Glad to see you are playing 2/5 now. That's a very nice win rate for Vegas 2/5. Good job!

Sent from my SM-T700 using Tapatalk

it's interesting that the variability of your true winrate doesn't add more noise to your expected earnings over the next 1000hrs.

A .95 confidence interval for your mean winrate is: $23.5 +/- $6.25.

Assuming $172 / hr standard dev going forward a .95 prediction interval (taking into account the variability in your winrate prediction) over the next 1000 hands is:
$23.5 +/- $12.56

Assuming $172 / hr standard dev going forward but ignoring the variability in your winrate prediction (i.e., we assume the mean is $23.5 with certainty) a prediction interval over next 1000 hands is:
$23.5 +/- $10.88

So if knowing your winrate is 23.5 with 100% certainty would mean you could expect $13/hr to $34/hr instead of $11/hr to $36/hr. so... meh.

I would have intuitively thought that the uncertainty of our winrate would have played a bigger factor in the uncertainty of our results going forward...

yep so if you're winrate is estimated over N hours and your predicting an interval over K hours then the adjustment to incorporate the variability of your winrate is:

sqrt(1 + K/N)

so in this case you have sqrt(4/3) = 1.1547

so the standard deviation around your true winrate is 172 / sqrt(1000) = 5.44 and your standard deviation around your observed winrate is 5.44 * 1.1547 = 6.28

Not sure I'm in agreement with this; my guess is when the end comes it might surprise people how fast it came.

All I can do is base my predictions based on what is trending in my market, which is:

- 2 of the local rooms flat out closing recently, with the rumour being that another in a high profile location will also eventually close (although admittedly I'm not sure if that rumour is true since it seems so outlandish)
- at least 2 local rooms recently severely downsizing number of tables (might be the case at other rooms too but I don't know for sure)
- Limit game (the entry level game for Texas Hold'em as far as I can remember) doesn't run anymore in my room (guessing similar in other local rooms)
- due to lack of tables, nightly tournament in my room has long been cancelled (I'm assuming similar in other downsized rooms)
- my room has once again increased its rake

Nothing trending in a good direction here.

Gwouldn'tsurprisemeatallifpokerwascompletelydeadhe rewithin5years,butI'vealwaysbeenaskyisfallingtypeo fguyG

With regards to predicting ones WR moving forward, the state you expect the game to be in >>>>>>>>>>>>>>>>>> all the SD math being shown.

GimoG

As someone that has been regging these games for the last few years- this is definitely true.

Hard Rock usually gets 1 5/10 game at night during weekdays. Isle will have 1 at night and 1 during the day. PBKC rarely gets a game. Most weekday nights- youll have two 5-10 games to choose from. Neither is very good. Lots of good local pros and traveling pros from Germany, Czech, etc.

Game composition is generally all pro and 1 maybe 2 spots.

2-5 is still great here but to pretend like 5-10 in Florida is amazing or something just isnt the case.

It is no coincidence that far and away my worst year on my giraffe was 2015 ($7.90/hr = 2.63 bb/hr over 582 hours), where my room was often down to just a single table the vast majority of those hours.

G(1)tableselection(2)seatselection(3)nothingelsemu chmattersG

yeah this is very true. all this inference power disappears if the distribution has changed enough over the period you sampled and the period you're predicting. i think the best you can really do is look back over a period you feel is consistent with the game state you're in now, then predict your hourly over the next few months. but that's probably not very useful.

Heh, I know right? I sincerely agree. Funny to think I'm about to hit 2k recorded hours and my first 1K hours, where I knew absolutely nothing, my WR was almost twice that of my current WR in my second thousand hour group.

My go-to strategy has always been 1 best game and 2 best position. I might have to finally learn all this math you kids keep talking about. What's it called again, "pot odds"?

I wish I know what game selection means.

Been pretty much playing single table games almost 90% of the times if not more.

Forces you to be a better player though.

Where are you at?

Think growth or decline of your game is tightly coupled with the economic climate in the 50-100 mile radius of your rooms. Need to be close to large population with solid disposable income.

I'm in the burbs just outside of DC. All the tech workers (like me), lawyers, lobbyists etc. Contribute to a very healthy poker economy.



