Ladder Advantages To TCP/IP Players--What To Do?

[Col Deadmarsh]

Those of you who have played a lot of games at TH have realized that the more you play, the more points you rack up. Therefore, the player who has 100 wins and 20 losses will have more points than the player with 10 wins and 2 losses. Now, I realize that those of you who play a lot should be recognized for being veteran CM players and the best of the lot because you've earned it by putting in a lot of time to the game. What I have a problem with is the people who are just as good at the game but only play PBEM games. They should be recognized too but using the current system of having only 1 ladder for all of us prevents this from happening.

Would having 2 ladders, one for TCP/IP and one for PBEM be an end-all solution? No. But it would greatly even things out. There are TCP/IP players who probably report 3-4 games a day as opposed to the PBEM player who reports 3-4 a month. These numbers are way out of balance and end up significantly skewing the player ratings on the ladder.

Since the TH ladder is based on the Swiss chess system, I'm assuming it would be extrememly hard to "adjust" the system to make it fair even if we have a top mathmatician in this community. The goal in this case would be to reward win percentages more so than they are now instead of rewarding the person who plays the most games. Of course like I said before, those people who invest so much of their time into this game should be rewarded for doing so and maybe this isn't the right solution to the problem.

Therefore, it seems the only alternative to fixing this problem, (and it IS a problem) is to have a seperate ladder for each style of play. As a PBEM'er who averages 4 games at a time, I don't mind being ranked against someone who has 15 PBEM games going on at once. This is a lot better than trying to compete with someone who plays 4-5 TCP/IP games a day.

The end result here is that the best of both styles of play will be rewarded by having their name at the top. And if you play both, then you simply report which style of game you played to the appropriate ladder.

We all want to be recognized for our talents at this game and this proposal would give that to us.

[Stalins Organ]

I'm only on hte ladder to get differnt opponjents. The rankings don't mean much to me.

[Panzer Leader]

Why is it important? But anyway, wouldn't it work to have rankings based on win/loss ratio, like a percentage? Of course, it would still mess it up if some bloke broke the record after the 61st game. Just ask yourself: What would the Babe do?

[Grunto IV]

they should base standings upon w-l record with a minimum number of games to qualify.

andy

[Sock Monkey]

Well, I signed up for the ladders simply to find PBEM opponents, so the rankings don't matter overmuch to me, either. That being said, what is the point of maintaining a ladder with a ranking system if it isn't adjucated fairly? Everybody playing the game has to be competitive at some level, and some people really get a kick out of that, so'd I support dual ladders if it addressed that problem .... Not that I'd have to do any of the work, mind you Anyway, if nothing else, it would be an efficient way of finding an opponent for the type of game you want.

[Treeburst155]

I just deleted a huge post because it wasn't clear after rereading it. I'm going to try again.

The chess rating system used by TH is the absolute best system for determining relative skill. However, it takes a VERY long time for a person's rating to accurately reflect his skill if playing PBEM only. TCP/IP players will hit that point MUCH sooner than PBEMers.

Because the starting point of 1,000 is a fairly low rating most people's score will increase over the long haul. That's why the TCP/IP players are pulling ahead of the PBEMers. Their "long haul" is not as long. Eventually the TCP/IP players will quit going up as a group. Why? They will have reached the point where their true skill levels are reflected in their rating. At this point the TCPers average rating will be static. Their ratings will spread out and the men will be separated from the boys among the TCPers.

Meanwhile the PBEMers will literally be years behind in their "long haul" to an accurate rating. Even the best PBEMer will be far behind the bulk of the TCPers because he hasn't played enough games to reflect his skill level yet.

Eventually the PBEMers' average rating will catch up to the TCPers but it will take a VERY long time. The PBEMers will never catch the TCPers in the amount of games played but they will hit the point where their ratings are an accurate reflection of skill. When the PBEMers hit this point they will find themselves right up there with the TCPers even though the TCPers have played 10 times as many games.

Something else I find interesting is that the TCPers will hit the point where it is undesireable to play a PBEMer. The gap in ratings will be too great. A loss would devastate the TCPer's score while a win against the PBEMer would help little. The TCPers will be forced to play each other only, if they're concerned about ratings.

In effect there will be two different ladders going on and the gap will be obvious. You will know by the number of games played and higher ratings who is in the "TCP League" and who is not. This condition will persist for at least a year I would think, but the average PBEM rating will slowly catch up to the TCPers. In 3 years there will be no obvious delineation between a TCPer and a PBEMer. The average rating of both groups will be about the same.

