|
PCM 18 AI and Man Game DB
|
| TheManxMissile |
Posted on 24-04-2020 12:42
|
Tour de France Champion
Posts: 18187
Joined: 12-05-2012
PCM$: 0.00
|
jandal7 wrote:
George Bennett 37th in Catalunya in a field which he entirely beat later in Japan (barring Uran and Tenorio). Both 6 day MO/HI races.
Bennett isn't quite a Top5 ranking rider (sorry!) and for that race with it's 40km ITT i can see why the AI maybe didn't do him justice compared to Japan with it's 3km ITT and MTT which much more suits his abilities (i.e. in Japan i'd say he was a Top3 favorite and performed as such, compared to Catalunya where he was a low Top10 level and then vanished).
Ewan, a Top5 ranking rider, in a race where he is a Top3-5 favorite (like Bennett in Japan) who then vanishes completely from the Top15. That's the comparision i'm after (again, no offence to Bennett)
|
| |
|
|
| SotD |
Posted on 24-04-2020 13:19
|
World Champion
Posts: 12188
Joined: 29-11-2006
PCM$: 2980.00
|
Tamijo wrote:
Testfile
4.0 pancake flat 172 km. No turns in last 10 km.
No garanti it will be different, but from a stage profile point of view we need to know not to use many hours if it is to no use, can't eliminate daily form but otherwise this should be a perfect flat route.
https://www.dropb...6BRua?dl=0
For panflat stageraces would it be an idea to eliminate daily form? Would that give better results? I mean, if the randomness is this big, maybe removing daily form would give atleast somewhat better results... Just throwing ideas up. I don't want to remove daily form for any other type of races obviously as they seem to work well, but those few races marked for sprinters only could maybe use that.
|
| |
|
|
| Croatia14 |
Posted on 24-04-2020 13:59
|
Directeur Sportif
Posts: 9099
Joined: 13-03-2013
PCM$: 2100.00
|
@SotD: I look mostly as whether Flat, Acc, Res and Sprint work together. Obviously nobody really knows how it work out exactly, but it's your and our task to figure out how. I can only talk about what I think works best from what I saw, but I could be horribly off. F.e. I think that Bewley would absolutely dominate with a leadout like f.e. Oss - Houle - Bewley. But in testing that could turn out completely wrong. I think the most important thing is that all 3 have a similar frame but are clearly other tiers of sprinter levels, so that you can properly build up a train. With an 84 Sprinter I'd go with 76/77/maybe 78 Sprint - 80/81 Sprint - 84 Sprint. Similar Acc deficits.
I still stand by my point that the leadout system isn't worse than in PCM15, it's just that nobody was adjusted this time. This time it benefits second tier sprinters, back then they were more useless than they should've been. Sprinters failed in PCM15 as well, I remember roturn's Groenewegen rants. There were other flaws like the best sprinters being too good. Now it turned around for some, which is good for other people. But overall I don't think it's worse (saying that with Kump that completely stopped trying anything in the hilly sprints he excelled at in PCM18 and also getting beaten left and right in Vuelta-Sprints where he should've been 2nd best by a landslide).
@TMM: I can't but I think if sprinters leading each other out have too similar acc one wears the other out when he accelerates. That's where my theory of Nelson leading Ewan out breaking Ewan comes from.
also @TMM: basically any puncheurs stage race is as bad as what you are referring to...
|
| |
|
|
| Ad Bot |
Posted on 28-11-2024 11:20
|
Bot Agent
Posts: Countless
Joined: 23.11.09
|
|
| IP: None |
|
|
| Tamijo |
Posted on 24-04-2020 14:08
|
Team Leader
Posts: 7406
Joined: 14-07-2015
PCM$: 599.00
|
SotD wrote:
Tamijo wrote:
Testfile
4.0 pancake flat 172 km. No turns in last 10 km.
