Pente variant proposal- ratings point gambling using a doubling cube
Posted:
Jun 18, 2008, 2:47 PM
I recall reading that in some circles Pente has been played using a Backgammon doubling cube ( http://boardgames.about.com/cs/backgammon/ht/doubling_cube.htm )while wagering on the outcome. Over the years, many variants of Pente have been proposed (quite a few of them now playable at DSG), most of them in order to help offset the P1 advantage. So, here's another one...
P2 starts each game with control of a doubling cube, which can be used by whichever player controls it at the time in order to increase the ratings points which the winner of the game receives. Figuring out how much of a factor each "double" would add to the ratings would be a detail to be worked out, of course. As soon as the player who controls the cube uses it, the other player either accepts that the stakes in the game have now risen or else concedes the game. If the player accepts the double, control of the cube passes to that player.
Let's say your opponent moves and now has 2 open 3s but you see a way to extend threes which you have already built in order to eventually capture enough pairs to win. A good time to whip out the old double, huh?
Or say you know the hammer really well. The position shows up in the game and you have the handle of the hammer. Another good time for a double...
My sense is that this doubling cube addition would help the top players to increase their rating points based on both their ability to "see further and more clearly" and based on their ability to psych their opps. I'm not opposed to other variations as being perhaps superior (including, of course, the points based on sets proposal, AKA UP ente).
I think this might be a fun one to explore as an additional variant, though.
Of course, a doubling cube could be an additional feature which would add spice to all of the variants, not just Pente...
Message was edited by: watsu at Jun 18, 2008 8:49 AM
Retired from TB Pente, but still playing live games & exploring variants like D, poof and boat
Re: Pente variant proposal- ratings point gambling using a doubling cube
Posted:
Jun 19, 2008, 4:09 AM
A weighted rating system on individual games is a much better way to properly apply rating changes to a given game than is a set based system. The set based equation, though an improvement, is not an elegant ratings system.
Here is a simple weighting solution to the ratings equation that in my opinion fairly adjusts the existing systems flaws in ratings adjustments. Don't replace the existing formula with a new one, simply apply a weighting to it's end results to reflect the value of P1 win\loss vs P2 win\loss
It is my opinion that a win as P2 is roughly 5 times the value of a win as P1, and the same ratio applies to losing as P1 vs losing as P2.
x = ratings increase/decrease amount under existing system
P1win = ratings adjustment at a 1:5 ratio(meaning a win as P1, and a lose as P2 has 1\5 comparative value of a P2 win, and a P1 lose) P1 win/P2 loss formula amendment: x * 0.5 = P1win
P2win = ratings adjustment at a 5:1 ratio(meaning a win as P2, and a lose as P1 has 5 times the comparative value of a P1 win, and a P2 lose) P2 win/P1 loss formula amendment: x * 2.5 = P2win
Example: Bob and Lee have just finished a game. Lets say Bob won and Lee lost and Bob's rating changed +8 to Lee's -4 under the standard ratings adjustment. (the actual ratings for each player is irrelevant)
Existing Standard Result Example: Winner Bob +8 Loser Lee -4
Weighted adjustment applied to existing equation’s results with winner as P1 will result as follows: Player 1 Win: Winner Bob +8 * 0.5 = +4 Loser Lee -4 * 0.5 = -2 Weighted adjustment applied to existing equation’s results with winner as P2: Player 2 Win: Winner Bob +8 * 2.5 = +20 Loser Lee -4 * 2.5 = -10
These factors were derived through a differential equation specifically designed to maintain a 5:1 ratio between winning as P1 vs winning as P2, and losing as P1 vs losing as P2. The equation is applied to the results from original standard ratings change. If the actual win\loss ratio for P1 vs P2 is something other than a 5:1 then the equation can be altered to fit whatever ratio actually exists. I thought a post application would be easier to apply to existing results rather than revamping the original equation. A max change limit of 30 points can still be applied.
These ratings changes reflect real game value differences of wining or losing as P1 and as P2, that is if the differential ratio is 5:1. Other ratios could easily be applied. It is simply my opinion that 5:1 is roughly accurate.
