| View previous topic :: View next topic |
| Author |
Message |
sadistic
Alter Hase
Joined: 17 Nov 2004
Posts: 820
Location: Florida, USA
|
 Posted: 11.04.2007 01:18 Post subject: |
 |
|
| The only thing that is absolute at gravon is that I have drilled the most and highest quality females. The rest is fuzzy. |
|
| Back to top |
|
 |
acerimmer
Alter Hase
Joined: 18 Nov 2004
Posts: 725
Location: England
|
| Posted: 11.04.2007 02:31 Post subject: |
|
|
| Quote: |
| I would counter that WSC does work as a fun system for competition. It's basically a point accumulation contest. It's not going to produce a statistically precise result like Kleier, but it does tend to be a reasonable reflection of player strength to within 10-20 positions. |
I agree. But let me add that I think the wsc is a good form indicator also. Any player in form.......will be highly ranked.
| Quote: |
| whereas Kleier attempts to reconcile the entire body of loosely related game results. Kleier does a good job at that, but doesn't really do much better at producing the true list of how players stack out at a given time. |
I think your wrong there darth. The kleier rankings for the current year are a good indicator of a players strength. No ranking can accurately predict a 100% true rating on any given day. Every player has good and bad form. This is why the ratings are averaged out over a 1 year period......which thus takes into account for good and bad form throughout a given time period. Currently the ratings show the ratings of players over a time period of 3.33 months. By the end of December......the ratings should be pretty damn accurate. (Apart from players who play very little games......which are not truely accurate IMO)
| Quote: |
Look at Vauban as an example. He has a rating of 1849, but has played only one player over 1500, and his average oppenent has a rating of 1383!
|
This is one of my original points!
These players who have played so little.......have unrealistic rankings.
In my opinion......the number of games played before getting a ranking......... should be increased......before these players get a ranking.
Currently it is 20 games. This is ok for real life tournaments......as most players would find it difficult to play more than 2 tourneys per year. This would mean a new player would be playing for 2 years before obtaining a ranking.
For Gravon........things are a little different........it is much easier to play games online.....than it is in a real life tournament. For this reason, the ranking level should be increased to a more appropiate level. I would like to see a 100 game limit before entering the rankings (although a 50 game limit would be a step in the right direction.)
IMO: The 20 game limit is far too low to give an accurate rating. One bad loss and the rating would tumble.
A 50 game limit would be much better......but still this would not give a totally accurate figure.
A 100 game limit......would be the best minimum number of games before being ranked......as any higher number would eliminate a large percentage of players. Still not totally accurate after a mere 100 games.....but at least the rating would be a decent indicator of a players current strength.
I think the ideal number of games played.....to get a good accurate idea of player strength.....would be in the region of 300 to 500 games.
No one could realisticly play 300 to 500 games on form alone. Anyone playing consisantly well for this amount of games.....is a damn good player.
One idea.....is to have bands of awards........linked to an accuracy percentage (which is based basically on games played.)
maybe have the current 20 game limit......as the basic rating with a 20% accuracy figure.
then the next band would be a bronze rating......for players playing between 50 and 150 games. This could have an accuracy rating of around 50%
The next could be a silver rating......for players playing 150 to 300 games.
accuracy rating around 80%
The next rating could be the gold rating......for players playing 300 to 600 games.
an accuracy rating around 95%.
Lastly....maybe an Ultimate rating......for players playing in excess of 600 games per year. accuracy rating for the current year of 100% seeing as after such a large amount of games.......their rankings would never really move anywhere.
| Quote: |
Achilles is another. He has a rating of 1189 with 2683 games played. If he were struck by lightining today and emerged as the best player the Stat world had seen he would have to play until he was an old man before Kleier recognized it, whereas he would probably go to the top of WSC very quickly.
|
Your argument is correct......for the overall ratings.
