If you equate poker to a financial investment you will realize that this is factually untrue. STD Dev of differing investments does not mean one investment is "better" or "worse" by itself.

I will grant that if there are 2 investments with equal expected returns (win rate if you will) we would prefer a reduced std dev if possible.

But just saying reducing std dev is the most important thing doesn't make any sense. I can invest in 30 day Treasury certificates that have virtually no STD Dev but those aren't maximizing returns (win rate)

Welp, now that I've written this little calculator......

Scenario, you're a winning 1-3 rec with a steady $21/hr WR and 55BB/hr StDv, you play 600 hours a year. 5% of your year the game goes off the hook with a straddle, deep, and heavy action. You like to book solid winning years, so you question whether the extra EV is worth the risk. Do you play those 30 hours in a wild game or table change to protect your consistent stream of fun money? If we assume in the wild game you have a $48/hr WR and 80Straddles/hr StDv, how does this affect your results?

Regular Game for 600 hours
EV= $12,600
60%confidence interval = $9,200 - $16,000
90%confidence interval = $6,000 - $19,200
99%confidence interval = $2,200 - $23,000

YOLO for 30 hours + Regular Game for 570 hours
EV = $13,410
60%confidence interval = $9,400 - $17,400
90%confidence interval = $5,600 - $21,200
99%confidence interval = $1,200 - $25,600

So does that mean in the scarier game you're 99% sure that for the year we'll win somewhere between 0.7 bb/hr and 14.2 bb/hr?

GsothistellsmeIshould...um,whatdoesthistellmeIshou lddo?G

It could be because we're both trying to have conversations with other people involved who may not be arguing the same things, but I think you are not getting it. I hardly care about SD when I am thinking about how I play the game. I care about SD when I am deciding whether or not to play the game at all.

Link to post?

This may be true but is not necessarily true. There are cases where aggression (even "good" aggression) may either lower variance or raise it. For example, frequently ending hands on earlier streets may decrease variance, while things like neutral EV pre-flop opens, all other strategy the same, will increase variance.

All other strategy the same, I'm pretty sure VPIPing more necessarily increases variance. The LAG would have to be playing differently on the hands both he and the TAG are VPIPing, which is the case for players that are conscious of their overall strategy; however, by my understanding of the terms' meanings there is nothing necessarily different about a LAG's vs. TAG's play style beyond pre-flop.

This is because folding has an EV of 0 on that street which is always equal to the result of 0, so the variance is also 0.

correct, play it safe all year, 99% sure you'll win between 1.2 bb/hr and 12.8 bb/r.
play it safe 570 hours, YOLO 30 hours, 99% sure you'll win between 0.7 bb/hr and 14.2 bb/hr

(under the given assumptions, obv)

you can lead a GG to water....

Ok, fair enough; the comparisons show not much difference in the min/max between the two.

However, the real take away is the min/max overall is insane (the whole point of the recent variance talk). Given these mins/maxs (the min where you are just barely breakeven, the max where you are an absolute stone cold crusher), it's pretty clear everyone should either (a) add two more part time jobs on top of current job and never play poker again vs (b) quit their current job and play poker full time and (c) something in between. (i.e. pray you run well, it's likely the only thing that matters)

Gnothatingatall,justsayin'G

Browni, the mpethy posts were over time, and scattered about. Someone better at search than me would have to link them.

Lags and tags are playing their hands differently post flop. Notably bc lags have a wider range. They check more. They bet smaller. Etc.

You are on the right track though in realizing it has to do with non-showdown winnings. That is why a good lag can have lower variance than a good tag.

To be clear, I am talking about variance from expected win rate. That is not the same as variance from expected value. This has always confused me when talking about variance wrt poker.

Here's a rio article. Focus is in mtts but Phil's op and Ben Tollerene's first paragrah in his response are what mpethy was saying years ago. Still heavily debated as you can see.

http://www.runitonce.com/mtt/theory-variance-vs-style/

The water comment was intended as a light needle, I wasn't trying to be overly negative.

I don't draw the same conclusion that you do though.

Same scenario, but now we're talking about a guy whose a full-time player for 8 years :

hourly WR in ()

Regular Game for 16000 hours
EV= $336,000
60%confidence interval = $318,400 - $353,600 ($19.90 - $22.10)
90%confidence interval = $301,600 - $370,400 ($18.85 - $23.15)
99%confidence interval = $282,400 - $389,600 ($17.65 - $24.35)

YOLO for 800 hours + Regular Game for 15200 hours
EV = $357,600
60%confidence interval = $337,000 - $378,200 ($21.06 - $23.64)
90%confidence interval = $317,400 - $397,800 ($19.84 - $24.86)
99%confidence interval = $294,800 - $420,600 ($18.41 - $26.29)

I was thinking up a reply, but I realized I made an assumption on what you meant here. Can you explain for me to be sure?

heh, you’re going to catch me. I am not good at this and am paraphrasing from memory how I understand it (eek)

I think I should have said there are 2 expected values. There is an expected value hand vs hand and there is an expected win rate. Stdev bb/100 refers to win rate. Expected value hand for hand influences variance in expected win rate but it’s not the entire picture bc non showdown plays a part in wr. I think we misuse the term variance when we refer to AA losing to JJ but we use it correctly when we refer to winning/losing 100bbs in 100 hands.

