|
2015 Database v4: New OVL rating
|
|
| Ad Bot |
Posted on 25-11-2024 22:59
|
Bot Agent
Posts: Countless
Joined: 23.11.09
|
|
| IP: None |
|
|
| SportingNonsense |
Posted on 24-04-2015 11:10
|
Team Manager
Posts: 33046
Joined: 08-03-2007
PCM$: 200.00
|
* As is an early release, we should all keep an eye out for errors to report - trying to compile a short list to submit. e.g. One thing I am particularly noticing thus far is a lot of riders who should have extra XP but wouldn't have gained a level seem to be missing the XP increase. I will come back with as full a list as I can see, assume it is helpful if everyone keeps an eye out and reports here.
It was something I initially overlooked. I thought I had added it all in, but perhaps I didn't save it. Don't bother creating a list of such riders though - I can double check later that the XP levels are right. It will just be a case of copying and pasting from one file to another, for me to fix it.
But yes, any other errors that anyone spots are welcome.
|
| |
|
|
| Ollfardh |
Posted on 24-04-2015 11:52
|
World Champion
Posts: 14563
Joined: 08-08-2011
PCM$: 9100.00
|
Well, the Cerny error I mentioned earlier, not sure if you saw it or not.
And I hate how my talents are looking compared to real life. Stuyven, Theuns, Debusschere, Verhelst, Ruffoni, they're all so disappointing 
Changed my sig, this was getting absurd.
|
| |
|
|
| Kami |
Posted on 24-04-2015 12:25
|
Classics Specialist
Posts: 3485
Joined: 19-06-2009
PCM$: 200.00
|
For some general feedback:
1. Hilly riders
Mountain and sprint are pretty vital stats for a puncheur due to MO/HI rating in some hilly races, and hilly races generally ending in a small group sprint. It seems like both these stats are undervalued for puncheurs. When testing this out by increasing/decreasing the stats of the same rider (Ulissi):
| Flat | Mountain | Hill | TT | Stamina | Resistance | Recovery | Cobbles | Sprint | Acceleration | Fighter | Downhill | Column1 | | 76 | 76 | 80 | 62 | 83 | 71 | 72 | 51 | 80 | 75 | 75 | 63 | 62 | | 76 | 56 | 80 | 62 | 83 | 71 | 72 | 51 | 66 | 75 | 75 | 63 | 62 |
The first rider has 20 more mountains stats and has 14 more sprint stats, yet their averages are 79.90 & 79.85.
2. Climbers/GC riders
I don't really know what happened with these riders, but to say it bluntly, they don't really make sense.
| Last name | Flat | Mountain | Hill | TT | Stamina | Resistance | Recovery | Cobbles | Sprint | Acceleration | Fighter | Downhill | Column1 | OVL | | Schleck | 71 | 85 | 50 | 50 | 77 | 82 | 82 | 53 | 62 | 79 | 70 | 64 | 73 | 84,13 | | Schleck | 71 | 85 | 75 | 74 | 77 | 82 | 82 | 53 | 62 | 79 | 70 | 64 | 73 | 84,13 | | Spilak | 72 | 85 | 50 | 50 | 80 | 81 | 78 | 70 | 67 | 74 | 71 | 68 | 77 | 83,82 | | Spilak | 72 | 85 | 80 | 77 | 80 | 81 | 78 | 70 | 67 | 74 | 71 | 68 | 77 | 83,91 | | Phinney | 73 | 82 | 50 | 50 | 77 | 74 | 76 | 58 | 72 | 75 | 68 | 71 | 83 | 80,63 | | Phinney | 73 | 82 | 77 | 83 | 77 | 74 | 76 | 58 | 72 | 75 | 68 | 71 | 83 | 82,17 |
Changing hill & TT stats has almost no influence on the overall unless it's the highest stat (Phinney).
Could anyone confirm these things in their DB, so it's not just my excel messing up. (I tried it multiple times just to make sure, but the results still seem strange).
