ELO on luck based games

[FrankJones]

....

Also w/ reduced k factor as an alternative solution (the recommended was 5. 4x lower than that of other games) trying to resolve the same issue it would take way more games to reach these levels. Thus causing issues when trying to filter for players bc there could be solid players ranked low that just didn't play hundreds of games yet. You've mentioned this before Frank in this thread or another but also it is more fun to actual see the ranking move with progress than and more relative to other games on the site rather than only a few points or even decimals at a time.

Agreed; I dislike tinkering with the k-factor. I want a high k-factor that brings people to an accurate Elo faster rather than slower.

Going back to the theme of "different people like different this" - Personally, I dislike high luck games, generally speaking. I try to avoid them. So, the BGA Elo formula doesn't bother me.

That being said, I can relate to what you are saying.
Open-face Chinese poker has even high variance than CaptainFlip. It might be the highest-variance game on this site.
There have been occasions where the Elo#1 player had exactly 300. I've never seen anyone get near 400. I've never seen higher than 327.
(Okay, current leader is at 343, but that is not sustainable. 2nd-10th are rated 265-308.)
If I were not removed for inactivity, I would be 3rd.

Do I like the fact that I often lose to players who are weaker than I am at the game? No. (As far as I am concerned, I'm the best player on the site). My solution? I rarely play. When I do play, I go to every effort to avoid playing anyone rated below 250, because the way the game is structured, the better player probably struggles to win 65% of the time. My own win rate is less than that. I'm at 64.2%.

And the route to 300 was frustrating, for all the same reasons. I needed to get on a bit of a lucky streak to get near 300, and then I would lose once and go crashing back down to 280. Frustrating. So, I rarely play.

It should be noted:

The reason for the high variance in this particular game is the absurdly low number of hands played in a "game". On BGA, a "game" of open face Chinese poker consists of 6 hands. The variance and luck from one hand to the next is extreme, but in the long run, the better player will win more points. Problem is, winning one "game" of 6 hands by a score of 180 to 100, and then losing the next "game" of 6 hands against the same opponent by a score of 100 to 99 still results in a "win" and a "loss", even though one player is ahead by 79 points (which is a HUGE amount for 12 hands). This is not a game meant to be played in tiny sets of 6 hands. It is similar in absurdity to having a hearts or spades match consist of one hand. Highest score after 1 hand wins !

There is an option to play this game to 12 hands, but barely anyone uses it, and I've even seen a suggestion that 6 hands is too many ( !???). A hand takes a minute or two. It's insane to me that people cannot be bothered to play 12 hands and some think 6 is too many.

On various apps on which this game is available, matches are set to 100 hands. Even then, a notably weaker player can win 35% of the time.

What is my point in this digression?
I suppose my point is, games with so much luck that the standard Elo formula fails are:

1) Poorly designed games (my opinion), and
2) Not my preferred type of game (fact).

[ChiefPointThief]

I strongly disagree with ofcp being even close to the highest variance game on this site. Even with 6 hands. A factor with elo is not only luck factor but player pool. For instance two similar games with high variance I’ve played and reached high 200s which was good enough for number 1 are Golfie & great American fox hunt. Both of these extremely lucky games. But if the player pool was a little higher I’m sure I could reach higher elo in both

I also knew ofcp before bga and thought it was to go to an amount of points (100) not a specific amount of hands. So that was different when I played it here

[FrankJones]

I strongly disagree with ofcp being even close to the highest variance game on this site. Even with 6 hands. A factor with elo is not only luck factor but player pool. For instance two similar games with high variance I’ve played and reached high 200s which was good enough for number 1 are Golfie & great American fox hunt. Both of these extremely lucky games. But if the player pool was a little higher I’m sure I could reach higher elo in both

I also knew ofcp before bga and thought it was to go to an amount of points (100) not a specific amount of hands. So that was different when I played it here

Playing to 100 point is also a joke, and that can actually end the game SOONER than 6 hands. In fact, the game here is ALSO played to 100 points. Each player starts with 100 points, and if one player gains 100 points (which can happen in 3 hands), the game ends WITHOUT even finishing all 6 hands.

You really think OFCP is "not even close" to the highest variance game here ??? Is this site filled with silly high-luck games where it's a struggle for anyone to get to 300 Elo?

How well do you know OFCP? I'm an elite level expert player, and I am very confident in saying that the variance in that game is HUGE. Even without the 6-hand or 100-point limitations, it's a massively high variance game. Hundrds of points can swing in a very short number of hands, and yet the difference between a millionaire and a break even player is 1.0 points per hand.

