I've put together an excel file that generates a list of riders to create a fantasy database.
Instructions:
- Press the F9 key and wait for the file to finish calculating to create a new list of riders
- The sheet titled StatsDB contains the basic database so you can filter it and look at the rider stats
Sheet CopyDB contains the data formatted correctly to copy into the excel editor
- Don't go to the other sheets unless you want to break something (or change the formulas, but good luck working out my mess of sheets ).
Current Features: (To be updated)
- Generates a realistic list of over 3000 riders
- Names, Ages, Birthdates, Nationalities, Potentials and Stats for each rider
- National Champions
- Teams
---
Currently it works pretty well, but it'd be great to get some feedback on this to hopefully improve it, Some points I'd like to discuss:
- Are TT/PRL related enough? Particularly for prologue specialists
- Should there be more stage racers with TT stats over 76
- Should there be higher fighter and downhill stats for random riders with lower main stats
- Should sprinters have higher PRL Stats
- Should riders from obscure nations not appear/appear much less often?
- Any other points
If there's interest I plan to see if it's possible to add potentials, national champions and possibly teams in the future, as well as any other suggestions.
Credits to Cyanide for the rider names, the EP for data on ages and nationalities and the few people who have posted similar things in the past (although I find those are outdated and not on the same scale, so I don't feel this is the same thing).
Small update, young riders are now less good, punchy sprinters have better hill stats, stage racers have better TT stats and proportion of nationalities have been updated (more Slovakian, Czech and Croatian riders, less New Zealand, Danish, Rwandan, Guatemalan and Hong Kong riders).
Aren't hill stats for climbers too low? You also may want to raise stamina for them and have more variety for REC - as in, bigger gaps, and in general, REC is far too low to be the top stat for it.
Last but not least, I don't see enough correlation between TTR/PRL and FLA.
Puncheurs:
Spoiler
I don't like that on average their RES is a lot lower than climbers'. In PCM 16, that is definitely going to make them not as good as they should in hilly classics in comparison to actual climbers, especially if their MO is lower than 70.
If the formulas produce a top punchy sprinter, I never expect their ACC to be lower than their SPR, and certainly don't want it to be lower than 70!
Time Trialists:
Spoiler
In only one case the prologue stat was higher than the TT stat among the very top riders; shouldn't happen. When it is lower, most of the time it's also quite a lot lower than what I'd deem realistic. In general, the differences between TT and PRL are too high. You'll hardly ever see riders with more than a 5 points difference between those 2 - while it's common the other way around - but in this case it happens on a regular basis for all rider types.
Cobblers:
Spoiler
Not sure I like the fact REC is generally too high for all of them, not like they need it. Also more cobblers should also be good sprinters. In that specific case, REC should also be higher.
Sprinters:
Spoiler
On average their TT stats are higher than their PRL stats. While the PRL stats are pretty much spot on, it's their time trialling that is too high.
81 sp 74 acc shouldn't happen irl, take the MG, those riders suck! Pretty much, their ACC is too low. Only a few lucky riders had their ACC higher than their SP in my case when it came to top sprinters, while a lot more had a lot lower ACC than their SP. Of course, variety is very welcome. It's of course nice to see sprinters with e.g. 80 sp 83 acc or types with 82 sp and 78 acc, but they shouldn't be as common.
Other:
Spoiler
Possibly raise the amount of riders with more than 82 FTR and make some riders that do not have more than 76 AVG get more than 80, and do the same for downhill, especially for riders from Belgium, Netherlands, Czech Republic and France, which we'll imagine are the cyclocross riders coming to the road.
The amount of riders from small nations is too high, you may want to adjust that part of the sheet basing your info on sta_region, gene_i_weight. Actually, not sure if you took a look at sta_region for the percentages of specialities riders from certain nations should have. If you didn't, check that out!
I thought I'd write a lot more, take it as a compliment. Seriously very cool project!
Edited by matt17br on 21-10-2016 21:07
Thanks for all the feedback Matt Here's an updated file with stats improvements based off your suggestions, as for the nationalities I've made it a bit less likely to get obscure countries and plan to update it later with a better system with weightings to certain rider types, etc. For now I also added charac_limit columns on the Copy sheet to make copying it into an editor easier.
Looks quite a bit better now! Seems like you didn't wait for excel to recalculate the stats and names of the riders in the screen you posted though, as only nationalities might have changed from an earlier batch of riders. That happened to me too yesterday when I didn't wait for the calculation to complete and for instance scrolled down with my mouse.
