Meta-Analysis Project: Help Wanted

[Irieguy]

With Big Brother squashing all its little brothers and forcing rakeback players to go back to work at Burger King, on-line poker has just undergone the most rapid and significant change since its birth.

I'd like to know what the state of SNGs is right now in terms of profitability. But I can't play 5000 SNGs in a month to find out.

I would like to gather a collection of SNG players who play at similar levels, with similar skills, under similar conditions and see how we do as a group over 3000-5000 SNGs between now and Christmas.

If I can get enough players, I would like to have two groups: an 800 chip group and a 1000 chip group.

Player Criteria:

1. You have beaten either an 800 or 1000 chip PartyPoker SNG structure over your last 1000 SNGs.

2. You can play a minimum of 500 SNGs between November 1st and December 25th. (I will accept data from Nov. 1st onward as long as you have been meeting all criteria and have been recording every SNG played.)

3. You know how to track your results consecutively without error or exceptions.

4. You play at least 4, and no more than 8 SNG tables at once when you play.

I will collect all of the data and combine it into a large results pool. The primary measure will be $/SNG.

I will collect the results anonymously. I am going to devise a reporting system that will allow me to inspect e-penises wihout knowing whose penis it is.

My goal is to have a reasonable understanding of how much $/SNG a skilled 2+2er can make while multi-tabling the 800 chip and 1000 chip SNGs on Party right now.

If you want to participate, PM me.

Irieguy

[microbet]

Interesting project. What if the previous 1000 had been played at multiple different buyins and the next 500 are also expected to be played at different buyins?

[Irieguy]

[ QUOTE ]
Interesting project. What if the previous 1000 had been played at multiple different buyins and the next 500 are also expected to be played at different buyins?

[/ QUOTE ]

As long as it's the same number of starting chips, that's fine.

I want people who are committed to playing a bunch of SNGs at a fixed chip-structure over a short period of time. If you move between $22's and $33's, for example, that's fine. $/SNG is going to be very similar anyway between those two levels.

Irieguy

[Irieguy]

Cool, 10 volunteers within the first hour.

I may be able to get over 10,000 SNGs of data at the 800 chip level and 5,000 at the 1000-chip level.

If you are a winning long term player who meets the above criteria, please contact me. I will get back to everybody within a few days.

Irieguy

[USCSigma1097]

Irie,

I'd be happy to help in whatever way possible but I 10 table, as I suspect that many other lower stakes SNG players do. Do you feel that ROI drop is statistically significant from 8 to 10 tabling as to not include that with your data pool?

Sigma

Why aren't you taking as a data item people's ROI before the drop?

[Big Limpin']

This project, collective sampling, has undoubtedly been on the minds of many of us, so having your name behind it should assure a good return. I'll offer ~2k by xmas.

Could we answer another question with this, of the earn rate difference between the "decrepid" skins and Party proper?

[Irieguy]

[ QUOTE ]
Irie,

I'd be happy to help in whatever way possible but I 10 table, as I suspect that many other lower stakes SNG players do. Do you feel that ROI drop is statistically significant from 8 to 10 tabling as to not include that with your data pool?

Sigma

[/ QUOTE ]

10+ tablers make up too small a percentage of SNG players.

Irieguy

[ QUOTE ]
[ QUOTE ]
Irie,

I'd be happy to help in whatever way possible but I 10 table, as I suspect that many other lower stakes SNG players do. Do you feel that ROI drop is statistically significant from 8 to 10 tabling as to not include that with your data pool?

Sigma

[/ QUOTE ]

10+ tablers make up too small a percentage of SNG players.

Irieguy

[/ QUOTE ]

Why does this matter? They're still playing SnGs and there's no significant difference between how they play and how 8-tablers play.

Also, why aren't you taking as an input people's ROIs over their last 1000 before the switch? Don't you want to know the change in ROI?

[Irieguy]

[ QUOTE ]

Why does this matter? They're still playing SnGs and there's no significant difference between how they play and how 8-tablers play.

[/ QUOTE ]

That is totally untrue in my experience.

[ QUOTE ]


Also, why aren't you taking as an input people's ROIs over their last 1000 before the switch? Don't you want to know the change in ROI?