No garanti it will be different, but from a stage profile point of view we need to know not to use many hours if it is to no use, can't eliminate daily form but otherwise this should be a perfect flat route.
https://www.dropb...6BRua?dl=0
For panflat stageraces would it be an idea to eliminate daily form? Would that give better results? I mean, if the randomness is this big, maybe removing daily form would give atleast somewhat better results... Just throwing ideas up. I don't want to remove daily form for any other type of races obviously as they seem to work well, but those few races marked for sprinters only could maybe use that.
We use deactivate random fitness - but as i understand it the daily form in single player we can't remove.
|
| |
|
|
| tsmoha |
Posted on 24-04-2020 14:54
|
Directeur Sportif
Posts: 11819
Joined: 19-07-2010
PCM$: 300.00
|
Croatia14 wrote:
also @TMM: basically any puncheurs stage race is as bad as what you are referring to...
That's true 
|
| |
|
|
| jandal7 |
Posted on 24-04-2020 15:02
|
World Champion
Posts: 11395
Joined: 17-12-2014
PCM$: 1020.00
|
TheManxMissile wrote:
jandal7 wrote:
George Bennett 37th in Catalunya in a field which he entirely beat later in Japan (barring Uran and Tenorio). Both 6 day MO/HI races.
Bennett isn't quite a Top5 ranking rider (sorry!) and for that race with it's 40km ITT i can see why the AI maybe didn't do him justice compared to Japan with it's 3km ITT and MTT which much more suits his abilities (i.e. in Japan i'd say he was a Top3 favorite and performed as such, compared to Catalunya where he was a low Top10 level and then vanished).
Ewan, a Top5 ranking rider, in a race where he is a Top3-5 favorite (like Bennett in Japan) who then vanishes completely from the Top15. That's the comparision i'm after (again, no offence to Bennett)
Yes you're probably right in that case, and certainly that he isn't quite as good as Ewan - no offence taken  forgot how lengthy that TT was, but the point still stands that I don't think the variance is limited to sprints necessarily even if it might be heightened there - more research needed possibly, but not sure how much it will help us in terms of where to go from there - unless for people this would be enough to want to leave PCM18?
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
5x  x5
2x  x2
2x  x2 |
| |
|
|
| TheManxMissile |
Posted on 24-04-2020 15:12
|
Tour de France Champion
Posts: 18187
Joined: 12-05-2012
PCM$: 0.00
|
As i said i don't follow along that well outside my own team. If it's a regular occurence, then whilst i still think it's an issue, it's just something to get used to. It's just feels so wrong when compared to playing PCM myself with a normal DB and not seeing these crazy swings without a crash involved.
Maybe i need to stop playing PCM again
______
Equally with all this, i'm not one of these people who'll quit the game either. At least i'm discovering all these things in a season where i knew i would relegate
|
| |
|
|
| baseballlover312 |
Posted on 24-04-2020 16:11
|
Tour de France Champion
Posts: 16429
Joined: 27-07-2011
PCM$: 10438.70
|
1.) I keep hearing people saying flat is more valuable or you're leadout riders are blowing up your train, but that's just not what I'm seeing in the data at all. I don't know why it's being used as a scape goat. Same thing with resistance. Ranneries has been more successful than AKA this season! Not in terms of just results per leader days, but legit points. Surely that's not because of backups...
Neither has ridden much since the last rankings, although the one race they rode together in the meantime (Romandie), Ranneries actually outsprinted AKA, so the gap is likely bigger in points now.
Ranneries - 73.36 OVL, 50,000 Wage, 22 Points
70 Flat (-7)
56 Hill (-8)
68 Stamina (-2)
65 Resistence (-9)
77 Recovery (-1)
78 Sprint (-2)
82 Acceleration (+5)
Kragh Andersen - 75.78 OVL, 125,000 Wage, 19 Points
77 Flat (+7)
64 Hill (+8)
70 Stamina (+2)
74 Resistance (+9)
78 Recovery (+1)
80 Sprint (+2)
77 Acceleration (-5)
Unless acceleration is suddenly OP - which supposedly is exactly the opposite in PCM 18 compared to 15, there is just no way backup stats are accountable for this. Even in PCM 15, where acceleration was OP, AKA was by far the more successful rider. Now, when flat and resistance are supposedly more important, and AKA has gigantic margins in both, he is worse.