The scoring system really does not reward a player for simply playing the game. It just needs many games played in order to reflect the truth. The quicker you play these games the sooner the truth about your skill is revealed. Remember, the 1000 starting point is a low rating. Most peoples scores will increase over time. The PBEMer just needs a great deal more time than the TCPer.

Treeburst155 out.

[ 05-29-2001: Message edited by: Treeburst155 ]

[yobobo@TH]

Colonel, I still don't believe in the separate leagues.

Why should a good player with 100 wins be ranked the same as a player with 10 wins? Even if both have a 70% win record? And what about win types and player rating you beat? This all comes into play. Also some PBM players finish a game in a week, some finish in 3 months, so how would that work :/ Should we have even more ladders, like 3 month PBMers, 2 Month PBMers etc..

One more thing, look at my rank and games played. I have plenty of games but yet my rank still sucks hehe.

[ 05-29-2001: Message edited by: yobobo@TH ]

[ 05-29-2001: Message edited by: yobobo@TH ]

[Redwolf]

Treeburst, I think your message incorrectly implies that the top players at TH prefer not not to play low-ranking players anymore.

That is not the case, I have a low score and play top players often and some I didn't play so far asked me for a PBEM. And that even though my score is low because I almost exclusivly play top players and my score is (IMHO) lower than my dangerousness would imply. Still, my impression is that most top players just want a fun game or at least take any game they can get.

I also think your basic assumption is incorrect. You still get enough points from playing lower ranked players and playing players of your own class is of course not guaranteed to succeed. IMHO, the relationship of points and statistical win chance is such that playing the lower ranking players would gain more points overall since most of the time their chances of winning or drawing are very low. Still, top players don't do that, either, they play each other as well.

While we're at it, I also found the top players to be very helpful and not shy to talk about their tactics.

The true "problem" (it isn't one, IMHO) here is that those who play masses of TCP games get very good very fast.

[Col Deadmarsh]

<BLOCKQUOTE>quote:</font><HR>Originally posted by yobobo@TH:

Colonel, I still don't believe in the separate leagues.

Why should a good player with 100 wins be ranked the same as a player with 10 wins? Even if both have a 70% win record?<HR></BLOCKQUOTE>

My example was a simple one so everyone would understand. Sure, someone with only 10 wins should not have the same score as the person who has 100. But let's take the TCPer with 200 wins and 100 losses vs. the PBEMer with 50 wins and 25 losses. Now, the PBEMer has played 75 games. He is no longer a newbie at this and his ranking should logically be somewhat the same as the TCPer, but it isn't.

Do you think it's fair for the PBEMer with a good record to be ranked under the TCPer with only an average record? Is there something we can do here? I read what Treeburst had to say and I thought that was really interesting that it would take the PBEMer a year if not more to catch up to the TCPer in points even though they have the same win percentage. Could you explain this system we are using so we understand how things are calculated. Tell us how many points we get for each win and how many are subtracted for each loss and what happens when there is a draw.

<BLOCKQUOTE>quote:</font><HR>Also some PBM players finish a game in a week, some finish in 3 months, so how would that work :/ Should we have even more ladders, like 3 month PBMers, 2 Month PBMers etc..<HR></BLOCKQUOTE>

Like I said, having another ladder is not the perfect solution but it would be better than grouping everyone together. I would not mind being put in a ladder with the PBEMer who plays 19 games at a time.

Maybe another ladder isn't the solution. Maybe the base number needs to be adjusted to move the PBEMers up to a number where they don't have to play for a year to show their true ranking. Off-hand, I would say that 50 games should give both TCPers and PBEMers a "true ranking." What are your thoughts on this?

If there are any mathmaticians around here that can validate what Treeburst said, come foward and do so. Maybe you can explain how things can be made fair for all parties. After all, not all of us joined the ladder to find people to play with. Some of us want to reach the top some day but with the way things are now, that would apparently take years to do so.

[ 05-29-2001: Message edited by: Colonel_Deadmarsh ]

[Olle Petersson]

<BLOCKQUOTE>quote:</font><HR>Originally posted by Treeburst155:

The chess rating system used by TH is the absolute best system for determining relative skill. However, it takes a VERY long time for a person's rating to accurately reflect his skill if playing PBEM only. TCP/IP players will hit that point MUCH sooner than PBEMers.<HR></BLOCKQUOTE>First I must admit I'm not on any ladder, nor do I know that chess system, but...