Someone who can maintain a 1.0 point-per-hand win rate can make a million dollars playing OFCP in one year. Someone who averages 0.0 points-per-hand rate will make ... zero dollars.

A single lucky/unlucky fifth street draw for both players can swing 60+ points, and yet massive profitability at this game can be achieved by ... outplaying your opponents by ONE point per hand in the long run.

That's not "high variance" to you?

[ChiefPointThief]

You may know more about ofcp than I do. But yeah. I would say not even close to the highest variance game. I put together big themed team events. One of them was all luck based games. I was way more confident going into my ofcp game than other games. Even just the 3 I mentioned above I’m more confident in ofcp (would still take captain flip in 2p)

[FrankJones]

You may know more about ofcp than I do. But yeah. I would say not even close to the highest variance game. I put together big themed team events. One of them was all luck based games. I was way more confident going into my ofcp game than other games. Even just the 3 I mentioned above I’m more confident in ofcp (would still take captain flip in 2p)

Can you help me a bit here? Perhaps with a list of 10 games here on BGA you consider higher variance? (I'm unfamiliar with most games here on BGA, so yes, it's quite possible there are a lot of higher variance games, I am merely skeptical because of my extensive knowledge of OFCP. I've played hundreds of thousands of hands. )

[ChiefPointThief]

You may know more about ofcp than I do. But yeah. I would say not even close to the highest variance game. I put together big themed team events. One of them was all luck based games. I was way more confident going into my ofcp game than other games. Even just the 3 I mentioned above I’m more confident in ofcp (would still take captain flip in 2p)

Can you help me a bit here? Perhaps with a list of 10 games here on BGA you consider higher variance? (I'm unfamiliar with most games here on BGA, so yes, it's quite possible there are a lot of higher variance games, I am merely skeptical because of my extensive knowledge of OFCP. I've played hundreds of thousands of hands. )

These are just some I snatched off my profile. Theres a ton!

Via magica
99
Flip 7
Hugo
The great American fox hunt
Yahtzee
Llama
Rolling pins
Go fish
Golfie

[FrankJones]

I'll give you Go Fish and Golfie.

Fox Hunt has only 3300 games played, which I think is far too few games to draw meaningful conclusions. I don't know the game, but low Elo ceiling indicates it's very possibly a higher variance game than OFCP.

The other 7 games all have higher Elos and Elo ceilings than OFCP, which makes me question exactly what makes them so much higher variance to the point where you would say "OFCP is not even close".

[ChiefPointThief]

Several of these games are sit down and play games. You can lose to someone who has never played before on their 1st game. Firstly you need some understanding of how poker works. With a game like flip 7 you can sit your child down at the table and possibly lose to them. I am more confident that I will come out victorious if I sit a child down for the 1st time and play ofcp with them. As mentioned you can make millions off of ofcp I can assure you no one can make millions off of 99 and tons of other games.

Using what is the current highest elo to determine the variance of a game is faulty. There are other contributing factors. For instance, sobek 2p I’m good at. My ceiling was 600. Currently 500. When I got to 600 that is so far from the rest of the field that eventually I will fall. Certain points in each game w/ luck factors reaches this point mathematically. I’m very certain that if this game had a player pool the size of azul I would be a “master” at it because the elo would be in the field to more easily obtain this. But that doesn’t change how luck based the game is. It would still be a 3 luck game.

[FrankJones]

Several of these games are sit down and play games. You can lose to someone who has never played before on their 1st game.

I understand now. I didn't realize these games were of this nature. To me, these re not serious games, and not really pertinent to the discussion. So, I'll grant you that games a child could win playing for the first time most certainly have so much luck and so little skill that they have higher variance than OFCP.

My fault. I should have been more precise in my phrasing. OFCP (The pineapple variety, since non-pineapple OFCP is a very weak version of the game) is likely one of the most high-variance games on BGA among the subset of games that are also high-skill games.

Sure, a game of "war" (the classic card game) has higher variance, because it has zero skill. There are no decisions to make. Flip over the card and see who has the better card.

Firstly you need some understanding of how poker works.

Not exactly. To play OFCP, one needs to know definitions of poker hands, and the point-values assigned to them in OFCP. This has little to do with how actual poker works. Poker involves betting, folding, bluffing, bet-sizing, hand-ranges, and various other nuances. OFCP has none of these things. OFCP is not a poker game. It is a math strategy game that uses the definitions of some poker hands to keep score (in a way that is not applied in any version of actual poker games).

[That last portion was written just to clarify to others, who may not know OFCP, what OFCP is and is not. I'm well aware that you already realize all this, having played the game yourself.]