Unfortunately......it isn't true when it comes to the yearly rating.......a player like this......could just wait till the following year.....and then dominate without question. This is why the yearly rankings are much more accurate than the overall rankings. |
|
| Back to top |
|
|
unbiasbob
Alter Hase
Joined: 21 Jun 2005
Posts: 604
|
| Posted: 11.04.2007 03:07 Post subject: |
|
|
| I can list 3 guys who climbed up several positions in the all time Kleiers rating: Dupe, Warlord and Specter. This must have been really tough to do but clearly these 3 guys are enormously improved players from whence they were. (whence they were?) |
|
| Back to top |
|
|
lightwing
Alter Hase
Joined: 13 Jun 2004
Posts: 393
Location: rotterdam
|
| Posted: 11.04.2007 20:42 Post subject: |
|
|
How about:
Ranking requirements = #month * 5? |
|
| Back to top |
|
|
unbiasbob
Alter Hase
Joined: 21 Jun 2005
Posts: 604
|
| Posted: 11.04.2007 22:07 Post subject: |
|
|
| Hi dudes, I'd like to chime in on this subject from my own personal situation. I have targeted playing bottom 30 players about 85-90% of the time this year. I have had decent success thus far winning % wise. I am ranked 82 out of 119 on the Kleier's ranking. I believe this to be accurate. I managed to climb from about #200 on WSC to #22 on WSC. The fact is that I am nowhere near this level in reality. I have now dropped back to #37 in that rank system but this is still far too high. So in short, someone like me can bottom feed and get to this high level when in reality, I am a #80 and no higher by my own admission. I am curruntly seeking financial aid for Stratego U so that I can be trained by the master Sadi phd. Sadi, if you are watching, do you know of any programs whereby I can seek this aid and get training by you? i would be forever grateful |
|
| Back to top |
|
|
sadistic
Alter Hase
Joined: 17 Nov 2004
Posts: 820
Location: Florida, USA
|
| Posted: 11.04.2007 22:35 Post subject: |
|
|
| All you have to do is come in my gameroom and watch the master at work. My hoodazzling trickery is something that all human beings should witness at one time or another during their lifetime. My newest fan, wellington, will be the first to tell you that he is utterly in awe of my stratego playing skills. |
|
| Back to top |
|
|
darthremark
Fortgeschrittener
Joined: 01 Sep 2004
Posts: 60
|
| Posted: 12.04.2007 01:09 Post subject: |
|
|
| Quote: |
| The kleier rankings for the current year are a good indicator of a players strength. |
Not necessarily, as examples have shown, and plenty more exist.
| Quote: |
| No ranking can accurately predict a 100% true rating on any given day. |
This is exactly my point. I shouldn't have to say anything more, but since you replied to me line by line... 
| Quote: |
| This is why the ratings are averaged out over a 1 year period... |
Actually, the yearly charts were put in place to generate excitement and competition because the all-time Kleier tends to stagnate, and exploited properly, WSC frontrunners become very hard to catch.
| Quote: |
| ...which thus takes into account for good and bad form throughout a given time period. |
Which is why they are never accurate at any given time and never will be. The rating is under constant "refinement", all the while the players skill is going up (or down). It's a moving target that cannot be hit by a formula. Kleier is a statistical tool. An attempt to reconcile a body of incomplete, uneven, loosely related data. It does a great job spitting out a useful number, but it's nothing more than a mathematical "best guess" based on said body of data. And it is open to skewing. I beat a guy about 500 points below me tonight and my Kleier 2007 rating went up a point. Give me a break.
| Quote: |
| Currently the ratings show the ratings of players over a time period of 3.33 months. By the end of December......the ratings should be pretty damn accurate. |
Hmm... Sometimes I think we are saying the same thing and then you go right past it. The accuracy you describe would only exist if all players participated similarly for the period in question.
All of these ideas are nice. If Spion could do it, probably the best solution would be to set up a customizable filter for the data. That way the all-time Kleier would be used for everything and users could choose to display only games from the last 12 months, or players with over 600 games, or any number of other variables. The data is powerful. The formula is not magic.
| Quote: |
| Unfortunately......it isn't true when it comes to the yearly rating.......a player like this......could just wait till the following year.....and then dominate without question. |
Good grief. Is your goal domination or an accurate rating? I don't want to wait for some arbitrary time window to know the rating. I want an accurate rating now! ..of course the formula can't provide it.
| Quote: |
| This is why the yearly rankings are much more accurate than the overall rankings. |
?? First you say we must wait a year for the yearly Kleier to get enough data to be "pretty damned" accurate. Then you say the all-time Kleier, which has far more data, is not really accurate at all? You don't get it both ways, Ace. Sorry mf. |
|
| Back to top |
|
|
sadistic
Alter Hase
Joined: 17 Nov 2004
Posts: 820
Location: Florida, USA
|
| Posted: 12.04.2007 02:40 Post subject: |
|
|
| It's much easier just to calculate who has drilled the most females. |
|
| Back to top |
|
|
art
Alter Hase
Joined: 16 Mar 2005
Posts: 325
Location: USA
|
| Posted: 12.04.2007 04:27 Post subject: |
|
|
Are you the father of Anna Nichol's baby? Science lies.