If the system in use is somewhat like the handicap system used in golf, which I do know something about, why not just enter at a better rating and let those who get a bad record lose points and positions, while those who are good advance?

Cheers

Olle

[Treeburst155]

Redwolf brings up a very good point.

A player who plays many games against humans will improve his skills faster than people who don't play so many games. This would make for a higher TCPer average score than the PBEMer average score. The TCPers play better from experience. The more games (against humans) you play, the more mistakes you make, the more you learn. This seems fair to me. The TCPer average rating will still achieve a static level eventually. I can't expect to compete with Tiger Woods on the golf course when he plays daily and I only play once per month.

I'm not saying that TCPers won't play PBEMers. I'm saying that when their ratings get high enough (1400-1600) it would not be wise to play PBEMers (especially ladder newbies) IF rating is important to the TCPer. Right now there really isn't that much of a gap. Swamp has played 124 games with a rating of 1458. This is still an average rating. His rating hasn't hit his true skill level even yet. This is aggravated by the fact that his actual skill is probably still increasing. This goes to show how long it takes to obtain an accurate rating when everybody starts at 1,000. Still, the points Swamp risks by playing a 1,000 pt newbie and losing are significantly greater than the points he stands to gain by beating the newbie.

If you were to take up with a chess club your rating would be fairly accurate within 20 games or so because you would be playing others who have played thousands of games. Their ratings are accurate. Your score adjustment after each game would therefore be more accurate. As it pertains to the TH scoring system even Swamp is still a newbie having ONLY played 124 games and all those against people whose ratings are not yet accurate. Everyone is a newbie!

The TH scoring system is by far the best way to determine skill but it has the significant disadvantage of taking a VERY long time to work it's magic. This is mainly because everybody is a newbie to the scoring system. The system is adjusting ratings based on inaccurate ratings of both players. This WILL sort out eventually. With every game you play your score is adjusted to better reflect your skill level.

The TH system is the best IMO inspite of the time it takes to work. In the meantime you have players' Win/Loss record and percentage. In the end, W/L percentage will not be as accurate as rating because the percentage doesn't take into consideration the skill of a player's opponents. Right now however it is the best way to judge skill while the rating system sorts out the players. If the player has 10 or more games played you can get a fairly accurate picture of how good they are. BTW, people like Swamp who win 77% of their games can expect to eventually be rated way up there, (3000?)provided of course they can maintain this percentage against people fairly close to themselves in rating.

When I browse the TH ladder I take into consideration games played, Win/Loss record and percentage, and finally I look at rating because that is the least important stat at this point if you are trying to determine skill levels. Being at the top of the list impresses me little at this point. Having played 124 games and winning 77% DOES impress me.

It would be nice to be able to sort the ladder in various ways rather than just have the highest rating at the top. It would also be nice to be able to filter out those with less than x number of games played. If a guy has played only 5 games I may not want him cluttering up my ladder sort as he hasn't played enough for me to be interested in his stats.

So how about it Yobobo? If the guy who is 10 and 1 could sort the ladder based on W/L percentage and exclude people with few games played (3-0 for example) he would find himself at the top. This thread would not have happened.

Treeburst155 out.

[ 05-29-2001: Message edited by: Treeburst155 ]

[Redwolf]

<BLOCKQUOTE>quote:</font><HR>Originally posted by Treeburst155:

Swamp has played 124 games with a rating of 1458. [...] Still, the points Swamp risks by playing a 1,000 pt newbie and losing are significantly greater than the points he stands to gain by beating the newbie.

<HR></BLOCKQUOTE>

You still fail to consider the win chance. Whether it is more risky or less risky to play a newbie instead of an almost equal player depends on the relationship of points to be gained and the win chance against that player.

Example: If you get 3 points from beating bigmob, but only one from beating fraggything, because fraggything is lower rated, you still can't tell which one is better to play without saying something about the win chance. If your chance of winning against bigmob is 60% and the chance to beat fraggything is 90%, then you know that from playing 10 games against bigmob you get 18 points, from fraggything you would get 9 points. What is what your assumption implies.

However, it is just your assumption, without prrof. If the win chances and points are different, it may pay better to play the lower ranking player. (Besides: the TH system works on victory levels, not just won/lost).

