| Author |
Topic: More statistics please! |
1165 replies
|
|
|
|
Most supporter credits spent?
|
|
|
|
Quote ( Tomek Kiełpiński @ August 24th 2018,17:00:42 ) Most supporter credits spent?
I risk to say that's Stu Foster :D
|
|
|
|
Quote ( António Pereira @ August 24th 2018,16:49:59 ) #Most races done as a GPRO Supporter - just to see the ones who were the great contributors (€) for GPRO development
Quote ( Tomek Kiełpiński @ August 24th 2018,17:00:42 ) Most supporter credits spent?
Well, sort of.
It will be more polite to put "More races done as a GPRO Supporter" . which is slightly different from "most supporter credits spent" - doesnt include liveries bought, etc...
but can be that too...
|
|
|
|
I've got a suggestion and that is the amount of career testing laps completed on each manager's profile page.
We've already got this on the racing resume, but I'm just too lazy to count that all up. Should also make for interesting comparisons ;)
|
|
|
|
|
Maybe a statistic showing who has had the same driver for the longest uninterrupted time would be nice?
|
|
|
|
- We need a Unique General Rank Number from all our presentation,
...like Chess.!!!
(easily you can define all the rules & parameters to create it)
...it will be very useful to many things.
:)
|
|
|
|
Can we have some statistics on the weather in Beautiful Sithonia, Chalkidiki please?
I feel I lost track of those stats.
A notification would be useful too
|
|
|
|
Quote ( Ioannis Kalogirou @ September 24th 2018,16:15:13 ) - We need a Unique General Rank Number from all our presentation,
...like Chess.!!!
At risk of being called a simpleton for not immediately agreeing with you, here goes: chess' ELO rating system is a relative, self-correcting rank based on a very tiny set of finite conditions: win, draw, lose.
Quote ( Ioannis Kalogirou @ September 24th 2018,16:15:13 ) (easily you can define all the rules & parameters to create it)
As such, i hardly think it's essy to implement in GPRO's ecosystem, where (a) many of the managers never race against each other, and (b) the concepts of winning and losing are much much more complex.
Do tell how easy such a thing would be (i think it won't).
Quote ( Ioannis Kalogirou @ September 24th 2018,16:15:13 ) it will be very useful to many things.
Also very curious what useful-ness there is in such ranking.
|
|
|
|
ELO-like rating systems are used in multiple players competitions as well. The two most common ways are:
1) treat a 40-people race as 40*39/2 duels and every person gets 39 wins/losses from a single race.
or
2) Compute the expected place and compare it to the real one. [The ELO changes are defined for win,loss,draw but are based entirely on the difference of expected score against the other player (which can be any number between 0 and 1) and the real score. Extending it to more outcomes is not a problem.]
|
|
|
|
|
Élő (Árpád) point system already at FIFA rankings. :)
|
|
|
|
Quote ( Ioannis Kalogirou @ September 24th 2018,16:15:13 ) We need a Unique General Rank Number from all our presentation,
...like Chess.!!!
Yes!!!
Quote ( Miel Soeterbroek @ September 24th 2018,17:49:01 ) i hardly think it's essy to implement in GPRO's ecosystem, where (a) many of the managers never race against each other, and (b) the concepts of winning and losing are much much more complex.
Absolutely, but who cares? It will be arbitrary, pointless, and fun! :-D
|
|
|
|
- We need on all existing Lists to have (After the visual Part) Like our Name and position,
...all our Team Members Names and Positions at a Row.!!! (with Green as usual)
...it's Easy & Simple.-
:) :) :)
|
|
|
|
next time like see which manager have 100% safe promotion in next seasson if not have a debt
we know it is one first rules which read newbies in FAQ and mentor program.
It can be gold star symbol .
In other way what think about add new special achievment - win race summary in all group for exaple give there Pro /gb/RaceSummary.asp?group=Pro&Season=66&Race=14 - link for last race Katerina Kakni win this race and be also first from all pro groups, think it can be fine achievment ... like see your comments
|
|
|
|
Quote ( Krzysztof Wnuk @ September 26th 2018,09:29:07 ) like see your comments
Good idea. Already introduced ;-)
/gb/Stats.asp?type=achievers&aid=316#scroll
Quote ( Krzysztof Wnuk @ September 26th 2018,09:29:07 ) next time like see which manager have 100% safe promotion in next seasson if not have a debt
nah
|
|
|
|
Quote ( Ioannis Kalogirou @ September 26th 2018,07:16:00 ) - We need on all existing Lists to have (After the visual Part) Like our Name and position,
...all our Team Members Names and Positions at a Row.!!! (with Green as usual)
And friends. And all Elite. No limits! I've got a wide screen! :-)
|
|
|
|
|
Well there definitely should be a page where you can see which teams all of your group mates are in. It's a basic thing to have and it's relevant because it's good to know if you are racing against team mates.