Re: Pente variant proposal- ratings point gambling using a doubling cube
Posted:
Jun 19, 2008, 4:47 AM
A more exact ratio could be determined by a comparative statistical analysis of games in the DB where similarly ranked players(at the time of their game) above 1700 faced off. I picked 5:1 simply because in my opinion, similarly ranked players will win 5 times more often as player 1 as they will as player 2.
I would eliminate similarly ranked players below 1700 simply because the random elements of luck become more a part of the game in those ranks below 1700 and would not represent a true evaluation of the weighted differential between a player 1 game vs a player 2 game.
Posts:
2,241
Registered:
Mar 4, 2007
From:
San Francisco
Age:
45 Home page
Re: Pente variant proposal- ratings point gambling using a doubling cube
Posted:
Jun 19, 2008, 5:55 AM
>so does this mean that the 5:1 ratio should only be applied in matches between 2 players that are both 1700+? and that players 1699 and below should have a different ratio? say a 2:1 for example.
~Zoeyk
Scire hostis animum - Intelligere ludum - Nosce te ipsum - Prima moventur conciliat - Nolite errare
Re: Pente variant proposal- ratings point gambling using a doubling cube
Posted:
Jun 19, 2008, 6:33 AM
No. I simply believe that the randomness of the quality of moves represented by lower ranked players would not offer an accurate statistical evaluation for determining a differential that purely reflects the win-lose ratio between black and white games of evenly matched players.
The ratio would be the same for everyone, but would be determined by the more revealing and statistically viable data derived from more experienced players.
Posts:
2,241
Registered:
Mar 4, 2007
From:
San Francisco
Age:
45 Home page
Re: Pente variant proposal- ratings point gambling using a doubling cube
Posted:
Jun 19, 2008, 7:16 AM
Ok sounds good,..now some people (such as my self) are not good at understanding what the math scrypt you wrote might mean in terms of actual points gained/lost. wonder if you could humor me by saying what the actual point out comes would be if say;
bob 1950 plays as white (player 1) ~vs~ Lee 1750 playing as black (player 2).
and lets say bob wins,..whats his new rating? and lets say bob loses,.whats his new rating?
and lets say Lee wins,..whats his new rating? and lets say Lee loses,.whats his new rating?
and now,...same questions, but this time bob plays as black and Lee plays as white. (this is reversing rankings across seats)
this is just so people (such as my self) get a more clear idea of what a 5:1 ratio will really mean, for those who are interested in your idea for a ratings solution, but dont exactly understand it all.
perhaps give also examples of not only multi brackets playing each other but players with in the same bracket.
sorry if im making this harder than it is lol
Scire hostis animum - Intelligere ludum - Nosce te ipsum - Prima moventur conciliat - Nolite errare
Re: Pente variant proposal- ratings point gambling using a doubling cube
Posted:
Jun 19, 2008, 11:48 AM
Hey zoeyk and richardiii, Thanks for your replies in this thread (seriously). I'd like not to get too far away from the proposal at hand, however. I like rich's proposal, which I believe I've seen before (probably in the ratings thread, I think) but I'd like to hear what player's thoughts on the option of the addition of a doubling cube to any/all Pente variants would be. Not that I'm not also interested in more on Rich's proposal. But maybe we can split that out into a new thread?
Personally, after some further thinking on the doubling cube idea I have the following additions-
1.) Would you play and/or prefer to play your variants of choice using a doubling cube as an additional option? In general, I would prefer to play with it- the exceptions being when I was rusty or playing against someone who I knew was SIGNIFICANTLY better. I'd bite the bullet and play the better players with it anyway, because that's how you improve your game, improve your strategies and make your rating more accurately reflect your true skill level. I wouldn't necessarily enjoy the whuppings much, though. I'd happily prefer to play poof pente with a doubling cube (versus without) against anyone, period, no exceptions. Please, please, knock some rust off my poor poof Pente game...