[/ QUOTE ]

There are lots of things I want to know.

The purpose of this analysis, however, will be to determine how much $/SNG a group of winning SNG players earn over a very large number of SNGs between Nov. 1 and Dec. 25.

It is best to keep the focus and scope of this thing very narrow, for a host of reasons.

Irieguy

[DMACM]

I can participate probalby 500-1000 22s.

[Irieguy]

Send me a PM so I can have you on my PM list for instructions.

Irieguy

[ QUOTE ]



3. You know how to track your results consecutively without error or exceptions.



[/ QUOTE ]

I don't meet any of the qualifications and I don't mean to hijack your post but would somebody be willing to tell me how to do this. PM me if that is best.

Thank you

[Indiana]

I'm assuming you have a statistician with a strong background in the caveat's of Meta-Analyses? Are you familiar with propensity scoring?

Indy

[microbet]

One thought is that eventhough it is only a short time between Nov. 1st and whenever you start, you might be hearing more from people who have started out well than otherwise.

[axeshigh]

[ QUOTE ]
One thought is that eventhough it is only a short time between Nov. 1st and whenever you start, you might be hearing more from people who have started out well than otherwise.

[/ QUOTE ]

Just as in every study made in the history of the universe, there is a bias in the selection of the subjects.

[Indiana]

Actually, Irie. This is not a Meta-Analysis. Its just an observational study. An example of Meta-Analyses is when you are combining many "controlled" clinical trials. Here you are just combining data from each player where nothing is controlled within player.

Indy

[axeshigh]

[ QUOTE ]
Actually, Irie. This is not a Meta-Analysis. Its just an observational study. An example of Meta-Analyses is when you are combining many "controlled" clinical trials. Here you are just combining data from each player where nothing is controlled within player.

Indy

[/ QUOTE ]

You're nitpicking like mad today...

[Indiana]

Well, you call it nitpicking, I call it statistics. What Irie proposes won't prove anything. Its just an observational view of some data. Only hypothesis driving, nothing else.

Indy

[mmbt0ne]

Yeah, if you could just never post again, that would be great. I've only got 1 more class in this stuff, and it's hurting my head just to think about one more semester of it. /images/graemlins/frown.gif

[splashpot]

[ QUOTE ]
One thought is that eventhough it is only a short time between Nov. 1st and whenever you start, you might be hearing more from people who have started out well than otherwise.

[/ QUOTE ]
I volunteered and I've had a terrible start to Nov. Possibly the worst start to a month I've ever had.

[Irieguy]

[ QUOTE ]
One thought is that eventhough it is only a short time between Nov. 1st and whenever you start, you might be hearing more from people who have started out well than otherwise.

[/ QUOTE ]

That's a very good point. But selection bias is going to be horrible here no matter what I do.

My goal here is not to run an air-tight statistical analysis. It's to combine data sets from winning players who volunteer before they play their sets (or most of their sets, as you point out.)

What we'll have in a few weeks is a set of data that looks like it will cover well over 20,000 SNGs from a relatively small and somewhat similar group of winning players.

It will give us a better idea of how much money you can make playing SNGs on Party right now than any other way I can think of.

It's just data. It is going to be helpful to me in a number of different ways and I think it will be helpful to others.

Irieguy

[microbet]

[ QUOTE ]
[ QUOTE ]
One thought is that eventhough it is only a short time between Nov. 1st and whenever you start, you might be hearing more from people who have started out well than otherwise.

[/ QUOTE ]
I volunteered and I've had a terrible start to Nov. Possibly the worst start to a month I've ever had.

[/ QUOTE ]
Sincerely, that is highly commendable.

[Irieguy]

I have 35 volunteers so far, pledging over 22,000 SNGs of data over a 7 week span of time. Now I'm looking to get a pool of over 30,000 SNGs.

My biggest fear is that the people who lose will not report their data. Hopefully, the fact that I will be collecting the data anonymously will help with that.

Thank you all for your help, and thank you in advance for maintaining the integrity of this data set by agreeing to report if you lose.

