|
Breakaways / Peloton AI - your experiences ?
|
| Popeye |
Posted on 23-06-2013 15:01
|
Under 23
Posts: 77
Joined: 03-07-2008
PCM$: 200.00
|
Because of the importance of the balance between the breakaway and the peleton I have decided to create a specific thread for that.
I would like to gather examples from the community and opinions as to whether the AI works or not.
However to make it a little scientific please include the details of stages played. i.e. teams played difficulty level etc.
It would be great if a few would undertake to race the same stages multiple times and at different difficulty levels.
Edited by Popeye on 23-06-2013 15:02
|
| |
|
|
| NTTHRASH |
Posted on 23-06-2013 15:12
|
Classics Specialist
Posts: 3192
Joined: 09-09-2012
PCM$: 3540.00
|
I play on Hard, and I realize that even though i am chasing at 88 effort or so, I'm only gaining 5" per KM. And that is troublesome. If I max out my effort, half of my riders fizzle out with my chase and we still only gain 15" per KM. Some examples.
Stage 1 of Tour De France. Played with SKY because I felt like using Froome. Last 10 KM, the break was about 1' out. I was in my train for Boasson Hagen, even though he probably would get 5th or so, ideally. So then, at 3KM to go, the break was still 29" ahead! I wondered how it could be, and just launched everyone into a mass chaos sprint. Greipel stayed in wheel until the flamme rouge, where the break was officially caught. Greipel launched, cleared the pack bu 19" and everyone else rolled in at 45 km/h.
Stage 2, breakaway was 3'30" when I started chasing. I caught up 1' by the time we reached 10 KM to go. The break stayed away and won 1'50" or so ahead, and only Froome and Boasson Hagen could even sprint from the effort I used. I exhausted my riders trying to catch a weak break.
Just a little too hard for me.
"America. Show a nipple on television and the whole country goes ape-shit." -DubbelDekker
|
| |
|
|
| Jaixo |
Posted on 23-06-2013 15:40
|
Amateur
Posts: 13
Joined: 13-07-2009
PCM$: 200.00
|
Thanks for making this thread, as I've been quite curious if this is really a problem. I've obviously read the many posts stating this issue. I play career on Hard difficulty, and I'm just finishing the Tour with Saxo Bank. I've done every single WT stage and the classics + random races in 3D race mode and have never experienced a "overpowered" breakaway. They've almost always been pulled in by the AI, especially if it was a flat sprint finish.
I've rarely had the need to do any pulling myself, but in the rare occasions where it was necessary, it worked just fine.
The only time I was surprised with a breakaway winning was in E3 harelbeke where Boom attacked alone with ~80K to go and finished 2 min ahead of Cancellara. |
| |
|
|
| Miguel98 |
Posted on 23-06-2013 15:52
|
World Champion
Posts: 10497
Joined: 23-06-2011
PCM$: 200.00
|
I think it depends on the quality of the rider's in the brekaway itself. For example, I played the Champs-Elysses stage and the breakaway won it, rider's like Kangert, Rogers and more we're in that breakaway. While in a CT race, you have low and medium 60's rider in the break and they get caught fairly easily. |
| |
|
|
| JoshK |
Posted on 23-06-2013 16:46
|
Amateur
Posts: 8
Joined: 22-06-2013
PCM$: 200.00
|
Normal difficulty. With the real name / stats DB.
Stage 1 of Tour De France, jumped on a breakaway right at the start with T. Danielson. Along with me was Roche, Mestre, Nocentini, Quintana.
I kept my team Garmin at the front of the pack but not pacing. With 165km left the 5 man break had a 5'27" lead.
I'm thinking Quintana is the favorite to win if the breakaway holds. But I'm new to this.
With 140km left we had 7'16".
With 100km left we had 5'07". LTB and ARG were the ones doing the pacing at the front of the peloton. OPQ joined at aboutt he 92km mark and it was down to 4'51".
With 50km to go we had 2'50".
With 25km to go we have 1'36".