Though a few points still:
- The mo-hi difference is now generally lower for top climbers but I feel there's still too many cases where the difference is more than 6/7 points or so. In general though, the difference is good like it is because climbers are OP in PCM 16. Just make sure to lower the cases where the difference is more than 6.
- RES differences between time trialists are only just a tad too big. I feel a bit lower would be right spot on.
- Too many good time trialists among cobblers! And same for prologue.
- I like the amount of sprinters among top cobblers has been raised, I like how many cobblers are good sprinters, but I feel there should be more great sprinters among them. See Kristoff, Degenkolb, Sagan, Boasson Hagen etc. nowadays.
- RES should be drastically raised for cobblers. It's rare to see > 70 RES riders among > 76 cobblers, it'd actually be preferable to see cobblers have the best RES on average.
- Too many top sprinters are decent cobblers. This might be a contradiction to what I wrote earlier, but I think the amount of great sprinters among top cobblers should be raised, and not the other way around.
- I still think you should raise sprinters' ACC by 1/2 points on average. Still see too many sprinters with worse acc than sp.
- You might want to increase sprinters' STA variety in order to create more Guardini or Mareczko types.
- In general, while it might well be a coincidence, there seems to be riders that are going to be too good for the competition in their field, while nowadays there's no sprinter or no climber who are going to win all their goals races. Take Medved among climbers, he has the best FLA, MO, HI, DH, STA and RES in the top 12, as well as one of the best TTs and PRLs. A rider like that could well wreck the opposition in hilly classics given how PCM 16 works. Take Fervenza for sprinters too, best SP, ACC, FLA, 3rd best HI even, solid enough backups - talking about top 12 sprinters again. That said it's true that it's always recommended to make small edits to these riders if you are to release a db based off this excel sheet and I'm not sure at all if you can actually influence it.
Great improvements in all the other areas I mentioned.
Thanks Matt I've updated it again according to your suggestions, as well as fixing some bugs like riders being aged 2016 (thanks fjhoekie) and the nations not on the Copy sheet.
Moving back to nationalities, I've now added in weightings so riders from certain countries are more likely to be certain disciplines, for example riders from Flanders are likely to be cobbles riders, Colombian riders are more likely to be climbers, etc. I also added it generating a few CX riders as that was a very nice suggestion
The nationalities might still not be perfect though, so suggestions on what to change with the weightings and amount of riders from each country are welcome.
Next I plan on hopefully adding preferred weathers/temperatures, heights, weights and potentials. By any chance does anybody know which table in the db editor contains the different weathers? Edited by trekbmc on 23-10-2016 21:19
Wow very, very cool project, kudos to you. I actually often think that playing with unknown names would be more fun since it adds some unpredictability. Can this be also used for PCM15? I guess so, although the stats may not work exactly as intended.
"It’s a little bit scary when Contador attacks." - Tommy V
Shonak wrote:
Wow very, very cool project, kudos to you. I actually often think that playing with unknown names would be more fun since it adds some unpredictability. Can this be also used for PCM15?
Thanks mate, can definitely be used for PCM15, I've actually been testing it in PCM15 due to that being a bit easier for me.
There may be a slight issue with the game handling res stats differently, honestly not sure but maybe I could make a version that handles them differently for PCM15, shouldn't be a problem for the game afaik though.
As for getting the db in the game, go to the sheet CopyDB on the file, and open the excel editor (I recommend using the EP as a base, but Official Release will work to), then copy the rows from the file into the corresponding rows in the editor (make sure that when the row has copied it doesn't have a link to the other file but the data itself, I usually copy into an empty post in daily and then across to the excel editor so this doesn't break) and delete any excess riders from the base db at the bottom of the table.
Once you've copied the data across to the excel editor, you'll need to do the teams and potentials manually (to save time on this maybe just play with random potentials, until I add potentials to the file). Once you have done that export the db and you should be good.
I feel like there is way too many cyclists generated with high flat (>75) and abysmal TT and backup stats (often even below 60). There should be more correlation between FL and TT, RES stats. Also, I don't think the difference between SPR and PRL stats should be more than 10 so often.
Overall I think there are still too many good riders from obscure nations. If I filter just nations for which I cannot name a cyclist from, I usually get around 10-15 with average above 75, which I think is slightly too many.