Let's stop the rating nonsense. I propose that the calculations continue as is, but instead the ratings only be posted to the internet once a month.
Wouldn't that drive you guys all nuts? Or, I mean more so.
No thread is complete without the infamous deli-sandwich87 "lol", the incoherent knee jerks of hihoben, or the grammatically incorrect, meaningless dribblings, of the one down under.
Put the rating system into an interactive 'always live' web program for which a user can insert or delete any battle results one wishes, and see how he ranks a top his make believe fantasy land of King of all Gravon.
Charge $5.00 fee for each "interactive session", $50.00 for a yearly subscription.
Honestly, my only gripe is that I wish to see the actual ranking points that existed on the date that I played the opponent and not his current ranking points. So when you click on your nick and scroll up to old dates you''d see that days rank. My fantasy is to visit 1300 before the arthritis kicks in.
Just kidding to you all. You're all invited to the party that I'm having in Iraq when they become our 51st state. 
Last edited by art on 13.04.2007 18:22; edited 1 time in total |
|
| Back to top |
|
|
acerimmer
Alter Hase
Joined: 18 Nov 2004
Posts: 725
Location: England
|
| Posted: 13.04.2007 00:43 Post subject: |
|
|
| Quote: |
The kleier rankings for the current year are a good indicator of a players strength.
Not necessarily, as examples have shown, and plenty more exist.
|
What examples are you talking about? Which players in your opinion.....should be higher or lower than their rank?
(leave out the guys who have played less than 100 games.......as I agree that players playing less than 100 games could have a inaccurate rating...........once a player reaches 100 games.....it is much harder to hold an inaccurate rating.....and less examples would be found.)
| Quote: |
No ranking can accurately predict a 100% true rating on any given day.
This is exactly my point. I shouldn't have to say anything more, but since you replied to me line by line.. |
Although it is not 100% accurate.......I do think the ratings are accurate to a +/- 1 % (once a player has played around 300 games for that year.)
| Quote: |
| Actually, the yearly charts were put in place to generate excitement and competition because the all-time Kleier tends to stagnate, and exploited properly, WSC frontrunners become very hard to catch. |
Regarless of why the yearly rankings were added (I do not know if this was the reason why it was introduced)............the effect is that we have a much accurate ranking system than the overall rankings.
| Quote: |
.which thus takes into account for good and bad form throughout a given time period.
Which is why they are never accurate at any given time and never will be. The rating is under constant "refinement", all the while the players skill is going up (or down). It's a moving target that cannot be hit by a formula. Kleier is a statistical tool. An attempt to reconcile a body of incomplete, uneven, loosely related data. It does a great job spitting out a useful number, but it's nothing more than a mathematical "best guess" based on said body of data. And it is open to skewing. I beat a guy about 500 points below me tonight and my Kleier 2007 rating went up a point. Give me a break. |
No ranking will ever be accurate on any particular day. Sometimes we have good days and sometimes we have the bad days. The only thing that can be done to predict a rating........is to get a statistical figure for how many good days we have......compared to the bad ones. Better players have many more good days.......but still have the odd bad day. Bad players will have many bad days.....but will beat even the best......from time to time.
You beat a player 500 points below you and your ranking went up 1 point........that sounds about right. What do you want.....a medal?
You should beat players who are 500 points below you........around 90 to 95% of the time. With this in mind......why should beating this player, change the rankings dramatically? Beat a player who is 500 points above you......and see your ranking rise dramatically (as it should).
| Quote: |
The accuracy you describe would only exist if all players participated similarly for the period in question.
|
This is why I would like to see the rankings have a minimum games before being ranked......of 100 games.....thus this would make the rankings much more accurate. But yes.....I agree......the low number of games played by some players......give some players inaccurate ratings.
| Quote: |
Good grief. Is your goal domination or an accurate rating? I don't want to wait for some arbitrary time window to know the rating. I want an accurate rating now! ..of course the formula can't provide it.
|
You ask for perfection....this will never happen. What we have......is a very good system that is not perfect.....but is the CLEARLY the best system available. It is good to a low +/- %. (depending on a resonable minimum number of games played per player)
The only way that I can see......how the rankings could be improved to predict the current rating of each player.....is by making the rankings a rolling 1 year ranking. In addition to this maybe have a top 100 rating for each month.........example: march 06 to march 07 top 100
april06 to april07 top 100.
| Quote: |
This is why the yearly rankings are much more accurate than the overall rankings.