2.) it's hard to tell whether this is actually a variant or not- it would certainly alter the DB of some/most/all of the variants (think D-Pente in particular) it was combined with, but OTOH it doesn't alter the rules of the game, just the relative preference for playing as P2. So, in that sense it could be viewed as more of an alterative rating system (which should be kept separate) than a variant.
3.) How much would a double be worth? Thinking further about this, I thought a resonable first approximation would be to make the initial doubling worth less than 1 additional game. The reason for this is that P1 is faced with a Hobson's choice- accept the double or resign- and if a series of sets are being played a single game with a single doubling involved shouldn't be sufficient cause to declare a winner of otherwise equivalent sets. So, let's say the first double would be worth somewhere between 1/2 a game and 2/3 of a game- enough so that 2 single doubling wins from the same P2 would count as an additional game won, perhaps slightly more than that.
But now, let's think of the case of the second and third doublings. My sense is that three doublings would be a good maximum number per game- it would leave P2 with an advantage but not add sufficient doubling potential to hugely skew the ratings- which would encourage abuse.
I think the second and third uses of the cube would count as full doubles. P1 had no good choice on the first double, but had better have a good reason to redouble. Giving it full weight allows P1 to "call a bluff", so to speak.
How would this work in practice? Minimum point gain (I believe) on a win is 3, correct me if I'm wrong, I haven't looked at the algo in a long time. Anyway, call it three for sake of example. A 3pt gain with a single doubling could become instead a 5 pt. gain. Two doublings would increase it to a 10 pt. gain, and 3 to 20 pts. A (probably rare scenario- barring manipulation) thrice doubled game would be worth nearly 7 times the value in ratings points of a standard game- however, P2 ALWAYS determines which games would fall into that category and which games stay at the standard value (or at the 3x+ level).
4.) a note about ratings manipulation- it should be obvious when if/it happens because we have an open DB record of every game played on site. The player rating system is only as good/useful an indicator of relative skill as the players allow. If a couple (or small group) of players only play amongst themselves in order to get artificially high standings in the ranking system, it can easily be spotted, pointed out and then ignored as a meaningless exercise/ private club. In other words, why should I care about some 2500+ player if I never have the opportunity to win/lose a few points against him/her? To my mind, ratings are only useful in the sense of allowing players who intend to match skills across the board to get a better sense of what they are up against. If I face a 2500+ player (inflated or not) for points I'll be wearing my lucky game hat and not be playing while tired or rusty. I expect you would be as well. And yes, I think it's possible that there are alrady some "bad apples" among the ratings. Who cares. If you play within the larger community and play against others who do, you know ratings shift within a range, which narrows (but hopefully improves until peaking out) with the number of games played. If you've ever played (online or offline, for points or not) a single other player or among a small group of players, you know that the "king of the hill" can really be pretty small beans. It's only when that "king of the hill" steps out into the wider world that their crown comes to signify (or not) anything more than "best player in the hood".
Retired from TB Pente, but still playing live games & exploring variants like D, poof and boat
Posts:
2,241
Registered:
Mar 4, 2007
From:
San Francisco
Age:
45 Home page
Re: Pente variant proposal- ratings point gambling using a doubling cube
Posted:
Jun 23, 2008, 7:21 AM
Hes right tho rich, meaning you should had posted your idea in a new thread. this thread was labeled to talk about a double cube proposal. and if he (the thread creator) wants to stay on that topic then ide suggest reposting this in a more correctly labeled thread for a more thorough discussion. just an idea.
Scire hostis animum - Intelligere ludum - Nosce te ipsum - Prima moventur conciliat - Nolite errare
Re: Pente variant proposal- ratings point gambling using a doubling cube
Posted:
Jun 24, 2008, 2:54 AM
regarding the original post,
if i was playing black against someone of skill roughly equal to my own, i don't think i would get much of a chance to use the doubling cube. the problem with this solution is it depends on the likelihood that p1 will make a big enough mistake to turn the advantage over to p2, which as skill level goes up goes very near to zero.
i find it interesting that g-pente, which i consider the fairest form of pente, is probably the least played variant of all the serious ones. i think d-pente is great but it changes the game. g-pente doesn't change the game at all, just restricts white's 2nd move, and since tourney rules already restricts it to begin with, it's not a drastic change. i think from now on, i'm gonna play g-pente only.