And keep in mind AKA gets way more opportunities for me than Ranneries, who is a last resort leader option and a throw in most of the time. They've ridden together exactly once, and have not made a train together, so it's not that Ranneries is riding away from AKA in a train, which would make sense given the acceleration gap.
I understand these small examples have sample size problems and are somewhat anecdotal, but I follow my own riders closest obviously, and so I feel it's only right that I only speak on the riders I know well enough to judge.
2.) You cannot be expected to micromanage a leadout like Croatia suggests, with the perfect riders slotted to consume the perfect amount of energy, because we do not control what the team does on the road! If a good leadout is good flat/sprint riders, then we can account for that. If it has to be micromanaged more than that, we just can't control that in this game as managers. The chances of the game actually using our domestiques as we want in the train is miniscule, because this is PCM. So that's not a solution imo.
knockout wrote:
Tamijo wrote:
The late hill at stage 4, world often allow a minor split so a true sprinter stage best without late climbs (even small), if we do not lift low MO
Seemingly good fair sprints with a lot of room for moving and changing position, often very close photo finish.
If that small bump towards the end of the stage is ridden in a way that MO instead of hill then the game design is imo completely ridiculous and shouldnt have ever be used for the mg
And I absolutely second this. That just makes no sense whatsoever.
Edited by baseballlover312 on 24-04-2020 16:14
RIP Exxon Duke, David Veilleux, Double Feature, and Monster Energy |
| |
|
|
| roturn |
Posted on 24-04-2020 16:12
|
Team Manager
Posts: 22246
Joined: 24-11-2007
PCM$: 3900.00
|
Regarding the sprinter issue. Just went to Denmark Stage 2 and let me smile a bit with this discussion here.  |
| |
|
|
| Croatia14 |
Posted on 24-04-2020 16:34
|
Directeur Sportif
Posts: 9099
Joined: 13-03-2013
PCM$: 2100.00
|
@bbls point 2: You can try to when building your team in offseason according to the needs and then slot riders in adequately in race planning.
knockout wrote:
Tamijo wrote:
The late hill at stage 4, world often allow a minor split so a true sprinter stage best without late climbs (even small), if we do not lift low MO
Seemingly good fair sprints with a lot of room for moving and changing position, often very close photo finish.
If that small bump towards the end of the stage is ridden in a way that MO instead of hill then the game design is imo completely ridiculous and shouldnt have ever be used for the mg
And I absolutely second this. That just makes no sense whatsoever.[/quote]
Its just wrong though, small hills at the end will always be ridden in a way that if it has any influence it will be the hill stat that is influencing the race.
|
| |
|
|
| baseballlover312 |
Posted on 24-04-2020 16:39
|
Tour de France Champion
Posts: 16429
Joined: 27-07-2011
PCM$: 10438.70
|
Croatia14 wrote:
@bbls point 2: You can try to when building your team in offseason according to the needs and then slot riders in adequately in race planning.
Sure, you can try. And then the game picks the wrong order for a train or has the wrong guy pace and it's completely screwed up anyway. This was always the case, but when such micromanaging is required, the chance of AI failing to act out your plan properly, and thus making you even worse off, are even greater.
Its just wrong though, small hills at the end will always be ridden in a way that if it has any influence it will be the hill stat that is influencing the race.
Well I'd be glad to hear that, but it doesn't seem other PCM 18 reporters like Tamijo agree with you. I'm not sure who I'm supposed to believe here.