Irieguy

[pergesu]

[ QUOTE ]
My biggest fear is that the people who lose will not report their data. Hopefully, the fact that I will be collecting the data anonymously will help with that.

[/ QUOTE ]
I'll probably get broke, but I'll be willing to at least publicly humiliate myself as such.

[Irieguy]

[ QUOTE ]

I'll probably get broke, but I'll be willing to at least publicly humiliate myself as such.

[/ QUOTE ]

There is no public humiliation or celebration. This is important.

I will not know who you are when you report your data. It is important for everyone to realize that I am going to collect this data anonymously. I will only know who the group of players is... not their individual results.

Irieguy

[johnnybeef]

[ QUOTE ]
[ QUOTE ]
My biggest fear is that the people who lose will not report their data. Hopefully, the fact that I will be collecting the data anonymously will help with that.

[/ QUOTE ]
I'll probably get broke, but I'll be willing to at least publicly humiliate myself as such.

[/ QUOTE ]

aww pergesu, you are just saying that because you suck /images/graemlins/blush.gif

[davehwm]

I don't like the idea of taking different numbers of sets from different groups of people with different skill levels. I don't agree with the assumption that all of these people "play simliar". And this is just the beginning of the many problems with this experiment.

I wouldn't put too much time into it, because I don't think you're going to be able to draw many accurate conclusions from this set of data, honestly.

[Indiana]

Thank you davehwm. I've been trying to tell him this but u do a better job with articulating it.

Indy

[Melchiades]

When you see someone played 100 000 SnG's, won't you know its raptor?

[davehwm]

[ QUOTE ]
Thank you davehwm. I've been trying to tell him this but u do a better job with articulating it.

Indy

[/ QUOTE ]

Thanks, but I actually don't think I'm articulating it very well, myself. It just screams waste of time, to me.

Basically, you can't assume cause and effect at all from a study like this. The statement: "The split of the party poker skins caused a decrease in $SNG/hr" would be completely flawed, statistically and logically based on a set of data like this.

I think if this is cross posted in the Probability forum, some of the stat geeks would be able to give a multitude of reasons why it's not statistically sound. I could start, but I don't feel comfortable enough with all the terminology (it's been quite a while since I've takend research methods or worked in a lab).

[microbet]

I don't think he's trying to draw many accurate conclusions. I think he's trying to get into the ballpark on one specific thing.

[davehwm]

[ QUOTE ]
I don't think he's trying to draw many accurate conclusions. I think he's trying to get into the ballpark on one specific thing.

[/ QUOTE ]

Which is?

[microbet]

I'm just pulling this out of my arse, but maybe the current profitibility of SNGs at party.

[davehwm]

[ QUOTE ]
I'm just pulling this out of my arse, but maybe the current profitibility of SNGs at party.

[/ QUOTE ]

I don't think he can get an answer to that question without a pretty freakin' large margin of error based on the statistical flaws in the data collection process. He will most likely find that they are still profitable. He will not find a number that is in the ballpark, as to how profitable they are.

[Irieguy]

[ QUOTE ]
[ QUOTE ]
I don't think he's trying to draw many accurate conclusions. I think he's trying to get into the ballpark on one specific thing.

[/ QUOTE ]

Which is?

[/ QUOTE ]

When I am finished, you will know how much money a group of previously successful SNG players earned over 30,000+ SNGs in a short period of time.

I am not going to beg anybody to find meaning in that. If you feel that you are better off by not even knowing the results, then you should not look at the data. If you feel that you are better off by knowing, or if you are in any way curious... then you can look at it because I will post it.

That is all. This is not complicated, nor have I said I am setting out to prove anything.

I am collecting some data. I sincerely apologize if this angers some people.

Irieguy

[sofere]

I think its an interesting experiment. I don't think its one I'm going to draw any hard and fast conclusions off of though.

The goal of the experiment was to "have a reasonable understanding of how much $/SNG a skilled 2+2er can make while multi-tabling the 800 chip and 1000 chip SNGs on Party right now."

It will not (and I don't think its Irie's intention) to come up with an average of 2+2 ROI. It also does not seek to achieve a range of ROI at which one could expect to be.