With 10km to go we have 33". At this point I tell Tyler Farrar to follow Cavendish's wheel. I also launch an attack with Danielson.
Breakaway caught w/ 2.2km left. Tyler is out of position w/ 1.7 left I'm failing pretty hard at this.
Cavendish wins by a bike length, Kittel in 2nd, Griepel in third, Goss in fourth. Tyler Farrar in 10th =\ |
| |
|
|
| Popeye |
Posted on 23-06-2013 16:47
|
Under 23
Posts: 77
Joined: 03-07-2008
PCM$: 200.00
|
Great posts all.
@NTTHRASH
Thanks for the details. IMO your stage one experience is just about plausible. However your stage 2 experience was not realistic at all. 
So far in my experience, playing on normal (what a wimp). I have found that the AI is, on the whole, a lot better than previous years. I have had some nail biting finishes. Breakaways are usually caught around the 10-15Km. But I've seen everything from 28Km to the 1000m. A LOT of fun and MUCH MUCH better than previous versions where without fail the breakaway was caught at exactly 20KM. Every-time. There's seems to be much more variation.
That said I want to live with the game a little while longer before reaching any conclusions. |
| |
|
|
| Lachi |
Posted on 23-06-2013 16:56
|
Grand Tour Champion
Posts: 8516
Joined: 29-06-2007
PCM$: 200.00
|
If you want to test certain situations in the game, you should always play on normal (and it has nothing to do with being a wimp).
If the difficulty setting breaks the gameplay, it would be a different matter. |
| |
|
|
| Ad Bot |
Posted on 27-11-2024 13:40
|
Bot Agent
Posts: Countless
Joined: 23.11.09
|
|
| IP: None |
|
|
| muttisr |
Posted on 23-06-2013 17:00
|
Stagiare
Posts: 178
Joined: 03-07-2011
PCM$: 200.00
|
For me it looks very good. SO far the breakways have only managed to hold off the peloton on middle hard stages, too hard for the sprinters, too easy or to long from last climb to finish line for overall leaders, so in other words as expected.
Only one was at first stage of Tour San Luis, I launched a solo break away after 1m and won by 17 minutes, though it was the peloton wherenoone was riding, maybe because the best sprinter, by no doubt was at my teams were just looking at each other.. But all the temas weh |
| |
|
|
| welker3257 |
Posted on 23-06-2013 17:08
|
Small Tour Specialist
Posts: 2336
Joined: 29-05-2013
PCM$: 200.00
|
I think that the only problem with the breakaways winning is that when the break wins it is by like 30 min. For example in the Tour de Bretange a three man group won by 28 min. The other stages were fine and the break was caught.
Gig 'em Aggies
Fast N' Loud Cycling Project - ICL
|
| |
|
|
| Popeye |
Posted on 23-06-2013 17:12
|
Under 23
Posts: 77
Joined: 03-07-2008
PCM$: 200.00
|
Lachi wrote:
If you want to test certain situations in the game, you should always play on normal (and it has nothing to do with being a wimp).
If the difficulty setting breaks the gameplay, it would be a different matter.
Fair comment. |
| |
|
|
| ruben |
Posted on 23-06-2013 18:03
|
Grand Tour Champion
Posts: 7721
Joined: 23-10-2006
PCM$: 200.00
|
career -> played entire giro and tour, breaks on flat -> almost no chance -> breaks in hilly stages, depend if favorite teams ride or not. Balance I found so far is pretty normal |
| |
|
|
| valverde321 |
Posted on 23-06-2013 19:27
|
World Champion
Posts: 12986
Joined: 20-05-2009
PCM$: 530.00
|
There is definitely an issue with breakaways and simulation though.
I think pretty much every stage I simulated ended in a breakaway.