Edited by Kami on 24-04-2015 12:30
|
| |
|
|
| Avin Wargunnson |
Posted on 24-04-2015 12:38
|
World Champion
Posts: 14236
Joined: 20-06-2011
PCM$: 300.00
|
It seems logical, as SN spoke about lowering the importance of "secondary" stats and maximized the importance of the "main" stat. When you look on the table in the first post, everything bar the main stat has minimal importance for calculating the overall stat.
I sense that as reaction to PCM fucking up the secondary stats, which were close to useless in PCM13, but i dont see it as viable solution for this "problem".
|
| |
|
|
| cio93 |
Posted on 24-04-2015 12:39
|
World Champion
Posts: 10845
Joined: 29-10-2007
PCM$: 500.00
|
That is quite obviously due to the nature of the average formula, which after the first main stat doesn't distinguish between important and secondary stats.
So if enough secondary stats are pretty high, the remaining important stats are irrelevant for the calculation.
That has always been the case though, just a bit less obvious because we didn't factor in resistance last year.
|
| |
|
|
| ggDonovan |
Posted on 24-04-2015 12:55
|
Breakaway Specialist
Posts: 897
Joined: 08-08-2011
PCM$: 200.00
|
Another feedback, it's about my feeling of a general inflation on the average:
DB 2014, average of AVG/OV of contracted riders in 2014
Calculated w/ AVG: 71.63
Calculated w/ OV 72.79
Dif: 1,16
DB 2015, average of AVG/OV of contracted riders in 2014
Calculated w/ AVG: 72.00
Calculated w/ OV: 73.28
Dif: 1.28
% Increase AVG from 14 to 15: 0,517%
% Increase OV from 14 to 15: 0,673%
I think there should be a tendency of keeping the two averages (AVG/OV) as close as possible so there is an equal impact while going to the renewals and allocating the RD.
|
| |
|
|
| Gustavovskiy |
Posted on 24-04-2015 13:03
|
Team Leader
Posts: 6036
Joined: 20-07-2008
PCM$: 200.00
|
Perhaps we could add a few exceptions to the formula like MON + TT, (MON or HIL) + SPR, COB + SPR. Add a bit to the coefficient of the 2nd best value in this cases so it differentiates them from pure climbers/cobblers/TTists.
|
| |
|
|
| sammyt93 |
Posted on 24-04-2015 13:33
|
Classics Specialist
Posts: 3634
Joined: 03-07-2012
PCM$: 300.00
|
Currently my team isn't in the DB as all the Sauber riders are down as Aegon riders, that confused me at first until I searched Sammarinese riders and realised what had happened.
|
| |
|
|
| Heine |
Posted on 24-04-2015 13:41
|
Grand Tour Specialist
Posts: 4116
Joined: 08-04-2007
PCM$: 200.00
|
To get a perfect and "fair" average/overall you would need to have formulas based upon several factors. You would in general need them to weight differently based on type of rider. Making it extremely hard to implement. The current suggestion weakens some kind of riders while it strengthen others.
What would be great, is if the db could automatically identify rider type (stage racer, climber, puncheur, sprinter etc) and then choose avg formula based on that (and race days).
BUT SN is doing this free of charge, we can't expect everything to be perfect. Hell, even keeping the game going at the pace it does is impressive.
If it is possible to implement it would be great if it couold be based on main stat + relevant secondary stats while still giving a high overall if main stat is high. However I'm not able to come up with any good suggestion other than my thoughts here ;-)
|
| |
|
|
| Ian Butler |
Posted on 24-04-2015 13:43
|
Tour de France Champion
Posts: 21854
Joined: 01-05-2012
PCM$: 400.00
|
Nice to see quite a bit of Argentine riders in that database. Hope to get a few in the team at least 
Wouldn't know anything about the stats discussions, so nothing to say there! |
| |
|
|
| Jesleyh |
Posted on 24-04-2015 13:48
|
Tour de France Champion
Posts: 15274
Joined: 21-07-2012
PCM$: 200.00
|
Thanks for the DB, SN, will be fun to look into it.
@Avin
The secondary stats seem to have more impact in PCM 14. At least I can confirm that the RES stat is fully working now.