I’m very close to approaching no f**king clue of what I’m talking about though. I used to know a lot more about this when it was fresh in my mind from college and when bright minds posted here often and I was truly interested in solving poker. (Most of that has died for me)

I’ve since learned to work on mental game, study common high level spots, and just shoot the ball

So this is not a strat thread, and I therefore don't want to go too far down this rabbit whole, but this understanding is incorrect. LAGs depend more on non-showdown winnings and when played back at often don't have a hand worth showing down.

The reason that this matters to variance is that LAGs are all-in with cards to come less often, and therefore are less frequently at the mercy of the deck.

Garick I meant to +1 your Johnny quotes earlier. Really incredible finds that I’m sure took some work. How many posters have we seen that same flow with? I’d guess hundreds. Except most totally fall off the radar at the downswing and never come back.

Only true sadists like johnny, squid, etc. keep on going. Not a compliment guys, I don’t know how you do it.

huh, had the popcorn all popped waiting for your explanation on that one. just shoot the ball is a nice save tho.

First - lol English. Variance has a ton of meanings, with at least a couple of meanings in the legal realm.

However, itt and in the forum in general, the word tends to be bastardized into a proxy for "run-bad" (and much less often - "run-good"). If we think our results suck over the last X hours compared to what we believe is our true edge, we say here the variance has been tough, or curse our variance, or poker variance is a killer, etc. This usage is related to the formal statistical definition, but people often use it itt without understanding that definition, even informally.

That definition, from the wiki page: In probability theory and statistics, variance is the expectation of the squared deviation of a random variable from its mean. Informally, it measures how far a set of (random) numbers are spread out from their average value.

For example, if a poker player had the following hourly results over 3 hours of play:
+400, -200, -800

The mean of those results is -200. The variance of these results is the average squared distance from the mean, or: (600^2 + 0^2 + 600^2)/3 = 240,000

Standard Deviation is the square root of variance, or in this case: ~489.90

research the Central Limit Theorem if you want to read up on how we can use observed mean and StnDev to make confidence interval estimates. Understand the CLT and a few distribution graphs and you are most of the way to earning 3 statistics credits (in the 80s, anyway).

(I'm ignoring the term "sample variance" in this poast as it is beyond scope and my understanding is rusty anyway.)

This.

I've said it before and I'll say it again, your games sound truly horrible. :/

Can anyone confirm sailboat's stdev bb/hr figures are not completely arbitrary? When I ran sims months ago I was supplied by this thread completely different numbers.

One thing I've realized extremely quickly is that table selection is just as important a skill as actually playing poker and it reflects in the winrate.

A lot of things go into it, scouting out the stacks, listening to the excitement, average age of players, attire, drinks, noticing who the regs are and how many of them are at the table, etc. Also noticing when the table changes from good to bad and vice versa, and noticing whether your table image is ruined because of run-bad or run-good. Probably forgetting to mention at least another dozen factors.

I would also add that this aspect is doubly as important at low stakes.

I have sat at some truly horrible tables, but probably still not on gg's level.
PS. sorry for the derail.

It's so damn important. If you have a choice between 2 tables (and I often will get a choice between maximum 2 tables) and one is 10bb/hr while the other is 5bb/hr EV just based on the villains/lineup - making a very careful decision here is important.

Something else I do a lot of during a session is move around the table. My room turns over rec players like crazy. If there's two empty seats to my left, I'll often move two to my left so that when a new player sits (good chance they're a fish) they'll have to sit to my right. Small stuff like that. As well as being constantly vigilant as to ensuring you're in the best game in the room at all times or at least doing something about getting moved.

here is a guide to calculating your own StnDev / hr

http://web.archive.org/web/200502110.../mmessay8.html

if you use this method on your net results, then divide your stnDev/hr by BB amount to get stdev bb/hr

Agreed 100%, a large portion of my success early on at 1/2 was great table selection, especially on weekends when the games can go from completely average to completely insane.

Too difficult for me lol.

with out derailing this thread even more I can honestly say that far more important than table selection and seat jumping (sooo many horribad regs literally play musical chairs all night long) is being a sport.

Fact - over the last couple o years I have literally table changed a handful of times. And I never seat jump. I get seat jumped regularly

What I am tho is a nice guy. I by no stretch of the imagination go full DGAF, but I am pleasant, I never seat jump fish, and I dont bum hunt, and never berate/talk strat. Because I am a nice guy and stick by these rules I play with a lot of action players that dont like playing with the dbags

Just sayin

+1 million

If you post up (or pm) results from 30+ sessions I can do it for you. Need a list of hours and results with a heading on what bb it is for those results.

Yeah, I might make 1 move to get a nit on my left and some decent stacks on my right, but mostly it's to get a good TV spot if it's game time. But, mostly shouldn't sweat it when you have a big edge on the pool.

That being said, time of day matters. Saturday at midnight in a large pool, if they don't hate money at your table, probably should try a table change.