nizzolo
140
80
79
Bennett
139
79
79
Lay
139
79
81
Ulanowski
139
80
77
Suryadi
139
79
76
Selig
139
79
81
Carsi
139
80
82
Guerao
138
83
76
Van Heerden
138
81
72
Vingerling
137
79
79
Avelino
137
83
78
Lavoine
137
80
78
Meyer
137
80
76
Vanderaerden
137
79
75
Reinhardt
137
79
84
Manarelli
137
79
79
Mohs
136
82
80
Nelson
136
81
81
Vanoverschelde
136
80
75
Drapac
136
80
79
Castaneda
135
79
77
Lo Cicero
135
84
78
Vantomme
135
81
76
Salleh
135
80
80
Tzortzakis
132
79
74
Kip
132
80
82
Rowe
130
80
70
Guillen
129
81
82
Awang
126
79
83
Thorsen
114
81
83
EDIT: BBL has noted that FL and RS can't be everything since AKA hasn't done that well - don't have answer for that beyond saying that I think it is a factor not the factor.
So I was struck by how Cunego's tests seemed to argue for lead outs and Alex's against it.
So I set up a 20 team race with every rider with 73 Fl, 73 RS all the rest 65. Except I gave 1 team an 81/81 SP and plus 3 reasonable lead out riders and another team an 81/81 SP leader and no lead out. The goal was a simple clean test of lead out vs no lead out. I used 100km flat course to avoid any risk of leadout riders being dropped and a 10KM straight 4 lane finish.
On the surface the results seemed to support the lead out - the rider with lead out won all 4 trials. In two cases the rider with the strong lead out got a great lead out from his top lead out man and won. But in the other two his lead out was not particularly effective, in both cases his lead out rider left without him. But he still won partly because the sprinter with no lead out tried to form a lead out with a 65/65 sprinter and that clearly isn't good.
Because the second sprinter tried to form a lead out I abandoned this test because it was testing good lead out vs bad lead out and not good lead out vs no lead out.
So then I added two more teams with 81/81 sprinters and the same quality lead out trains. So we have 4 equal sprinters, 3 of whom have equal trains and one who has none.
I was hoping I would get 2 or 3 lead outs and one rider surfing (presumably the one with no lead out riders).
So I ran 4 tests with this, and in each cases two of the teams tried to form leadouts typically with mixed success. One of the teams never tried t form a lead out. More on that in the next post but the weird thing is it wasn't the team with no lead out riders.
In two of the four tests one of the teams trying to set up a leadout was the team without any qualified lead out riders. In one of those they had the last surviving lead out.
In 3 of the 4 cases the team with the last lead out train lost when a sprinter came off their wheel. In all 3 it was because the lead out dropped them off too soon and all they did was lead out their opponent. In one of these it was the bad lead out team so maybe not surprising. So the good lead outs were successful 33% of the time. Which isn't great but I think I need to run it more times as if the led out rider wins the next one it goes to 50% and that is a different answer
So I think I have seen enough to say a good lead out is better than a bad one. But still not sure if having a lead out is better than no lead out.
I also think real life sprints have changed in the last few seasons - typically there is only 1 surviving lead out train and it is often very messy. So maybe credit to Cyanide and they are trying to reflect this.
I would have done more of test 2 (and maybe I will) but I got distracted by a different issue. More on that in the next post since this is kind of long already.
Edited by Ulrich Ulriksen on 02-05-2020 04:13
OK, volume 2: So while I was doing test 2 in the post above I made an error and it led me to something else weird ( a bit like the discovery of penicillin and almost as important)
In setting up the lead out for one of the teams I accidentally put the stamina of the best lead out rider at 77. I didn't notice but when I looked at the favorites screen that rider was listed as 3 star favorite ahead of his own teammate. So a rider with 77/77 SP/AC was ranked ahead of a rider with 81/81 SP/AC because his stamina was higher (77 vs 65) in a 100km stage. All other stats were lower or the same for the 77/77. I reset set the stage multiple times and this was always true. I then lowered his stamina, at 70 he was no longer favored but at 73 he was. So Stamina is reasonably heavily weighted in the setting the favorites. By the way the 77/77 guy was also listed as the team leader in the game (rider 1).
So first, I was interested in whether there was any basis for that - does ST matter in a sprint of 100KM. The answer is no. I tried multiple different scenarios and in every one riders with the same SP and AC finished the same regardless of ST. I think this is just a bug in the favorite algorithm. Or maybe the favorite algorithm uses a general formula that is not sensitive to stage length.
So does it matter? Does the AI behave differently. I don't think it does because of Stamina specifically but I do think being a "favorite" matters. A rider with poor stamina might be moved down the list unfairly in a short sprint resulting in less favorable treatment.
I think in Alex's work it was noted that STA might have an influence on lead out riders. But still don't have an answer to whether having a lead out helps.