Feyenoord(football) and Kelderman fanboy
PCMdaily Awards: 12x nomination, 9x runner-up, 0x win.
|
| |
|
|
| CountArach |
Posted on 24-04-2015 13:49
|
Grand Tour Champion
Posts: 8290
Joined: 14-07-2008
PCM$: 200.00
|
Just building on ggDonovan's stats on the inflation of average, here's some stats:
Total riders with 75+ as their average in 2014 - 185
Total riders with 75+ as their overall in 2015 - 454
77+ average in 2014 - 71
77+ overall in 2015 - 181
79+ average in 2014 - 32
79+ overall in 2015 - 45
From there things obviously get much closer. It seems that most of the major gains come in riders who are brought into that 75-77 range. Most of the guys in there have at least one 76-77 main stat with good support stats. However, you also get lots of guys who aren't useful for anything more than domestique support who just have a main stat (there seem to be a lot of climbers particularly like this) of 76 and then relatively poor support stats (or only a halfway decent TT stat. Jackson Rodriguez (5979), Thomas Rohregger (855) and Tyler Brandt (6241) are all examples. Josh Edmondon (3103), Enrico Barbin (3258) and Marco Guillen (6490) are all examples. Now, if the fact that guys like this are going to be over 75 in their overall stat is factored in to renewals and they won't cost much more than the 60-70k that they probably deserve as a renewal then that's okay but if they are going to start demanding more than that because of their high average, then that is going to seriously screw with the market.
I don't know if you've already thought of this when it comes to renewals or even if the overall stat factors in at all, but it is something that you will have to be aware of.
It is also worth mentioning that some sprinters who only have high sprint/acc stats (like Theo Bos) seem to have very high averages. They are obviously going to be hard to fit into any ranking system like that, however.
|
| |
|
|
| Gustavovskiy |
Posted on 24-04-2015 14:08
|
Team Leader
Posts: 6036
Joined: 20-07-2008
PCM$: 200.00
|
Heine wrote:
To get a perfect and "fair" average/overall you would need to have formulas based upon several factors. You would in general need them to weight differently based on type of rider. Making it extremely hard to implement. The current suggestion weakens some kind of riders while it strengthen others.
What would be great, is if the db could automatically identify rider type (stage racer, climber, puncheur, sprinter etc) and then choose avg formula based on that (and race days).
BUT SN is doing this free of charge, we can't expect everything to be perfect. Hell, even keeping the game going at the pace it does is impressive.
If it is possible to implement it would be great if it couold be based on main stat + relevant secondary stats while still giving a high overall if main stat is high. However I'm not able to come up with any good suggestion other than my thoughts here ;-)
It's obvious we're not going to find a perfect formula, but the discussion goes only to aid SN on finding the best possible.
Also, as you suggested, SN's DB work, throughout the years has been absolutely outstanding. His expertise has been vital to the development of the game, and I think we should all cherish that, and thank him for that massive contribution.
That said, I think adding a few "IF" clauses to the formula could be doable. I don't have much time now, but could elaborate on that next Monday.
|
| |
|
|
| matt17br |
Posted on 24-04-2015 14:16
|
Directeur Sportif
Posts: 10525
Joined: 28-09-2013
PCM$: 200.00
|
Pelucchi's name is Matteo, I noticed his name is Matte in the db. Also, column1 should actually indicate prologue stat as far as I know? I noticed also that 2 riders have got 85+ in the Fighter stat. Portela already had 86 in last year's db and now has 89 but Grippo has got 86 and I was wondering if that is a typo or the result of some odd kind of training/development?
For the rest 
Goodbye real life! Looking forward to the transfer season.