RIP Exxon Duke, David Veilleux, Double Feature, and Monster Energy |
| |
|
|
| alexkr00 |
Posted on 24-04-2020 16:54
|
World Champion
Posts: 13915
Joined: 05-08-2008
PCM$: 300.00
|
baseballlover312 wrote:
Well I'd be glad to hear that, but it doesn't seem other PCM 18 reporters like Tamijo agree with you. I'm not sure who I'm supposed to believe here.
I'm curious what Tamijo is basing this on though.
It looks like in all the runs we have at least one guy with less than 60 mountain making it into the top 10. Surely if the mountain stat would have had an impact there we wouldn't have seen that.
|
| |
|
|
| jph27 |
Posted on 24-04-2020 17:44
|
Team Leader
Posts: 7339
Joined: 20-03-2010
PCM$: 900.00
|
alexkr00 wrote:
baseballlover312 wrote:
Well I'd be glad to hear that, but it doesn't seem other PCM 18 reporters like Tamijo agree with you. I'm not sure who I'm supposed to believe here.
I'm curious what Tamijo is basing this on though.
It looks like in all the runs we have at least one guy with less than 60 mountain making it into the top 10. Surely if the mountain stat would have had an impact there we wouldn't have seen that.
It depends how close to the finish it is and the steepness of the climb. If it's within 15km chances it's tackled at a high enough level of effort to use Hill rather than Mountain, and if it's below 5% then it should become about Flat (at least if my understanding of the stats is right). |
| |
|
|
| valverde321 |
Posted on 24-04-2020 17:48
|
World Champion
Posts: 12986
Joined: 20-05-2009
PCM$: 530.00
|
jph27 wrote:
alexkr00 wrote:
baseballlover312 wrote:
Well I'd be glad to hear that, but it doesn't seem other PCM 18 reporters like Tamijo agree with you. I'm not sure who I'm supposed to believe here.
I'm curious what Tamijo is basing this on though.
It looks like in all the runs we have at least one guy with less than 60 mountain making it into the top 10. Surely if the mountain stat would have had an impact there we wouldn't have seen that.
It depends how close to the finish it is and the steepness of the climb. If it's within 15km chances it's tackled at a high enough level of effort to use Hill rather than Mountain, and if it's below 5% then it should become about Flat (at least if my understanding of the stats is right).
I think if its for example 4% it would be a mix of Flat and MO/Hill (MO/Hil depending on the speed/effort)
between 3-5% is a mix of Flat and MO/Hill
|
| |
|
|
| SotD |
Posted on 24-04-2020 18:11
|
World Champion
Posts: 12188
Joined: 29-11-2006
PCM$: 2980.00
|
In terms of the argument that hilly results is equally bad I disagree. The best puncheurs perform the best. maybe not the same riders as ‘15 necessarily but the ones having the best stats for this game. When Bobridge fail to perform it’s easy to track back, that it is because og low MO. Those trackbacks I simply can’t find in sprints.
|
| |
|
|
| tsmoha |
Posted on 24-04-2020 18:16
|
Directeur Sportif
Posts: 11819
Joined: 19-07-2010
PCM$: 300.00
|
SotD wrote:
In terms of the argument that hilly results is equally bad I disagree. The best puncheurs perform the best. maybe not the same riders as ‘15 necessarily but the ones having the best stats for this game. When Bobridge fail to perform it’s easy to track back, that it is because og low MO. Those trackbacks I simply can’t find in sprints.
That's true as well. I was just referring to the fact, that puncheurs, who was supposed to do well (before I knew the MO stat is so important now), do a s**t in stage races. Obviously, I know by know, that Bobridge became a useless 700k rider because of his 69 MO, yes. As for sprints, it's different indeed. Guess after the TT randomness, Cyanide invented the sprint randomness and next year, it's the climbers, too 
|
| |
|
|
| Kentaurus |
Posted on 24-04-2020 18:32
|
Classics Specialist
Posts: 3999
Joined: 26-07-2011
PCM$: 200.00
|
Sadly people need to stop looking at HL as the same stat it once was, it is no longer a primary attribute, but a support attribute. For most of the riders the late hill in stage 2 of Denmark would be ridden with their MO stat (climbing attribute). Those with a bad enough MO to force them to use red bar to maintain their position during the climb however would then use their HL stat.