Sent from my Nexus 6P using Tapatalk

Absolutely. Near Detroit here. During the auto-bailouts there was a huge influx of money to the tables as people took the severance offers, then it dried up quickly. Everything else has been pretty "meh" since then.

But you can see a visible uptick in the numbers of players at the table at the start of the month when the checks come it, then it tails off near the end. Like clockwork.

I live there as well these days. All the poker action up at MDL? (Have yet to play it)


Sent from my iPhone using Tapatalk

There was a known 1-3 "whale" back when I played in st Louis. Owner an apt building. We all knew when rent day was. He'd walk in and attempt to dust off 1-3k into a 300-500buyin game. Also made 2+2 boards for being so incredibly foul at the table that he'd drive other fish away, leaving you with ethical Qs (at least in his specific case, you could be a dick to him and stand up for everyone else and he'd keep giving out $$$)
.

A whale or two can make or break a poker economy.

Springfield IL had a home game I knew about but never played (2.5 hour drive further than a Dino). Lots of rich guys. Some bad lags prob made 80k a year min there playing 1-2x per week. Their whale who sometimes went elsewhere probably dropped 200+/year

Sent from my iPhone using Tapatalk

IMHO, I think that a lot of people in this thread put way too much emphasis of the math of standard deviation. Maybe I am biased though.

Any guesses s as to whether a 2/5 game 10% capped at $15 with $5 sit down fee is beatable ?

What bias do you have against standard deviation.

I don't think anyone would say treating hourly results as a stationary process is a perfect one but simple models can be useful as long as the user is aware of the limitations.

I think a lot of people who haven't spent any time studying predictive statistics think that a model that isn't perfect is useless. But this is not the case at all.

Agreed.
I'm also not sure that poker volatility follows a normal curve. If it doesn't, SDev is guideline that gets you blown up.
Has anyone done work on whether poker w/l tend to be Gaussian or non-Gaussian? If so, what is the proper curve like?

Like AI EV is also a crappy metric--by the time you're AI, a number of things, mathematically, have gone right.
Are there any things that attempt to measure $$ put in ahead paired w wins without showdown?? ((So if you bet 20bb Otr w a flush and are called by a pair, you're+ 20bb. Betting the FD on the turn for 10bb would be -8bb as his pair would be 80% to win).


Sent from my iPhone using Tapatalk

Central Limit Theorem in its weak form states that the average of repeated trials that satisfy certain mixing conditions (i.e., events separated by large time periods are independent) converges (in distribution and probability) to a normal random variable.

So yeah hourly results will converge to a normal distribution but that still doesn't say much about whether that normal distribution mean and SD are useful in predicting future mean and SD.

I just thought ATsai was just using a well timed statistics pun.

https://en.m.wikipedia.org/wiki/Bias_of_an_estimator

That or he was astonished at all of the energy being spent on these calculations instead of more important things like basic pot odds, bet sizing, and seat selection. Most complex math I focus on is fold equity calculations off the felt, and those are trivial (and much more important/insightful than my standard deviation)

Come on homie. You've studied some finance--Princeton is like economics central. Central Limit Theorem = lose all your money in the stock market.

And if it is a Gaussian distribution, is there an empirical study across players for this?

Given that trials look different, and stack depth paired w a variable amount to win/lose manifests itself in skewed statistics makes me inclined to think Normality does not hold for live poker (NL, at least, Normality is prob true in Limit)


Sent from my iPhone using Tapatalk

I struggle to count the pot accurately so I'm not holding out a lot of hope that this will make sense but:

Wouldn't a normal distribution only apply to the actually purely random events in the game?

Your overall experience of variance/swingyness may well not be normally distributed due to all sorts of non-random factors:

- your tilt
- their tilt
- being exploited
- exploiting others
- differing game dynamics over time
- changing relative skill level over time

I'm assuming there are other lower stake games running (i.e. your game isn't literally the lone table running in the room)? If not, the end is nigh.

Grepent,imoG

At least a couple of us have recently guessed that table selection is far and away the number one factor in winrate (this being a winrate thread).

Would be super interesting if RP had lone table running stats vs multiple table running stats (if there was a boom and post boom in his area).

GwishRPwouldpostgiraffesforbargsandlolzsG

These are the best things to work on wrt to WR imo