The point about a good ladder computation system is that the points you gain from playing a certain player reflect the abilities of that player so that you would statistically play any player without looking at the ladder score too much. 10 games against a 1010 points player should gain the same points for you as 10 games against a 1250 points player.

The top players at TH play each other and they play newbies alike. Given their play experience and assuming that they want to be competive, that says that in their opinion in TH system reached this goal. That is all you can ask for.

This doesn't touch the problem that good players may have a low score because they gon't play often, which is the original problem this thread discussed. But given that people who play often get actually better, I don't think that this is that much of a distortion.

[yobobo@TH]

So many variables here.

Not only win % but win type to %.

I can have 60% wins, but 80% of the wins are only minor wins compared to the player with 80% total wins.

[Capt. Vance Astrovich]

3 comments.

First, what about the person who plays a game BOTH PBEM and TCP/IP? Would that mean a third ladder for these combination games?

Second, why not provide a second listing or ranking based on % win/lose record. Just have it there as another system rating for people to use to compare.

Last, there could even be a ranking list that just rates people based on there last 20 games, or 50 games, or 100 games or some arbitrary number. Then whether it takes 1 month or 1 year to play the required # of games, it is a comparison of just that # of games.

[yobobo@TH]

We have that in place from 50-100 games played http://tournamenthouse.com/CM/TTopDogs.shtml I can add 20 or less etc..

[Redwolf]

Just provide a link to an ASCII-table with seperators that people can import into Excel or use with awk.

looser:looserscore:winner:winnerscore:kind-of-loss:date

That way people from outside could build analysts systems in websites.

BTW, what about storing who was defender and/or Axis in a game? More numbers to play with.

[yobobo@TH]

Colonel,

The bottom line is players like yourself that only play 9 games in as many months will climb very slowly, unless your winning every game. Compared to players like Redcoat, another PBM only player who has 16 games played in just over a month. But again I have 84 games played and am ranked 934. So the games played has not helped me.

Yes Redwolf,

I have that axis allied games played ready to go for some time now. Will try and get it in this week.

[ 05-29-2001: Message edited by: yobobo@TH ]

[ 05-29-2001: Message edited by: yobobo@TH ]

[Gen-x87]

This is a lot better than trying to compete with someone who plays 4-5 TCP/IP games a day. "

Damn who are these people?!?!?!?!?!? I play at TH and Ill be lucky to get 1 game anight in.

Anywho I joined to find a ready supply of human opponents. I do not see the purpose of having a second tier for PBEM players. What happens when you mix and match?!?!?!?!?!?!? Should we then make a third tier for PBEM\TCP players?!?!?!!?!?!?!?!?!?!?!?!?!?

If you are so worried about your ranking then maybe you should take a look at why you play the game.

The only reason I use the ranking system is to get an idea of how good my opponent is. If they are in the lower rankings like sub 900 then I usually try to experiement a bit more. If they are 1100+ then I bring what I know usually works.

My ranking is like 950 and I could care less. I play for the fun of it

Gen

[Treeburst155]

Redwolf,

Good point about the variable degrees of victory. That does throw an added complication into things I hadn't thought about.

If I have a rating of 1,500 and a newbie comes along to play me I am taking a great risk because I have no idea of his skill level. Is he new to CM or is he just new to the ladder? Whether he is very good or not the points I stand to gain will be less than the points I stand to lose for any given level of victory because of his low rating. I will admit that the multiple degrees of victory complicate this characteristic of the chess system in ways I can't quite predict since I'm not a statistician. I would also admit that the ladder newbie is PROBABLY not as skilled as the old timer on the ladder. This makes it a fairly safe bet for a win of some degree for the old timer when he takes on a ladder newbie.

I'm not on the regular ladder but I've got lots of experience with the game. If I were to get on there and play the higher rated players they would indeed be taking a risk with their rating that they are unaware of. Not that I am THAT good, but I'm definitely not the newbie my 1000 pt. rating would imply.

The best way is just to have a sort function for Win/Loss percentage I think. That way the guy who is 50 and 25 will be right up there with the guy who is 200 and 100.

Another way to assess skill is to have tournaments where everybody in the tournament plays everybody else one time. All participants would start with zero points. A player would receive 1 point for a minor victory, 2 for a tactical, 3 for a major, and 4 for a total victory. Draws would score 1/2 pt. for each player and losses score nothing. Whoever has the highest total at the end of the tourney wins!

This would assess skill level rather quickly between relatively small groups of players.

