I had intended to run a little minigame as an offseason distraction and test of PCM24, with the mangame pelotonn but time got the better of me.
I have done and will continue to do some PCM24 testing. I started by looking at how the Medium Mountain stat might be assigned. To do that I did some analysis of the PCM Daily 2024 DB. I figured they knew what they were doing and have had a year to test since I believe it was added in PCM23.
My first test was to look at what stats best predict the Medium Mountain value. I used all riders in the DB where the higher of HI or MT was greater than 72. Figuring we don’t care much about lower quality climbers. While I tried a few variants the obvious one worked best. Averaging the Hill and Mountain was the best predictor of a rider’s medium mountain value. It works so well it is my guess that is how the PCM team developed it.
There are two simple options here, you could round everyone down or round the 0.5 average riders up. The latter works a little better. The actual medium mountain average was 73.43. The table below shows the performance of the predicted medium mountain under the two scenarios.
Approach
Correlation with Actual Value
Average Calculated Med. MT
Round All Down
.9897
72.46
Round 0.5 Up
.9903
73.01
So rounding up works pretty well and I think would be a fine approach.
The predicted values are still a little lower than the actual. I looked at where that gap was and it was primarily in riders with an average of HI and MT in the 70’s. For the higher end guys and the lower end guys the base formula, with the round down, works pretty well. I came up with a third scenario which says round down, but if your average of HI+MT is between 70 and 80 then add 1. So a 74/75 HI and MT and a 73/75 HI MT would both get 75 MM but an 80/82 and a 80/83 would both get 81. This slightly improves the prediction as shown below.
Approach
Correlation
Calculated MM Average
Round All Down
.9897
72.46
Round 0.5 Up
.9903
73.01
Round Down + 70-79 adjustment
.9913
73.28
Given the small gap between the actual and predicted I think another approach that would also add a little variety would be to allow managers to add the extra points. You would do the average and round down and then give manager’s flexibility to add some extra points. Something like: you get 1 point per rider with an average between 70 and 80. But you could use them on any rider and you could use up to 2 points on 2 rider. This would push the overall average close to the predicted value but give managers a chance to influence their riders. This could be done after transfers.
In summary, 3 options I would suggest based on this:
1. MM = (HI+MT)/2 round 0.5 up
2. MM = (HI+MT)/2, round all down but add 1 where the base formula yields a value between 70 and 79
3. MM = (HI+MT)/2, round all down and give managers a point allocation to spend
My plan is to implement one of these in the 2023 Mangame DB and run some test races as a next step. Interested in thoughts on which races or how I should approach that.
Whichever method we use isn't going to create any Medium Mountain specialists until we have riders maxing using new stat gain files.
Would it be a good idea to let managers choose which stat out of Mo, MM and Hi the new stat goes to, making the replaced stat the MM if that's not chosen. E.G. if I chose the average to go to Mo, then the current Mo stat would become the MM stat instead.
Then FA's would just get MM as the average by default.
It could be incorporated into the stat gain choice PM's for the year it gets implemented as that's when the DB would be getting updated already.
Nice this testing is already done and you give us some insights, thanks for this.
Important to know again is also to understand what the attributes mean and how they work IMHO. I am sure most of us already are aware but I still wanna point this out again:
- Mountain, Hills & Medium Mountain have nothing to do it with the steepness of a climb as long as it goes uphill
- Also it only indirectly has to do with how long a climb is (see next)
- the different attributes define the speed uphill with a certain rider intensity
Mountain Stat > defines the speed/ability uphill up to around 75%-77% intensity
Medium Mountain Stat > defines the speed/ability uphill around intesity of 85%-87%
Hill Stat > defines the speed/ability uphill at 95 % intensity and more
the values there are more estimated values by me and what i could find but definitely around that. In between those values its alway a mixture of the stats.
Just a quick summary (what everyone probably already knew ) . What results out of that is that it defnitely make no sense at all when Medium Mountain is lower than both other stats. Possible but kinda weird.
Also shows that the duration of the speed while climbing is not just defined by these stats but also heavily depend on energy stats like Stamina, resistance and so on. A puncheur with high hills is useless if their res is low since he can only use his Hill stat very shortly. With the new medium Mountain stat i think its a bit easiert to plan and judge climbers/puncheurs
Thanks for doing these tests!
I also hope there will be some testing time available to find out if riders like e.g. Eastman (82Mo/71Hi), who don't seem to have an issue in the current game version (even getting stage podiums in some really hilly stages), will still be useful with such an implementation of MM (would be 76/77 depending on the scenario), or if they get pretty much useless with that new stat added.
