IMO, you should decide what you want to measure first before coming up with a system to measure it 
All the formulas are nice and well intended but based on guesswork in terms of which factors are significant and what their relative weights should be. If any system was going to be designed and implemented, a more rigorous approach should be used. I imagine the data exists to be able to back these factors out so if RP ever wants to implement something like this, it should be possible to do so.
the theory behind the formula is that tournament skill and cashgame skill, are both counted about equally, also there is a cash and win multiplier, which can be between 0 and 100% according to the ratio of how much cash and wins you got, obviously since no cash and no win is way more frequently happening this is also counted into it, the rarer the wins/cashes the stronger it counts, also when taking the average like you would normally do, there will be a flaw in the system that makes the big wins count less and the more you play the more this will affect it, also for this problem i have thought of a additional calculation that will erase this problem. at last we have a chip multiplier, this is based on the ranking we do have right now, this will increase the points you have by a certain % based on how much chips you have, this because chips are not the only thing that matters but it is certainly one factor that matters.
on top of those things the first 2 are + points and the other 3 are x points. this beause if you only have +, it wonāt be about good play anymore but only about the amount you play. and if there is only x, it wonāt work anyway. also there is not only + but also - to earn, this is also to make good play favorite over much play, because if you play a lot but play bad, there will be more - points then + ponts and it will give you a negative score. there is another reason why there are much x instead of + points, this is because x makes the points count more equal then + does, and since the bankroll sizes are so much different, even a good formula based on only + and - will favor chips way over all else.
so the short version of this is: the things measured in my formula are cashgame skill, tournament skill, amount of cashes, amount of wins and bankroll size. and all have special formula infused to keep flaws out of the system.
i can understand that part, but what kind of stuff would you also add and why?
Yiz,
Now this 3-4 threads , same topic. The easiest way to please everyone is to have in settings ācriteraā for a "Custom Rank : " line, for each player. The means, in settings I checkoff criteria, then in a line for ācustomāā¦ it would let me see everyoneās rank based on the criteria I picked. No other player would have the same criteria, nor would know how I rank everyoneā¦ personally I like the approach of Replay having a formula that they think is important also, which could be in many/multiple forms.
donāt know if i get it wrong, but you mean you pick random stuff and make a formula of it which involves a ranking and donāt tell it afterwards which stuff is involved?
if i got it right, what would be the idea of picking random stuff? as it is meant to be as best as possible to provide the skill part of poker.
second thing is, what would be the use of not telling which things are involved, that way you canāt improve yourself when doing something wrong.
if i donāt understood well, explain me what i miss plz.
Iāll take this over to the thread for it for a more complete answer but basically what Iām saying is that all the formulas put forward include likely significant criteria but nothing that has been back-tested to see whether is actually significant or not. Even when we have intuited a significant factor, unless we run the data, we donāt know how strongly to weight them relative to each other.
This is a really cool line of inquiry for me and would love to go over it and see if its even possible to come up with a formula that makes sense. My interest stems from the research I did on equity pricing formulas back when I was pursuing my doctorate. You have to go into the piles of data to find which factors have the most explanatory power. Some factors overlap and influence others so then you have to adjust for those issues.
Of course, Iām not even sure a proper formula for āskillā can be found at all but the search itself would likely be worthwhile.
I think this is a cool concept if Iām getting what you are saying. Basically you are talking about setting up a screening system whereby a player could input the criteria he/she feels are most important and have the universe of players here be sorted based on those factors. People could use the factors they find most important and predictive to sort through and rank the universe of other players by their own custom set of factors. If wouldnāt be a public ranking system but could be useful nonetheless.
what if we came up with a balance of how much a player has won or lost in all of their games combined - weāll call it āchipsā for short - and then we just measure who has the most āChipsā
You might be on to something here. The simplicity of the idea smacks of pure genius
I think the poker skills used in MTTS and Ring games are very different. I donāt think you can have a common skill rating combined. It would have to be a rating for each. I prefer MTTs and would like to see a skill rating on my opponents, but only for MTT play. I donāt care about their ring game rating.
Also, I saw one proposed formula in an earlier post where āaggressivenessā was a criteria in the rating. I think agressiveness can be show of skill when used at certain times, but likewise at other times might show a lack of skill. You couldnāt just base it on how often someone raises.
I like all the ideas and Sarah may be onto something with a menu we could used to develop a rating using certain factors and omitting others.
I think this is correct. You may be able to have a combined ranking but it wouldnāt be as accurate as a ranking for the specific discipline. To go a bit farther down this line of thought, it would be hard to compare a heads-up specialist against an MTT player or 6-max SnG player. Any ranking system is going to have to make decisions on what it is measuring and how specific it wants to get. That is why it is both a difficult and interesting topic for me.
I really do like Sarahās concept with the 1 caveat that I would still want an opt-out for players who do not want their hands and play available for analysis. I will always start with the assumption that it has to be up to the individual player how much of their information they wish to share, if any.
As 1 final note here - I do think the single most valuable tool RP could provide to players is the ability to gauge our own play over time. Integrating an analytics program or making the hands played here exportable to 3rd party analytics programs would be of great help in terms of training and learning. Competing against others is certainly great but I think it is just as important to be able to compete against ourselves and improve our games over time. The only real way to do that is to review hands played and learn from mistakes we make. That ability would take the focus away from small-samples of results and allow players to look at their game on a macro-level.
4 threads, not helpingā¦
this 1 - this thread
player ranking, not ur momas army boots - ??