New tournament groups could then randomly be created from the same player pool and the process repeated again and again all the while maintaining a running total of people's accumulated scores. Allowing newbies into the "tournament system" presents some real problems however. Everyone must have the same number of games played in order to reflect true skill levels. The rating system's problem is somewhat similar. Still a huge tournament of this sort would be fun.

Treeburst155 out.

[Redwolf]

<BLOCKQUOTE>quote:</font><HR>Originally posted by Treeburst155:

If I have a rating of 1,500 and a newbie comes along to play me I am taking a great risk because I have no idea of his skill level. Is he new to CM or is he just new to the ladder?

<HR></BLOCKQUOTE>

That is of course a valid concern. But players new to the ladder are not the only ones with a score of 1000 and being more dangerous. In my case I would tend to rate me as more dangerous than my score implies, because my score is ruined by playing top players almost exclusivly. Besides a number of ruined games with low-ranking players where something went wrong so that no win for me was recorded...

However, there are plenty of around-1000 players in TH who have a game history and then you can easily see whether they play bad or just don't care about the score too much, which brings us to your next point...

<BLOCKQUOTE>quote:</font><HR>

The best way is just to have a sort function for Win/Loss percentage I think. That way the guy who is 50 and 25 will be right up there with the guy who is 200 and 100.

<HR></BLOCKQUOTE>

... because I think that is too much of an individual question.

What makes a dangerous 1000 points player for you, personally?

- whether he played high-ranking player often, no matter what the result?

- whether he once beat a high-ranking player? May have been luck.

- whether he lost against someone you already played?

I was dead serious that yobobo might offer a raw table so that each player may build his own system in a spreadsheet, programming language or piece of paper. But then, someone has to program the table export.

<BLOCKQUOTE>quote:</font><HR>

Another way to assess skill is to have tournaments where everybody in the tournament plays everybody else one time.

<HR></BLOCKQUOTE>

I don't think it is a good idea to push people into mandatory games of any kind, even if the issue we discuss would be a problem for me.

Having said all this, it is obvious that I care less about my ladder score than colonel and maybe you, so keep that in mind when my words look like downplaying.

[ 05-29-2001: Message edited by: redwolf ]

[yobobo@TH]

Need some input on this,

What if a player did not lose any points for a loss. For wins he still would only gain point amounts depending on his opponents rating and win type.

Hmm

[Redwolf]

<BLOCKQUOTE>quote:</font><HR>Originally posted by yobobo@TH:

What if a player did not lose any points for a loss. For wins he still would only gain point amounts depending on his opponents rating and win type.

<HR></BLOCKQUOTE>

People will not fight anymore to get a lesser loss (total, major, minor etc.) and the opponent will have much greater chances to get a surrender from some players without a sense of honour. If think you cannot use the different victory levels anymore if you introduce this, just binary win or loss.

Generally speaking I think we need the reluctance to report a loss otherwise reporting will get messed up for various reaons.

[ 05-29-2001: Message edited by: redwolf ]

[zahl]

<BLOCKQUOTE>quote:</font><HR>Originally posted by Treeburst155:

A player would receive 1 point for a minor victory, 2 for a tactical, 3 for a major, and 4 for a total victory.

<HR></BLOCKQUOTE>

There's one inherent problem in systems that take victory level into account. You could score four brilliant, but only narrow wins against state of the art opponent and then commit a major blunder in the fifth game, resulting in a total defeat. Your opponent would have negated all of the wins, all thanks to your horrible mistake in one game.

[Col Deadmarsh]

Okay, this is where I stand now...

After hearing what Treeburst had to say, I feel better about everything now knowing that someday (probably when CM3 hits the shelves) a PBEMer with the same skill as a TCPer will have the same points and be ranked accordingly. At least it evens itself out in the end.

Until that happens, I would like to see a toggle button that would show the top 20 in win percentage. This would give everyone a good idea right now of each player's skill level. Of course, there should be a minimum number of games played to be ranked with a percentage, otherwise everyone who's 3-0 will be at the top of the list. Maybe a good minimum number would be around 20-30 games played. What do you think?

[yobobo@TH]

But again win % means nothing.

If I played 20 -900 rated players and had 10 minor wins, and you played 20 +1050 rated players and made 10 total wins. What would the % mean? We would still show the same win %.

How about the top dogs link showing ratings from 20 and under games played, 20 - 50 games played or something like that?