If it's hard to fit in an existing page, it could be an all new one which is in the same category as "Money and car levels" or "Managers' staff".
Alternatively, it could be a new filter in the Team standings: "Show only teams with managers in your group". That implies a big limitation though, because you would only be able to filter by your own group and not be able see which teams are represented in, say, Elite.
|
|
|
|
Quote ( Ioannis Kalogirou @ September 24th 2018,16:15:13 ) We need a Unique General Rank Number from all our presentation
This was coded about 12-13 years ago, but was never published:
/Stats.asp?type=genrank#scroll
I don't even know how it works :)
|
|
|
|
Quote ( Vladimir Alexandrov @ September 26th 2018,10:51:18 ) I don't even know how it works :)
It does work (no matter how) but is a bit outdated :-)
|
|
|
|
Quote ( Tomek Kiełpiński @ September 26th 2018,10:02:57 ) Good idea. Already introduced ;-)
/gb/Stats.asp?type=achievers&aid=316#scroll
OK not see this :)
|
|
|
|
|
I'm assuming that this ranking stopped calculating after a certain season? Considering there are only 7300 people on it, the guy in last didn't do any races, and Roland isn't even on it :)
|
|
|
|
|
Maybe it got broken at some point. So the data there is pretty useless.
|
|
|
|
I reckon something like taking the avg score in each tier, then multiplying it with some sort of tier factor (elite being 1, and reducing for lower tiers), adding them up then multiplying by a highest position attained (1st in Elite = 1st, 1st in Master = 41st, etc) number. Plus some sort of bonus for wins in groups or championships or whatever.
Simple. I accept payment in cash or hot races ;)
|
|
|
|
|
|
Quote ( Vladimir Alexandrov @ September 26th 2018,10:51:18 ) I don't even know how it works :)
- First thanks Vlad for your Interesting.
...Now about the Rank Number my Idea is to have a flexible Number who moves Up-Down after Every Race. To create that first we must Define the Parameters to calculate. On Snooker is the results of 2 Years for example...
Parameters:
A. The Period who we Use our personal Results. That Period it must be Sort. My opinion is that 5 season Results are ok. (5 last seasons + the current = 6 in real)
B. We must define the Difficulty of Wins, Podiums, Points, Faster Lap, Poles and Others...
...from category to category. ...for example if Define a Win in Rookies 3 points on AMA must be 30 or more.!!! 10 times Higher (maybe 15 times) ...the same about win on PRO, Master Elite. The same and for all others Parameters. Of course maybe we need more Parameters about other achievements.
...after that easy we have a Number and the position on our Profile and see after every Race how he moves.
- Now that Number can Help to make all groups more Equals and fair, because easily you can put the Strong and weak managers who promoted & relegated on groups.!!!
...the sum of 40 Rank Numbers of a Group Show haw strong it is and after every season reset will be more Equals.!!!
...that's the fist thoughts about the theme who has a lot more to Thing.
...I can't involve more because my English as you Know are terrible...
:) :) :)
|
|
|
|
Quote ( Ania Piekarska @ September 24th 2018,18:14:21 ) ELO-like rating systems are used in multiple players competitions as well. The two most common ways are:
1) treat a 40-people race as 40*39/2 duels and every person gets 39 wins/losses from a single race.
There is a rating system like this in Deluxe Ski Jump's online game. (Could be somewhere else too.) It would be nice to count it from the start and have the peak ratings too, but there are data missing from some early races and it might need quite a lot of calculations.
|
|
|
|
There are several ways to create a ranking, not necessarily an ELO-based one. For example, a quite simple one is the WAR from baseball. Well, in baseball it isn't actually that simple because it requires you to estimate how better (or worse) was a particular performance compared with the average performer in a set of different variables (batting, base running, defense) and how do these variables matter on the outcome of a game.
However, getting the performance above replacement for GPRO is easier. From historic data it's easy to get what's the average placement per league from the average manager (although some data filling is required, because many managers haven't been in the higher leagues). That would give a numeric value to each performance on a race by race basis.