EDIT:
So I just did a test on that 2nd Denmark stage, using the control riders with edited attributes to be 60MO/85HL and 85MO/60HL (with 72 for everything else). With both riders maintaining position near the front and 99 max effort level.
I'm honestly not sure what attribute was used for that short climb? Both riders topped the climb with about the same yellow bar remaining, neither used their red bar. Both riders were fully ready for the sprint with about 2/3rds of their yellow bar remaining and full red bar.
Edited by Kentaurus on 24-04-2020 18:54
AZTECA - NBCSN  |
| |
|
|
| Tamijo |
Posted on 24-04-2020 18:57
|
Team Leader
Posts: 7406
Joined: 14-07-2015
PCM$: 599.00
|
alexkr00 wrote:
baseballlover312 wrote:
Well I'd be glad to hear that, but it doesn't seem other PCM 18 reporters like Tamijo agree with you. I'm not sure who I'm supposed to believe here.
I'm curious what Tamijo is basing this on though.
It looks like in all the runs we have at least one guy with less than 60 mountain making it into the top 10. Surely if the mountain stat would have had an impact there we wouldn't have seen that.
Or maybe some people just like to read out a PCM18 negative comment into everything I try to do or say.
A: Fact: there was splits
B: I suggest it could be a MO related, but maybe it was something else as we all know you will never be 100% sure, seen a trillion mystic splits in PCM15 impossible to prove why
The fact remains a late climb can make splits, so if you want a 100% safe mass sprint better not use late climbs too often.
What I did not say was that it preventer anyone from getting a result, or in
other ways relating it to rider or results at all I SAID THERE WAS SPLITS in my test on that stage.
It was a sprinter stage no time gabs, I could also have made a novel about that particular stage, but I just said there was splits.
|
| |
|
|
| alexkr00 |
Posted on 24-04-2020 19:07
|
World Champion
Posts: 13915
Joined: 05-08-2008
PCM$: 300.00
|
I was trying to say it's impossible the mountain stat was responsible for the split in my opinion. If that would have been the case, I doubt riders with such low mountain stats would have recovered well enough to get a top 10. It was long enough from the finish for they to recover a bit but I really doubt they would have recovered so much.
Maybe a testing scenario could also be to take a regular team instead of the control team, leave it on the preconfigured scenario and see exactly what the type of riders are losing energy there. Or even in the sprints, for example control Festina and see exactly what is happening with Coquard's energy. On this one, I agree with SotD, whatever happens with Coquard is very strange, sprint trains, hills, corners and whatnot included.
|
| |
|
|
| Kentaurus |
Posted on 24-04-2020 19:26
|
Classics Specialist
Posts: 3999
Joined: 26-07-2011
PCM$: 200.00
|
So I do think with the issues for sprinters, in the best simply not being consistent, that goes for most disciplines actually, that is the fault of a stupidly overpower daily form mechanic... with how closely bunched top riders tend to be (15+ riders) within 2-3 of the primary stat, this will be expected.
Lets say you have a rider with 79 as their primary stat on a +3 day. He will have better ratings than a rider with an 82 primary stat on a 0 form day or worse. This IMO is the biggest thing I hated about the recent games, and the inability of the modders to do anything about it. I don't mind some fluctuation in daily form, but the ratings changes given are just horribly out of proportion.
Even worse, we have no ability to turn this mechanic off either.
AZTECA - NBCSN |
| |
|
![imgur.com/KlJUqDx.jpg[/img[img]]https://imgur.com/yUygrQ.jpg](https://imgur.com/KlJUqDx.jpg[/img[img]]https://imgur.com/yUygrQ.jpg)