OFCP is an odd game in that is has a very high skill level and steep learning curve, yet still has tremendous variance from one hand to the next. It's also a game in which you can make the mathematically perfect play and still lose to an opponent who plays his cards blindfolded. (unlikely, but possible). I mean, a perfectly played hand can still foul and an opponent's horribly played hand can make something absurd like trip 3's up top after foolishly putting three 3's on the top row, which is an asinine play.

With a game like flip 7 you can sit your child down at the table and possibly lose to them.

To all their own, so to speak, right? I cannot imagine why anyone would want to look at a computer screen and play such a silly game. I could see doing that while having a few beers and listening to music and socializing. Plenty of silly non-strategy games fall into that category. To me, playing a game like that on the internet would about as interesting and fun as playing beer-pong on the internet.

I am more confident that I will come out victorious if I sit a child down for the 1st time and play ofcp with them. As mentioned you can make millions off of ofcp I can assure you no one can make millions off of 99 and tons of other games.

A person could make millions playing OFCP IF that player can win 1.0 points-per-hand in the long run, against opponents willing to play $50 a point. This is extremely difficult to do.

And yet, it's possible to win 60 points in a single hand. In fact, it just happened today. I was up 61 points after 1 hand in a 6-hand match, and my opponent conceded before we finished the 2nd hand. I mentioned "become a millionaire by winning 1 point per hand" merely to illustrate the stark contrast between the minuscule per-hand differential needed to make a fortune with the staggering number of points that can be won or lost in a single hand, let alone a match of 6 hands. I used to play 400 hands a day and still have some bad days, despite generally being one of the best players on any platform on which I have played the game.

Actual poker is similar in that it is a game of high skill - especially a game like pot-limit Omaha - and yet, even strong players can go on lengthy cold runs that last weeks. High-level professional poker players can have losing months, despite having made millions in their careers.

Using what is the current highest elo to determine the variance of a game is faulty. There are other contributing factors.

I don't think it's all that faulty. There is a high correlation between Elo ceilings and skill-factor of the games.

OFCP has a reasonably sized player pool here. It's not for lack of players that OFCP pretty much caps out around 300 Elo. It's because the variance is too high in the game itself, exacerbated by the absurdity of calling a game "6 hands" or "100 points", which would be like playing 6 hands of Texas Hold'em and giving a win to whoever has the most chips after 6 hands.

For instance, sobek 2p I’m good at. My ceiling was 600. Currently 500. When I got to 600 that is so far from the rest of the field that eventually I will fall. Certain points in each game w/ luck factors reaches this point mathematically. I’m very certain that if this game had a player pool the size of azul I would be a “master” at it because the elo would be in the field to more easily obtain this. But that doesn’t change how luck based the game is. It would still be a 3 luck game.

That's a rather bold claim.
There are games here that have a reasonably high skill-to-luck ratio AND a large player pool that do not have any master-level (700 Elo) players.

Also, "luck 3" is not a well-defined term. None of those numbers are. What does that even mean? To me, the amount of luck in a game is defined not by some relatively arbitrary number from 1-5, but rather, by empirical evidence of the possibility and difficulty of having any players reach specific Elo thresholds, such as 500 or 700.

[FrankJones]

I also knew ofcp before bga and thought it was to go to an amount of points (100) not a specific amount of hands. So that was different when I played it here

We may have played on some of the same apps. One very popular app has an option to play a match to 100 points. But this is also silly. It could end before 6 hands (I've had this happen), or it could drag on for 120 hands (bizarrely, I've had that happen too).

The true measurement of skill is to play thousands of hands and then evaluate. On the app I play on (the one I referenced in my first paragraph), rather than set the match to 100 points, we just set it to "ongoing indefinitely". I had 4 concurrent games going against a very strong player, and at one point I was up 1500 points. This amounted to less than 1 point per hand, but at $50 a point, he would have been minus $75 thousand to me. And playing 4 concurrent games on an ap doesn't take long to get to 1800 hands, or however many hands it was that we played when I was up 1500 points.

Actually, I just checked. I'm still ahead by 1600 points, but we have played 4500 hands. So, I am +0.355 points per hand, and yet would still be ahead $80 thousand if playing $50 a point.

That is very high variance. Just because a child playing for the first time could not win does not make the game any less high-variance.

That being said, if one were to explain to a new player the rules (which could be done in a few minutes by simply showing a single-sheet printout of the definition of each poker hand and the points those hands score in OFCP), that new player most definitely could win his first hand, or his first match to 6 hands.