Re: Pente variant proposal- ratings point gambling using a doubling cube
Posted:
Jun 24, 2008, 4:16 AM
Hi Karl, Never having played Pente at the top level, I can't say how often top players would find a use for the doubling cube against other top players. However, it seems to me as though a 1900 level player would have a greater incentive to play against a 1750 level player given a way to gain more points from a P2 win than can be gained currently. Also, mistakes may be rare, but they aren't unheard of. I beat richardiii in a rated game as P2 (once upon a time) when our respective ratings were 2091 and 1584.
And as I mentioned above, the doubling cube could be used for any Pente variant (including, of course, G Pente). Perhaps top Pente (and variant) players can mentally keep track of how frequently a situation arises against opponents of any skill level where they would gain a potential advantage in ratings by using a doubling cube. It wouldn't necessarily help that much in competitions between the top players, but I think it WOULD help distribute more rating points to the upper levels. Ultimately, I think this is one of the challenges Pente currently faces, because there isn't a whole lot of incentive for 1900-2000 level players to play against 1700-1800 level players right now. An established rated player of over 2100 is about as rare as a hen's tooth right now. Compare this with 2800+ ratings for the top level chess players. Unless I'm mistaken, the general rating method for games here is based on the chess algo. Anything that helps remove the current point ceiling and better distinguishes between the yellows and the reds improves the game. But again, never having played at the top, I can't estimate how much or how little this would actually make a difference to the game of Pente. D Pente at least it would completely (if subtly) change, though.
Out of curiousity, which games do you consider to be serious versus which not serious? To my mind, all the currently available options are serious and viable variants of stone games/Pente.
Message was edited by: watsu at Jun 23, 2008 10:25 PM
Retired from TB Pente, but still playing live games & exploring variants like D, poof and boat
Re: Pente variant proposal- ratings point gambling using a doubling cube
Posted:
Jun 25, 2008, 2:53 AM
watsu,
you make some great points. i like this doubling cube idea.
the only variant i don't consider serious is poof-pente, whose name disqualifies it from being serious even before the gameplay gets a chance to redeem it (although i'm fairly certain it doesn't). Here are the variants in order of fairness:
d-pente keryo-pente g-pente pente gomoku
and in order of seriousness (as in how worthy they are of my and anyone's time):
Re: Pente variant proposal- ratings point gambling using a doubling cube
Posted:
Jun 25, 2008, 6:56 AM
Karl, It sounds as though you are judging a book (poof Pente) by it's cover (name). Speaking as the online originator of the game (though not the name of it, which dates to the over the board origns of the variant) I suggest you might want to ignore the name and give it due considration based on the merits of the game. Various red Pente players including progambler and Up2ng have found it to be a worthwhile variant which does help P2 by adding to defensive options and making games more complex- much the way keryo does, though in a different way, of course.
If you haven't already, I suggest you might take a look at the puzzle I made (with some input by Up2ng). It gives a fair idea of how different (and complex) the variant can potentially get. P2 wins in a couple of different ways given optimal play by P1, but only after 26 moves by both players have elapsed.
and if you can ignore the name, I'd be happy to play a turned based set or two against you to attempt to give you a better idea of how the variant can progress. Though some games are no different than standard Pente games, so I can't promise it will become clear in a single set.
Retired from TB Pente, but still playing live games & exploring variants like D, poof and boat
Re: Pente variant proposal- ratings point gambling using a doubling cube
Posted:
Jun 27, 2008, 4:00 PM
Cool, karl, I sent you an invite for a poof pente set.
Getting back to discussing the doubling cube- I've had at least 3 games over the last week or so (3 hammers and maybe some others) where I would have used a doubling cube had I had one. I won two of them and made a silly mistake in the other one. However, if the stakes had been doubled, I might have paid better attention and one all three, who knows?
Retired from TB Pente, but still playing live games & exploring variants like D, poof and boat