If say the average over the 10k SNGs is 20% ROI, then it is reasonable to assume that a skilled SNGer can possibly obtain an ROI of 20% at the time of the experiment.

To me though it would basically be a nice way to measure my e-penis in the privacy of my own home without having everyone else glare at me and flame me for it.

Moral of the story, you whip out your e-penis in public, you wind up with e-herpes all you're left with is a flaming e-penis.

[microbet]

Well, the diffences in opinions on this issue make for a pretty big ballpark. I'm certainly not saying it will definitely show anything, but it might. Maybe he'll get 100,000 sngs. Maybe the results will be surprising in some way. Anyway, I don't mean to fight, I'm just phenomenally bored with my job.

[davehwm]

[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
I don't think he's trying to draw many accurate conclusions. I think he's trying to get into the ballpark on one specific thing.

[/ QUOTE ]

Which is?

[/ QUOTE ]

When I am finished, you will know how much money a group of previously successful SNG players earned over 30,000+ SNGs in a short period of time.

I am not going to beg anybody to find meaning in that. If you feel that you are better off by not even knowing the results, then you should not look at the data. If you feel that you are better off by knowing, or if you are in any way curious... then you can look at it because I will post it.

That is all. This is not complicated, nor have I said I am setting out to prove anything.

I am collecting some data. I sincerely apologize if this angers some people.

Irieguy

[/ QUOTE ]

If you are just collecting data, that is fine. It will not come close to showing you the profitability of SNGs, though. It will be a number that is flawed statistically because of the different sample sizes for each player and the skill difference between these players (and the levels they are playing at).

If I measure the batting averages for Barry Bonds over 500 at bats, Andruw Jones over 750 at bats and an above average minor league player over 1000 at bats, then average these for a batting average, I learn nothing. Especially if no one else observes these at-bats and I ask the baseball players to report the data to me themselves.

P.S. I'm not angry at all. I thought it was a pretty civil discussion.

[sofere]

I would love to see this separated out by buy-in, in addition to as a composite. The sample per buy-in would probably not be significant, but I'm just curious.

I also don't think it would be that much more work.

[faquewdikhed]

[ QUOTE ]

If I measure the batting averages for Barry Bonds over 500 at bats, Andruw Jones over 750 at bats and an above average minor league player over 1000 at bats, then average these for a batting average, I learn nothing. Especially if no one else observes these at-bats and I ask the baseball players to report the data to me themselves.


[/ QUOTE ]

What if all baseballs had 1 pound added to them, and the cumulative average dropped by .200. Sure it wouldn't mean jack [censored], but it would still be interesting.

[microbet]

[ QUOTE ]
[ QUOTE ]

If I measure the batting averages for Barry Bonds over 500 at bats, Andruw Jones over 750 at bats and an above average minor league player over 1000 at bats, then average these for a batting average, I learn nothing. Especially if no one else observes these at-bats and I ask the baseball players to report the data to me themselves.


[/ QUOTE ]

What if all baseballs had 1 pound added to them, and the cumulative average dropped by .200. Sure it wouldn't mean jack [censored], but it would still be interesting.

[/ QUOTE ]

Yeah, and what if you juiced the players, corked the bats, livened the balls and shrunk the stadiums? It wouldn't mean jack, but you'd sure get a lot of HRs. (from metaphor to hijack)

[Irieguy]

[ QUOTE ]


If I measure the batting averages for Barry Bonds over 500 at bats, Andruw Jones over 750 at bats and an above average minor league player over 1000 at bats, then average these for a batting average, I learn nothing. Especially if no one else observes these at-bats and I ask the baseball players to report the data to me themselves.


[/ QUOTE ]

I would suggest that the difference between the best 2+2er that has won $ over their last 1000 SNGs and the worst 2+2er that has won $ over thier last 1000 SNGs is smaller than the difference between Barry Bonds and a minor leaguer.

The self-reporting problem and all the selection bias problems are huge, but I have an idea about how to mitigate those concerns to the extent that I will be able to understand the data. Also, I personally know the majority of the players who will be reporting their data. So, while I will be asking Barry Bonds to tell me his batting average... he will be likely to tell me the truth since I am his batting coach. Andruw Jones can lie if he wants to, but I just paid for his lap dances last week. The minor leaguers are harder to trust, of course, but I don't think many of them are slipping into this data pool.