Also, I'd appreciate to get a timeline for the switch to PCM24+ before picking stat gains. I guess the question would be if we do a "big bang" (switch to PCM24 plus new OVL), or one after the other. Obviously the new MM stat will require adaptations to the OVL formula anyway, though...
Ezeefreak wrote:
Just a quick summary (what everyone probably already knew ) . What results out of that is that it defnitely make no sense at all when Medium Mountain is lower than both other stats. Possible but kinda weird.
All of what Ezeefreak wrote is correct and the % on effort is about correct there too. As a reminder, flat also has an impact on inclines <7%. But I want to particularly highlight this point, because while it is off for MM to be lower than hill AND mountain, there are some situations where MM being ABOVE both Hi/MO makes lots of sense and I hope those are considered with new riders in particular - although I don't know how to change that for our DB.
A separate point is that I think the formula for MM could potentially include flat stat as well. For example, a rider who has FL > HI > MO is a flat specialist VS. a rider who has FL > MO > HI might be a flat rider who struggles with high intensity, but can stamp out a hard pace (think IRL Mikkel Bjerg?). With this in mind, I've thought of the following idea that might help with the variety of riders and make MM stat dependent on rider type.
If FL > HI> MO; MM = 1/3 difference from HI to MO (closer to HI) - Flat specialist
If FL > MO> HI; MM = 1/2 difference from HI to MO - diesel engine
If MO > HI > FL; MM = 1/2 difference from HI to MO - All-around climber
If MO> FL > HI; MM = 2/3 difference from HI to MO (closer to MO) - long climb specialist
If HI > FL > MO; MM = 2/3 difference from HI to MO (closer to HI) - punchy classics rider
If HI > MO > FL; MM = 1/2 difference from HI to MO - short climb specialist
I know in advance this is hard and just a starting idea without action, but I wanted to throw it down in case it helps. Having not analyzed the DB, this may not even provide enough difference to be worth. But seeing as the difference between someone getting 80MM and 79MM is over 1million in training fees, finding the right solution is important. Appreciate everyone putting time into this and happy to help however I can.
1. The PCM Daily DB follows the rule about MM never being lower than both HI and MT
2. As noted MM can be higher than both HI and MT although I think that maxed at +2. I think this makes sense, seems like a guy who is good at the medium pace can't be that bad at the others. So I am not sure about "Medium Mountain Specialists". If you define that as 74/76/74 then I get that, but would struggle with something more exaggerated.
3. While I agree we will need to wait for stat gains over time to create variety my options 2 and 3 could both create riders with a MM above the higher of HI or MT in some cases.
4. Should be able to test Sean's scenarios against the PCM daily DB without too much trouble. Will look to do that.
2. As noted MM can be higher than both HI and MT although I think that maxed at +2. I think this makes sense, seems like a guy who is good at the medium pace can't be that bad at the others. So I am not sure about "Medium Mountain Specialists". If you define that as 74/76/74 then I get that, but would struggle with something more exaggerated.
3. While I agree we will need to wait for stat gains over time to create variety my options 2 and 3 could both create riders with a MM above the higher of HI or MT in some cases.
One quick note on this: There are a handful of exceptions to the +2 rule for MM, most prominently Finn Fisher-Black who's at 72/76/72 Mo/MM/Hi. But that's indeed very rare.
What is relatively more frequent are riders who have MM very close to one of Hill or Mountain but with a significant difference to the other. Examples of this at the higher levels include Kuss at 82/81/75 or Sosa at 77/79/73, as well as Pidcock at 74/79/79 or Lafay at 73/78/77.
One thought I had, which goes into a somewhat similar direction to what sammy and sean proposed, was to create the possibility of adjusting/decreasing the higher of the two stats to allow for such a distribution. So if a rider has e.g. 79/72 Mo/Hi, instead of going to 79/76/72, you could decrease Mo and increase MM by the same amount to arrive at 77/78/72. And maybe allow something like this to be used once per team or something like that.
Also, as a quick reminder on stamina and resistance, which Ezee already alluded to: Stamina defines how quickly the yellow bar decreases on efforts between 65 and 85%, resistance defines how quickly the red bar decreases on efforts of >85% (and also how quickly they fill back up). But both only come into play as the green bar decreases, which as far as I know is still determined by main stats.
Also, as a quick reminder on stamina and resistance, which Ezee already alluded to: Stamina defines how quickly the yellow bar decreases on efforts between 65 and 85%, resistance defines how quickly the red bar decreases on efforts of >85% (and also how quickly they fill back up). But both only come into play as the green bar decreases, which as far as I know is still determined by main stats.
So what then determines the yellow and red bars when the green bar is full?
redordead wrote:
So what then determines the yellow and red bars when the green bar is full?