Edited by matt17br on 24-04-2015 14:17
|
| |
|
|
| sgdanny |
Posted on 24-04-2015 14:17
|
Classics Specialist
Posts: 3591
Joined: 18-03-2014
PCM$: 200.00
|
Also it's Bartlomiej and not Bartlomieg Matysiak
|
| |
|
|
| fjhoekie |
Posted on 24-04-2015 14:21
|
Grand Tour Specialist
Posts: 4476
Joined: 25-07-2010
PCM$: 200.00
|
I'm not good at formulas or anything, but I do think some work is needed for a few sprinters. In my team Vaitkus (77SP, 77FL) has a significantly lower avg than Kobe Vonverschelde (80SP, 72FL). Vaitkus is way stronger in nearly all other stats, and results-wise he's far more valuable. No idea where this could be adjusted in the formula, but imo their avg ratings would be better if they were completely switched
EDIT: Also Orlov's exp doesn't seem to be added. Has definately done some racing last year 
Edited by fjhoekie on 24-04-2015 14:23
Manager of Team Popo4Ever p/b Morshynska in the PCM.Daily Man-Game |
| |
|
|
| Ollfardh |
Posted on 24-04-2015 14:39
|
World Champion
Posts: 14563
Joined: 08-08-2011
PCM$: 9100.00
|
Similar to others, I also have no exp added for Ruffoni and Verhelst.
Changed my sig, this was getting absurd.
|
| |
|
|
| SportingNonsense |
Posted on 24-04-2015 14:48
|
Team Manager
Posts: 33046
Joined: 08-03-2007
PCM$: 200.00
|
@ggDonovan @CountArach Irrelevant. Obviously the factors dependent on AVG previously will be adjusted. It doesnt matter if the top OVL value is 84 or 184. The Race Days included in the file already have been crudely adjusted to do just that.
What matters is that the general order of riders, when sorted by OVL, looks right.
Heine and Gustavovskiy have grasped the point of how the formula can be refined. It's all about Rider Types and IF statements.
@Kami The answer to your hill stat query is Di Maggio. No Mountain stat or sprint stat of note, and a lot of wins last season. The reality is that Mountain stat is not too important for the majority of hill races on the calendar, and sprint doesn't count for much in an uphill finish either. I'm not too worried about the 50 stat scenarios as they aren't a part of the game. As above, it's the order of riders that counts.
Edited by SportingNonsense on 24-04-2015 14:49
|
| |
|
|
| ggDonovan |
Posted on 24-04-2015 15:33
|
Breakaway Specialist
Posts: 897
Joined: 08-08-2011
PCM$: 200.00
|
SportingNonsense wrote:
@ggDonovan @CountArach Irrelevant. Obviously the factors dependent on AVG previously will be adjusted. It doesnt matter if the top OVL value is 84 or 184. The Race Days included in the file already have been crudely adjusted to do just that.
OK
SportingNonsense wrote:Heine and Gustavovskiy have grasped the point of how the formula can be refined. It's all about Rider Types and IF statements.
PCM categorizes the riders in 7 specializations:
- Fighter
- Puncheur
- Stage Races
- Sprinter
- Time Trial
- Northern Classics
- Climber
I've for example, for my personal use (can be shared, it's not top secret), formulas for every rider using this specializations.
Then I use a simple rule to sort the overall quality of the riders that is [1st best spec. + (1/2)*2nd best spec. + (1/3)* 3rd best spec.]/[1+(1/2)+(1/3)]. I think is difficult to have a rider good at more than 3 specialization.
This could be a way to improve the formula.
Edited by ggDonovan on 24-04-2015 15:45
|
| |
|
|
| Kami |
Posted on 24-04-2015 15:58
|
Classics Specialist
Posts: 3485
Joined: 19-06-2009
PCM$: 200.00
|
@SN
Fair point about the hilly stats and when looking at the OVL of the hilly riders, it looks pretty on point.
About the climbers/GC riders, i used 50 to show the extreme just for the general idea. Like you said earlier, when you order riders by OVL, Schleck's higher RES/REC shouldn't outweigh Spilak's TT/HI adventage when looking at the riders OVL.
Some other examples: Intxausti & Henao Montoya have almost the same overall (0.05 difference), while Henao has -6 in TT and the same MO/HI.
Dekker & Taaramae: Dekker has the higher OVL while Taaramae has the higher HI/TT.
There are more examples but it comes down that the OVL stat doesn't represent the GC riders strength very well. Wouldn't adding statweight to HI/TT when MO is the highest stat help with this even when the supporting stats are higher then HI/TT.
|
| |
|