The reason why I believe that combining this data is useful is because of the nature of SNGs. Variance is fixed by the structure, and performance boundaries per SNG are narrow because there are only a few hands being played.

Irieguy

[1C5]

Well some have corked the bats *cough Sosa cough* and many have juiced *cough Bonds and many others cough* so we are already 1/2 there.

[bones]

[ QUOTE ]
Andruw Jones can lie if he wants to, but I just paid for his lap dances last week

[/ QUOTE ]

I'm going to go out on a limb and say that this is the only time Yugo and Andruw Jones will ever be analogous.

[mmbt0ne]

[ QUOTE ]
When I am finished, you will know how much money a group of previously successful SNG players earned over 30,000+ SNGs in a short period of time.

[/ QUOTE ]

Will we know how successful they were earlier?

[raptor517]

i will play 5 109s to contribute. i will start with a win and end with a win, using my selective sampling techniques. im sure you will be able to grab oodles of information from that. holla

[The Yugoslavian]

[ QUOTE ]
[ QUOTE ]
Andruw Jones can lie if he wants to, but I just paid for his lap dances last week

[/ QUOTE ]

I'm going to go out on a limb and say that this is the only time Yugo and Andruw Jones will ever be analogous.

[/ QUOTE ]

Well except I'm not in this sampling thing. For several reasons.

*shrug*

Anyway, I don't think Andruw Jones was up in a scrippa club last night with a Peruvian cutie. So, you're right, this *is* his only chance to be analogous to ME.

Ship it batch!

Yugoslav

[bones]

[ QUOTE ]
Anyway, I don't think Andruw Jones was up in a scrippa club last night with a Peruvian cutie. So, you're right, this *is* his only chance to be analogous to ME.


[/ QUOTE ]

Andruw Jones doesn't stop at first and head back to the dugout.

[pergesu]

[ QUOTE ]
[ QUOTE ]
Anyway, I don't think Andruw Jones was up in a scrippa club last night with a Peruvian cutie. So, you're right, this *is* his only chance to be analogous to ME.


[/ QUOTE ]

Andruw Jones doesn't stop at first and head back to the dugout.

[/ QUOTE ]
PWNT BATCH!!!!11one

[ace_in_the_hole]

Just a curious observation that sorta relates to this data collection:

I think that this time of the year is going to be slightly more profitable than the rest as the holidays are giving people more down time and the cold weather in some regions will induce more indoor hours and in turn more internet poker for many people. Just curious if this is a concern, furthermore there could be an additional data collection after the new year with the same players that could also help determine if that time epriod was more profitable. I do understand that it is hard to derrive any concrete evidence due tot he variance but it may be sometihng to think about.

[The Yugoslavian]

[ QUOTE ]
[ QUOTE ]
Anyway, I don't think Andruw Jones was up in a scrippa club last night with a Peruvian cutie. So, you're right, this *is* his only chance to be analogous to ME.


[/ QUOTE ]

Andruw Jones doesn't stop at first and head back to the dugout.

[/ QUOTE ]

At least he's in the major leagues while you're ass in still back in AAA, /images/graemlins/blush.gif.

Yugoslav

[wuwei]

[ QUOTE ]
[ QUOTE ]
Anyway, I don't think Andruw Jones was up in a scrippa club last night with a Peruvian cutie. So, you're right, this *is* his only chance to be analogous to ME.


[/ QUOTE ]

Andruw Jones doesn't stop at first and head back to the dugout.

[/ QUOTE ]

I was going to ask why the trip report isn't up yet, but maybe this explains it /images/graemlins/blush.gif

[Nick-Zack]

Your PM box is full so I can't reply

[KennyBanya]

[ QUOTE ]
Your PM box is full so I can't reply

[/ QUOTE ]

Me too.

Irieguy, feel free to PM if you still need data. I currently play the 22s, but I was unable to PM you.

Thanks,

KennyBanya