New player ranking calculation - S_Sarah
another player ranking calc - Yiaz
I will respond to this, in my thread.
i do fully agree on this one, but the problem is, how can we test this without actually changing the ranking?[quote=ā1Warlock, post:46, topic:6209ā]
Of course, Iām not even sure a proper formula for āskillā can be found at all but the search itself would likely be worthwhile.
[/quote]
also fully agree on this one. i do also think there is no such thing as the perfect skill system, skill is a very very wide thing, and the only thing we can do is minimize the possible flaws in the rank and maximize the stuff involved in it.
i get there is a part of this thatās likable in the way of seeing which parts are going well and which need improvement, but while i like this as an extra addition, i really donāt like this replacing the system, this is because it will erase the ranking instead of replacing it, since there wonāt be a number 1 player, 2,3,4,5 etc. anymore, which means the public ranking what we all want to improve, will be fully erased instead.
isnāt that the same as we have right now? the bankroll is already all chips won and lost combined. donāt know if i understand wrong, but if so lets hear it .[quote=āSeville, post:50, topic:6209ā]
I think the poker skills used in MTTS and Ring games are very different. I donāt think you can have a common skill rating combined. It would have to be a rating for each.
[/quote]
i do agree on this one since they are very different, but i have also considered this in my ranking formula, i have gave ring and tourneys both + points while the other criteria got x points. the fact that they both have + means it wonāt matter if you choose to play ring, tournaments, or both, since it will add points both on the same way. the only flaw is that siince they are so different, they both have a different way of adding points. i do have tried of making them as close as possible however but they can never be 100% the same.[quote=āSeville, post:50, topic:6209ā]
Also, I saw one proposed formula in an earlier post where āaggressivenessā was a criteria in the rating. I think agressiveness can be show of skill when used at certain times, but likewise at other times might show a lack of skill. You couldnāt just base it on how often someone raises.
[/quote]
agreed. i have also mentioned this point on the thread.
really great idea![quote=ā1Warlock, post:51, topic:6209ā]
You may be able to have a combined ranking but it wouldnāt be as accurate as a ranking for the specific discipline.
[/quote]
i agree.
at last i like to say that for the ranking as we use it now the best possible thing we could do is getting a formula to getting as close to the true skill as possible and use this as the main/public ranking.
and as a additional thing the side rankings we talked about of a specific discipline will be also a great thing to add because like you said this is the only way to make it 100% accurate.
long story short: as ranking i think the best would be all things considered and as close as possible to the truth.
and i also like the idea of rankings of a single criteria, but only as a side rank, not the main one.
ā¦
if anyone has questions about these opinions and/or the ranking feel free to ask.
I still think what we really need is access to a more robust set of raw data.
Having a single skill rank number isnāt all that useful. For example, if you were ranked 500th, how would this help you get to 300? Would a single number tell you you are folding too much on the flop? Playing too many hands from early positions? Seeing too many flops from the small blind?
Everyone is stronger in some areas than in others. Let me see your whole stats, and let me figure out where you are weak and where you are strong.
Providing access to the complete stats would be much more useful, both when trying to assess your opponents and when looking for leaks in your own game.
true but what you mean are statistics instead of a ranking, i know you are talking about stats, but iāll specify it some more:
the ranking will be based on how well you played, the statistics show how well you played.
while this looks a lot like each other, there is a difference. the statistics and ranking are both based on raw material you talked about, the raw material infused in a (good) formula makes a ranking, the raw material infused with explanation and details makes statistics.
but i fully agree both are very good things, and statistics are very important to improve your game further, there are already statistics on replay, but there are very few, an expansion of it would be very useful.
My thoughts in regard to a ranking system are much simpler (KISS) and so, Iām thinking, perhaps stand a better chance of being implemented, for one thing:
Add another stat, publicly displayed, showing oneās average finishing position in games. This would include single table games and MTTs. Similar to how tournament points on are used for leaderboards, but basing it not on points or just the top 50%, simply the finishing position, from 1st place down to last place, and combine all games played to provide the average. These averages could then be used for RANK. One issue would be games in which there is no single winner, like ticket games, where there can be several ā1st placeā ticket winners.
The āRANKā could be divided up into different categories, such as SnGs, MTTs, and further, if so desired (to complicate things).
Not perfect, but perhaps better than no skill based ranking system at all.
This of course would not work for ring games, but those are kind of a separate category anyway, and treated as such throughout the site.
Yiaz,
I have read every reply in all 4 posts, others havenāt ā¦ I keep seeing duplication, plus your post has already dropped so low its off the radar kindaā¦ who cares whoās post is @ the top, I donātā¦
Seriously, we should have them somehow merged, and its a *itch to try and slog thru ANY long topicā¦ this damn forum uses unlimited scrolling rather than pages, once a topic gets too long its almost unreadable, as well as each catagoryā¦ unlim scrolling is DUMB and a pain in the ass.
i think you misunderstand what i meant, perhaps i havenāt specify my point well enough.
here are the details:
to make a long story short, i fully agree with you, but i just donāt like to see problems get replaced with other problems. but if you have a solution to solve this, i go without a doubt for the merge.
Yiaz,
naaaaā¦ leave it as isā¦ Iām just worried all this time spent, is for nothing.
oh ok, if you, me or anyone else may think of something, i would still like the idea
Thatās one of the reasons I was questioning whether there should be a āRANKā. It could at least indicate, in a very general way, how much need there is for improvement, compared to others. Also tracking rank, over time, could show if there has been any improvement.