Another possibility, which would produce a very similar result, is to compare the performance with a null hypothesis, with the score being related with how likely would you be getting a similar performance if the results were randomly assigned. For example, let's say I get a 20th place in Pro. Being in Pro if things were random has ~7% odds (I'm not including Rookie in these calculations). Now, getting a 20th in a race, again if things were random, is being ~50% percentile. Hence you'd get 93 + ( 7 * 0.5) = 96.5 relatively score that race.
Finally, we already have a metric that does the job. The team points per race stat is quite similar, albeit it gives quite lower relatively weight to the high leagues compared with the null hypothesis (or to any of the 2 previously mentioned approaches). Which isn't that bad anyways, because whereas the null hypothesis approach has better statistical accuracy, it produces too low (although statistically fair) weight to the lower leagues, which isn't that great for managers in said leagues, plus these metrics give more weight to a 40th place in Pro than a 1st place in Amateur. Again, makes sense, because in order to be in Pro in the first place you required to actually haul your ass off Ammy, but a 40th in Pro is easier to get than a 1st in Ammy.
The only problem with the team points per race metric, besides its relatively low weighting of the higher leagues, is that the game calculates it with the penalty for joining a new team. Many managers have lower than real because they were penalized each time they joined a new team. I think that part should be corrected. Also, it penalizes people who spend a lot of time in the lower leagues preparing a package for Elite, but then again, said preparation is done in order to pave their way to success in the high leagues, so it tends to balance itself.
|
|
|
|
Quote ( Edwin Silva @ September 26th 2018,18:58:41 ) WAR from baseball. Well, in baseball it isn't actually that simple because it requires you to estimate how better (or worse) was a particular performance compared with the average performer in a set of different variables (batting, base running, defense) and how do these variables matter on the outcome of a game.
WAR is a comparison of a player's value to that of a so-called replacement player, not an average player. So that an average player is greater than a replacement player and thus has positive WAR.
A comparison to average is certainly easier, both in concept and in execution, and so I like your thoughts in your post.
|
|
|
|
|
It would be nice to have the average speed for every league in "Track information".
|
|
|
|
Quote ( Brad Marshall @ September 26th 2018,20:49:09 ) WAR is a comparison of a player's value to that of a so-called replacement player, not an average player. So that an average player is greater than a replacement player and thus has positive WAR.
Thanks, Brad. I was under the impression "replacement" meant replacement from an average player from the league, but in fact it is as you said, where a replacement means an hypothetical reserve player being subbed in instead.
I want to add an ELO system isn't in reality that hard to implement. Ania's suggestion is quite good in that regards, although I'd keep it at W/L instead of predicted place difference (which is more accurate, but requires more calculation). The coding itself is quite simple. You can use for instance 10-20 seasons worth of history to produce the initial ELO seedings and from then onwards it would be a rolling one, not necessarily on a race by race basis but a season by season basis, in order to simplify calculations.
Each race requires a total of ~156k ELO calculations (for the whole GPRO structure), but each calculation is extremely simple, so it shouldn't take even 1s calculation.
That for the seeding period. Afterwards it's even easier. The ELO rankings should produce an expected W-L ratio per season vs. each manager, and conversely you have a real W-L ratio. The equation to update your ELO rating each season from real to expected outcome is super simple (NR = R + k (ER - RR)), where NR is new rating, R is previous rating, k is the k factor, ER is the expected W/L ratio, RR is the real W/L ratio).
If the team is interested, I could help with that. Also, there is a very handy webpage with codes for the ELO system in different languages (C#, C++, Python and Java) at https://www.geeksforgeeks.org/elo-rating-algorithm/ , although that's unrequired given the coding is very simple nonetheless.
|
|
|
|
Small suggestion for the manager profile page ('Other statistics' section): the manager's career average finishing position - only counting races they actually participated in.
It could also be added as a full-blown statistics table (with statistics per tier and everything) on the Statistics page.
|
|
|
|
|
Deutsch
Español
Română
Italiano
Português
Português
Français
Polski
Русский
Български
Македонски
Српски
Nederlands
Vlaams
Suomi
Lietuvių
Čeština
Magyar
Eesti keel
Slovenčina
Türkçe
Shqip
Ελληνικά
Hrvatski
Bahasa Melayu
Dansk
Pirate