Also every stage I tried to get my own rider in the break, they made it. I have never been brought back in all 4 of my attempts, so I still think its much too easy.
|
| |
|
|
| cogneto |
Posted on 24-06-2013 02:28
|
Neo-Pro
Posts: 303
Joined: 30-07-2007
PCM$: 400.00
|
I think its pretty real if you dont pay attention some times they might win if the pack dont want to chase cause 1 of there team is in the break away you must check that out cause that was the only time i had to pull them along . |
| |
|
|
| Jones27 |
Posted on 24-06-2013 21:35
|
Stagiare
Posts: 200
Joined: 02-05-2008
PCM$: 200.00
|
I think it Works pretty well. First of all, now you actually get some competition, and you have to fight to get the breakaway back (champ elyees should maybe Work another way around, havn´t played that yet, but offcourse it should be rare that a breaaway wins that stage).
You have to be aware of the new "analyse behaviour" thing. often Theres a reason that no one rides i the front. Fx I was second in the ranking, but did nothing, and the lead riders team did nothing too. Then I saw (in the behaviour thing) that it was because they wouldent do all the Work alone (there was only a few seconds between me and the lead rider in the overall ranking), they ecspected me to help. I then put two riders in front and in the same second the lead riders team helped me, and we worked together. This is good because the AI demands something of you, so you cant just lean back with all riders fit, for the final sprint, and make a way bether leadout train than the other teams. |
| |
|
|
| Housies |
Posted on 25-06-2013 09:33
|
Under 23
Posts: 89
Joined: 31-01-2010
PCM$: 200.00
|
I am doing a career with a custom team. But I grant you the riders are continental quality. Therefore I mainly do 2.2 races.
Breakaways rarely suceed there.
In my tour of malaysia (2.HC) I was able to make 2 wins in the breakaway. It were the 2 mountain stages. My riders were behind in the GC, joined the breakaway in the flat section of the race. Having 47 min before the last climb, it was easy to take the win with my average climber.
I gues it is fairly realistic. Why would they chase a minor rider 10 mins behind in the GC?
|
| |
|
|
| ThVoets |
Posted on 25-06-2013 09:47
|
Stagiare
Posts: 174
Joined: 11-06-2013
PCM$: 200.00
|
Housies wrote:
I am doing a career with a custom team. But I grant you the riders are continental quality. Therefore I mainly do 2.2 races.
Breakaways rarely suceed there.
In my tour of malaysia (2.HC) I was able to make 2 wins in the breakaway. It were the 2 mountain stages. My riders were behind in the GC, joined the breakaway in the flat section of the race. Having 47 min before the last climb, it was easy to take the win with my average climber.
I gues it is fairly realistic. Why would they chase a minor rider 10 mins behind in the GC?
A 47min lead is really realistic, yeah 
I was PCM Rulezz in a previous life
Winners never quit, quitters never win.
|
| |
|
|
| Kentaurus |
Posted on 25-06-2013 09:52
|
Classics Specialist
Posts: 3999
Joined: 26-07-2011
PCM$: 200.00
|
Housies wrote:
I am doing a career with a custom team. But I grant you the riders are continental quality. Therefore I mainly do 2.2 races.
Breakaways rarely suceed there.
In my tour of malaysia (2.HC) I was able to make 2 wins in the breakaway. It were the 2 mountain stages. My riders were behind in the GC, joined the breakaway in the flat section of the race. Having 47 min before the last climb, it was easy to take the win with my average climber.
I gues it is fairly realistic. Why would they chase a minor rider 10 mins behind in the GC?
Well the biggest reason, if you're only 10' back and have a 47' lead, then you going to be the race leader by 37'...
AZTECA - NBCSN  |
| |
|
|
| Popeye |
Posted on 26-06-2013 18:12
|
Under 23
Posts: 77
Joined: 03-07-2008
PCM$: 200.00
|
Thanks to everyone who participated.
The general feeling seems to be that the breakaway system/peloton AI is almost working, but there are occasions when the peloton simply do not chase hard enough and allow enormous gaps to build up.
Sound like a reasonable summary?