AFAIK, nothing.
Two things I want to address:
1) Currently, PCM24 fucked up sprints so bad that the only solution was to basically remove ACC influence entirely.
2) If climbing skill is differentiated into three stats instead of two, should training those become cheaper since it will impact overall performance less?
2) If climbing skill is differentiated into three stats instead of two, should training those become cheaper since it will impact overall performance less?
Or we keep stat gains and training like it is (so we can only train MO and HI in the future) and Medium Mountain gets newly calculated with a formula like e.g. (Mo+HI)/2 after every stat gains / training season?
redordead wrote:
So what then determines the yellow and red bars when the green bar is full?
AFAIK, nothing.
That is my understanding as well. So, at the start of a race, if two riders ride at 80 effort, their yellow bars will decrease at exactly the same rate not matter their stamina. However, the rider with the worse main stats will have to ride a higher effort to keep up with the stronger rider, gradually decreasing the green bar, making stamina and resistance more important as the race progresses. In contrast, if the rider with worse main stats has less green bar later on but higher sta/res stats, that can be an equalizer, as I guess it should be.
This is the sprinter who is in the wheel of his lead-out man and who falls slightly behind when starting his sprint.
...
The cause is essentially a problem from the "Acceleration" attribute and how it's utilized in the engine.
When you are well protected in the wheel of your pilot and he still has red energy, and suddenly you start your sprint: you have kind of a double penalty:
- Wind effect: sprinter loses wind protection.
- ACC effect: sprinter takes some time to reach max power (depending on ACC's value).
So while your pilot is at maximum or nearly maximum power, you have your sprinter far from his maximum value AND penalized by the wind : this is enough for him to lose a few meters to his pilot, instead of passing him directly as you would expect! The lower the ACC value is for the sprinter, the more we see the problem.
Regarding the Wind effect, not many things to say, it is the desired behavior. Regarding ACC effect, the actual timing to reach maximum power is probably too long and it's the heart of the problem here.
What we’re working on actually is to diminish the time to reach this maximum power, and it has a clear positive effect on that problem.
First minor side-effect is that the ACC attribute will be less important, but it could be acceptable (and we can say it was maybe too important).
They also mention that the root cause of this issue had been present for years, it just hadn't been as visible due to other factors.
cunego59 wrote:
They also mention that the root cause of this issue had been present for years, it just hadn't been as visible due to other factors.
While I don't see any reason for them to lie, I want to illustrate my point to show how far removed from reality it can get now (and even include SP being largely irrelevant too):
In my current career, my main sprinter has 76sp/73acc, and my main climber 57sp/60acc.
If I set my climber to follow in the slipstream, and I start the sprint at a reasonable point, my climber has no issue following right behind my sprinter all the way to the line.
Yes, there is a higher bar usage, but depending on the pre-sprint situation that can be offset to the point that I can use my main sprinter as a leadout, and get my GC rider in the Top 5 of a .Pro sprint every now and then even at my ridiculous difficulty setting of 1.25...
Now this is obviously less of an issue for AI vs AI racing as it won't exploit the system like this, but frankly it requires just as much scrutinty as MM imo.
2) If climbing skill is differentiated into three stats instead of two, should training those become cheaper since it will impact overall performance less?
2) If climbing skill is differentiated into three stats instead of two, should training those become cheaper since it will impact overall performance less?
Or we keep stat gains and training like it is (so we can only train MO and HI in the future) and Medium Mountain gets newly calculated with a formula like e.g. (Mo+HI)/2 after every stat gains / training season?
I kind of like this idea, basically you get 0.5 MM for every point of HI or MT, you might just have to track the half points. You could then extend this concept to PRL when you train TT or SP or pair stamina and res to make training them more attractive. If you didn't increase budget this might pull money away from HI and MT if other strategies become more attractive.
Also, as a quick reminder on stamina and resistance, which Ezee already alluded to: Stamina defines how quickly the yellow bar decreases on efforts between 65 and 85%, resistance defines how quickly the red bar decreases on efforts of >85% (and also how quickly they fill back up). But both only come into play as the green bar decreases, which as far as I know is still determined by main stats.
So in a relatively easy stage with a sprint finish or a single climb stamina doesn't matter much because you the riders go from easy effort right to max effort, but is a classic with lots of moves and counter moves the stamina comes into play since more of the race is a moderate to high effort levels. is that the right way to think about it?
My thinking on the versions is that we target the change for the 2026 mangame season, by which time maybe we can use PCM25 which might fix the sprint bug?
) . What results out of that is that it defnitely make no sense at all when Medium Mountain is lower than both other stats. Possible but kinda weird.