?? First you say we must wait a year for the yearly Kleier to get enough data to be "pretty damned" accurate. Then you say the all-time Kleier, which has far more data, is not really accurate at all? You don't get it both ways, Ace. Sorry mf. |
The problem of the overall rankings......is that the players who have changed ability over the years.....are not represented accurately. Within the short period of one year.....this change is reduced dramatically.....AND THEN reset .......giving the player a fresh chance to provide more accurate data.
If you can't see the negative aspect of the overall rating compared with a yearly rating......then you must be the only player on gravon who thinks this. |
|
| Back to top |
|
|
eindhoven
Alter Hase
Joined: 22 Jan 2005
Posts: 269
Location: eindhoven, Lampegat en Philipsstad
|
| Posted: 13.04.2007 03:36 Post subject: |
|
|
PFFFFFFFFFFFFFFFF
it's all about how many games you play and against who !!!!
how more you play --- how less to loose/win
how less you play --- more to loos/win
always win = win
oponent lower ranked --- more to loose and less to win
oponent higher ranked --- more to win and less to loose
so real stratego-fans just play the game as much as they can and having FUN
j. cruyff |
|
| Back to top |
|
|
darthremark
Fortgeschrittener
Joined: 01 Sep 2004
Posts: 60
|
| Posted: 13.04.2007 04:59 Post subject: |
|
|
| eindhoven wrote: |
| it's all about how many games you play and against who !!!! |
Thanks Eindhoven, for that simple truth.
Ace, this isn't that complicated and I don't want to get repetitive. I'm not saying Kleier is off by a mile. It's *usually* a reasonable indicator of relative strength. I think you're trying to make it something that it's not though. You ask where players are at today, but by your own arguments you can't get there. You reject historical Kleier because it's not current, but you readily state that Kleier needs 300+ games to get a good reading. Man, someone can change a lot in 300 games. They might be white hot by game 300, but dragged down by early learning losses for example. There are any number of reasons why players might go up or down. The rating is a statistical tool, nothing more. They hold WC's every year don't they? Why don't they just anoint whoever Herman tells them is best?
I like the sliding 12 month chart. That's actually a really cool idea (albeit similar to my 12-month filter). |
|
| Back to top |
|
|
acerimmer
Alter Hase
Joined: 18 Nov 2004
Posts: 725
Location: England
|
| Posted: 13.04.2007 17:07 Post subject: |
|
|
We seem to be splitting hairs over something I think we generally (but not totally) agree upon.
I think the rating after 100 games is pretty accurate........although to get an optimum result......I think 300 to 500 is needed.
Only new players will be dragged down by early losses......and this problem rectifies its self on the following year.
The formula does not say who herman thinks should be number 1......it shows who the formula says is number 1.
The formula is very good.
Yes, they do hold WC's ! I might even get to one of those......one day.......maybe you should attend also?
The thing with the WC......is that the winner could be anyone of maybe 20 players who are good enough to win the WC. The player on form going into the WC would be favourite to win.......as past results do not effect the outcome.......and it is merely the best player on the day.....who takes the WC crown. This is not always the best way.......but it is an established way which is used in many competitions. |
|
| Back to top |
|
|
eindhoven
Alter Hase
Joined: 22 Jan 2005
Posts: 269
Location: eindhoven, Lampegat en Philipsstad
|
| Posted: 13.04.2007 17:35 Post subject: |
|
|
make it 500 and i agree with you totally...... Ace
Greatings from eindhoven |
|
| Back to top |
|
|
acerimmer
Alter Hase
Joined: 18 Nov 2004
Posts: 725
Location: England
|
| Posted: 13.04.2007 18:13 Post subject: |
|
|
wow..........eindy with a sensible post!
I concure! |
|
| Back to top |
|
|
|
Search
Memberlist
Usergroups
Profile
Log in to check your private messages
Log in