Another thing I noticed is that most of young riders have fairly high FL stat. Is that intentional? Also there are very few (almost none?) climbers with COB stat above 65. I think those climbers who have high FL, TT stats should also have higher chance for COB stat around 65-70.
Regarding the regional bias for rider types, I think it should be even stronger (slightly). Especially regarding such obvious cases like you mentioned. Other biases which come to mind (not sure if you implemented those) are
- Time trial focus in Australia, Denmark, UK
- Stage races/TT focus in USA
- Climber focus in Austria, Eritrea, Iran (there should be more riders generated from Iran too I reckon), Spain, Venezuela
- Flat/TT focus in Baltic countries, Finland, Ukraine, Sweden
trekbmc wrote:...make sure that when the row has copied it doesn't have a link to the other file but the data itself, I usually copy into an empty post in daily and then across to the excel editor so this doesn't break...
Thanks, might try an career with this feature Only first complaint I can see is the high MO/Hill ratio where several riders get 80+ in both stats. I can change that manually, so not a huge problem
I took another approach to looking at this. So I adapted a comparison file I created for my fantasy DB (see 2013 DB downloads). File is attached and you can use it if don't already have something like this. Unfortunately I don't have 2016 so I compared to 2015 which might cause some noise.
Your DB seemed pretty good in general. My specific observations:
1. Most of your stats are a little higher than the comparison on average particularly DH and CO, with an average 2 to 3 points above the 2015 DB (although maybe 2016 runs higher). Connected to this you have from 2% to 20% more riders over 70 in each stat.
2. The correlation between Hill and Mountain doesn't seem strong enough nor that between prologue and sprint. Prologe and TT looks good.
3. In the source DB a lot of riders are very bad in CO and MT with the game stats showing two peaks one in the mid-fifties and the other in the mid-sixties. You have more of a normal distribution. Not sure how much this matters since it mainly impacts riders with lower stats.
On your weather question, not sure if there is a table in the DB but if you compare the values in the DB to what displays in game I think you can figure out what they mean. When i looked at it in 2013 and 2014 there were only 7 values for each.
If you want to use my file:
I linked the fantasy DB tab to your file (if you update the links on this tab to link your live file it will update the comparison when you recalculate with them both open) and copied the comparison DB tab from the 2015 PCM DB opened in Lachi's excel editor. You can copy and paste from a 2016 DYN Cyclists to the "From PCM DB" tab to update. The rest of the file will update off that.
The two "summ" tabs show the profile of the stats. The Rider Comparison tab compares these two, probably the best stuff is summary stats at the bottom.
I used Ulrich's comparison file to compare it with the EP 2016 database. There are obvious differences because of the cut off at 60 we adapted, but some other features are possible to compare.
The averages, medians and deviations are closer than with the 2015 db. Still a tad higher (most notably DO and PRL stats, which are minor stats anyway and somewhat random, i.e. with high uncertainty).
The most interesting to me are these facts:
- Your number of FL values is consistently higher in the 70's range by about 1.5% (~500 riders), meaning that you have about twice as many (~35% as opposed to 17%) riders with FL>70.
- Profile of MO values is pretty much spot on, but the correlation of HIL and MO stats is lower than in the EP.
- Similar to what Ulrich saw when comparing with the 2015 database, correlation of TT and PRL is very good but the correlation of PRL and SPR stats is way lower in your case. This is something that should be addressed imo.
- You have quite a few riders with very low recovery. This might not be a problem, unless there are higher profile riders among them.
- As I noticed before already, the correlation between FL and TT is much lower than it should be imo.
- I realise it might not be easy to find the right balance, but there are other connections between stats like SPR/ACC and REC/MO, which are significantly weaker in your generated db compared to EP 2016.
Thanks a lot for all the suggestions guys, especially the comparison file is really useful.
I've updated it with these suggestions, the changes made are:
- Less high flat riders (turns out there was a typo in one of the formulas causing that)
- Hill/Mo and SP/PRL correlate better now
- For time-trialists and climbers/GC riders there is a better Flat/TT correlation (for other rider types I'm not sure there needs to be too much of a correlation, for example in the EP most cobblers have high flat but not that high TT).
- OP riders who beat the field in mountains and hills are less likely (although Valverde types are still possible, just they'll be unlikely to win GTs)
- Finally, I've added in potentials, they are currently very rough since I don't exactly understand how the game handles them or what is reasonable, so it'd be great if I could get an opinion on them.
).