This has now been included in the issue list for a future patch.
thanks again for all the input. |
| |
|
|
| jseadog1 |
Posted on 26-06-2013 18:30
|
Grand Tour Champion
Posts: 9596
Joined: 18-07-2010
PCM$: 13552.00
|
I just simulated another stage where the break won by 22 minutes, it was 21 minutes last time, and the peleton never even started gaining on the break. This is unrealistic.
PCM.Daily Survivor Season 2 Fan Favorite Winner
PCM.Daily NFL Fantasy Football Champion: 2012
PCM.Daily NHL Prediction Game Champion: 2013
PCM.Daily NFL Prediction Game Champion: 2012, 2013, 2015, 2016, 2021
|
| |
|
|
| Malkael |
Posted on 26-06-2013 19:28
|
Sprinter
Posts: 1697
Joined: 02-08-2010
PCM$: 200.00
|
Yes, that seems like a reasonable overall summary of the issue. After running a lot of sample flat stages as a weak team and leaving everything up to the AI I personally found that there maybe a few varied reasons behind why the breakaway is succeeding so often in 3D Races.
Events such as: (Normal, Weak Team, No Favourites, Little to No Burden by AI to Help Chase, Repeated Each New Flat Stage Over 20 Times)
1. Breakaways can require a lot of effort to nullify or bring back, which can caused absolute carnage to the Peleton. For example, quite a few times on the Champs-Élysées stage attempted and successful breakaways were caught in time. However, the Peleton was absolutely devastated by the high pace the Peleton had to ride to catch the break and stop new breaks forming causing several splits in the Peleton that might cause people to wonder if it were secretly a selective mountain stage and not a flat sprint stage.
Even then late breakaways on the last lap of the Champs-Élysées almost won several runs of my hands-off stage playthroughs. Those breakaways weren't caught by the pack but rather by the sprinters launching their sprints and leaving everyone else behind. By that time a lot of teams were lucky if they had two riders (including their sprinter) in the front group due to the constant high pace being ridden to catch attacks.
2. Sometimes Sprinter's teams are failing to contribute to the chase either at all or sufficiently enough to help wind the breakaway back in. In several runs of the final stage of the Vuelta it was left almost purely up to Saxobank - Tinkoff to try and chase down the breakaway, Radioshack-Leopard did near the end offer two or three riders but they never chased hard enough to take a turn relaying at the front of the peleton ahead of the several Saxobank - Tinkoff riders who had been relaying all stage without rest.
3. This ties in to my observations in 1., but especially if only one team is left to chase the Peleton will only match and not better the speeds (and so effort?) being ridden by the breakaway. Who could blame them when they could potentially blow their entire team up due to the effort required leaving the stage win for some other team to claim.
Some of the carnage I've seen was extraordinary enough, as seen in my Champs-Élysées example, with multiple teams helping. Assuming a single team could even catch the breakaway in my trial tests what is to stop another one from going away again and winning if that same team is still left to chase?
3. Weirdly, the Peleton actually managed to get nearer to, or chase down, the breakaway more often on Hard difficulty compared to Normal difficulty in my repeated tests.
4. Related to 3., the breakaway are able to match the efforts of the chasing Peleton and maintain their gap through to the end a bit too easily in many a test. In theory, as seen in real life, the Peleton should wear down the break through the weight of numbers. Four riders relaying at 80 should tire faster then eight riders relaying at 80 due to a greater of number of riders to share the exhaustion caused by the high effort.
I would be inclined to say that potentially the breakaway is not tiring enough after being out front for so long considering my experiences. Or it may be the reverse and the AI peleton needs to be programmed to commit more riders to share the burden, introduce a higher effort earlier, or both. Since at the moment too often the Peleton is lucky if it catches the breakaway within five in-game kilometers of the finish. Let alone the 10km or 20km in-game kilometers it arguably should often catch the breakaway at in flat sprint stages to draw a parallel with real life and previous PCM games.
Those are my observations so far after conducting many a repeated test, which I reserve the right to modify as I continue to conduct experiments should I come to a different hypothesis. |
| |
|
