FYP

GG - everyone can end a session at +200... it is just that the maniac over many sessions will have a wider range of results than the low variance guy.

Although CAD is a whole new wrinkle... math may be not applicable

We don't have "sevens" up here cuz they look too much like 1's.

Gisthataproblem?G

This is exactly the same thing I was saying earlier in the thread, about how standard deviation per hour is computed. bip! was saying that no one actually tracks their win/loss hour-by-hour or orbit-by-orbit; instead they track whole sessions and divide the session result by the number of hours. To me this seems like it gives inaccurate results for the standard deviation; but the only way to resolve this issue is to take careful notes of how much you win or lose every orbit or every hour.

I think this month I may start doing that, because it seems like a very fun experiment both for poker and for seeing whether the CLT can be applied to poker orbits. If I stick with it long enough to get a big sample size of orbits I may come back in awhile and post my results on here. bip, what do you think would be a big enough sample of orbits to be worth analyzing?

^^^^

Ah, that makes a lot more sense to me.

Tracking results hour-by-hour or orbit-by-orbit (hey, why not hand by hand?) seems like a neat idea, however I'm sure you'll quickly find it a big pain in the ass. Good luck to you if you do it!

GcluelessstatsnoobG

Here's a thought ... if we took an online dataset we'd have the results of every hand. It'd be trivial to calculate these kinds of statistics for both the high resolution "every hand", as well as by orbit or by hour (call it every 25-30 hands?) or another factor. If the various methods all come up the same or close, you wouldn't need to worry about the coarse resolution we get live.

Anyone have one of those DB's handy?

Because hands that you play UTG aren't supposed to have the same average win as hands on the button. So orbits are the smallest units that make sense to track.

Well I deal with a lot of big data-sets for a living and use Matlab every day, so it really wasn't a lot of effort once I sat down and decided to do it (thanks to a super bowl blizzard). I'd actually be fairly interested in looking at a couple other datasets too.

If you look at the rolling standard deviation plot, between about 2000 and 2500 hours I was putting in more time in the downtown rooms than I was in the local games. So a less wild game. The StDev is lower during that time period, and has really cranked up a bit more as I've gotten lazier about driving for a game.

This would be interesting.

I can't tell you a theory reason, but looking at my trailing average and StDev plots (and playing with other windows I didn't show), I'd expect you need between 100 and 200 samples to get a meaningful number. Smaller samples will be subject to a lot of noise.

I don't know @ how many orbits necessary, but I would guess something on the order of ~500

*to get stdev... WR will still be way uncertain (as discussed before)

500 orbits = ~167 hours = enough of a sample size?

Gseemssmalltome,butIsuckatstatsG

The finer sampling resolution doesn't really accelerate any of the stats confidence... just a curiosity exercise

Since we're again dealing with what is effectively a scaling issue, it would be the same thing if you looked at, e.g., your stack size once an hour as it would be your stack size once an orbit, since you're almost certain to get +/- 2 or 3 hands per hour every hour. At worst, we could save a lot of effort by doing for instance 50 hours vs the number of orbits seen in those 50 hours.

Bleh, all I'm saying is that a lot of our x variables will scale linearly. Hands/hour, orbits/hour, hands/orbit, etc. Use the unit that is linear and easiest to measure.

oh god

you're doomed

sorry to be a debbie downer but beating 1/2 and making a living off 1/2 are two entirely separate things and just the fact that you have a plan for how you're gonna play every session tells me you are not the 0.1% that can pull the latter off

also, even if the miracle happens, your best case scenario is making barely enough to survive while dealing with the emotional torment of a job that sometimes costs money... is that really what you want in life?

Or he could get a job and end up like this.

work is dumb but at least it's safe!

would much rather worry about how i'm gonna deal with the annoying intern than worry about how i'm gonna be homeless if this downswing keeps up

often forgotten: merely thinking about the latter slashes your winrate, and it's hard not to think like that when you're losing!

Being homeless is underrated.

everything DK says is right. keep your jobs. no money in poker. everyone's solid.

Might remote into work today, weather looks awful

Kinda gotta side with DK on this one.

1. We don't know the dudes full situation, but your margins are incredibly thin at 1/2 for a living wage.
2. 6k is totally fine for a Rec player with nothing coming out for living expenses. It's suicide for a pro.
3. You have a boxed recipe for yor play which is beyond bad.
4. I played poker for one year in school where it was being used for rent. I can't say I affected me a ton, but it is different and mentally Id say I'm a pretty unusual exception on that front.
Get a job in Vegas an grind in the side until you know what you're doing. Solid life advice.

Working into a job serving food or bar tending is definitely optimal imo.

What you can do is combine sessions so that each "new" session is about the same number of hours.

I.E., these are my first 11 session lengths.

5.00
4.00
7.50
9.50
9.00
6.50
13.00
11.50
9.00
8.00
11.75

Combine the first 3, combine the 4th n 5th, 6th n 7th, 8th n 9th, 10th n 11th, and then you have:

16.5
18.5
19.5
20.5
19.75

It's not perfect but closer to what you're trying to do.

But isn't what Vernon trying to do the opposite? (i.e. he'd like to break down those combined hours into their original hours, the only problem being that he doesn't currently have the results that go with the original hours)

GIwastoldtherewasgoingtobehalfnakedgirlshere,no?G