vpFREE2 Forums

For a Visitor - Does High Variance really matter?

#1

On my last trip to LV two weeks ago, I played 15/8 LD for the first
time and hit 4 deuces; which, of course, makes me like the game even more.

I knew that the game had a high variance, but I didn't know how high
until I came home. So now, I wonder...

... for a visitor that visits 2-4 times a year (I hope), does playing
a high variance game matter if the EV is also high? Since I played
this at the LVH, I also know that there are other possibilities there,
like JOB, PE & AA. In fact, AA has a higher EV and lower Variance
than LD. However, there are two factors that would make me want to
play LD more...
    The MG machines that have LD are TITO and come in .25 & $1 denoms;
while the machines that have AA are coin drop and are only in .25 denoms.

Just curious what the experts think...

Jeff

#2

use the sharpe ratio:

sharpe ratio = (er-1)/sqrt(variance)
for fpdw = .0076/sqrt(26)= .0015

the higher the sharpe ratio the better the play, assuming you have the
bankroll

kelly bankroll (~11%ror)= variance/(er-1) bets

On my last trip to LV two weeks ago, I played 15/8 LD for the first
time and hit 4 deuces; which, of course, makes me like the game even

more.

I knew that the game had a high variance, but I didn't know how high
until I came home. So now, I wonder...

... for a visitor that visits 2-4 times a year (I hope), does

playing

a high variance game matter if the EV is also high? Since I played
this at the LVH, I also know that there are other possibilities

there,

like JOB, PE & AA. In fact, AA has a higher EV and lower Variance
than LD. However, there are two factors that would make me want to
play LD more...
    The MG machines that have LD are TITO and come in .25 & $1

denoms;

while the machines that have AA are coin drop and are only in .25

denoms.

···

--- In vpFREE@yahoogroups.com, "JShort_DE" <jmdshort@c...> wrote:

Just curious what the experts think...

Jeff

#3

Everyone has to determine their own comfort level,
financially and psychologically.

We're single line dollar and multi-line quarters players
and we play whatever and wherever we get the highest
positive EV/hour (including casino incentives).

We've never run into a situation where the variance was
high enough to have it affect our decision making process.
However, there probably is some point that we would reject
a play for extremely high variance ... a 101% ER on a game
that only paid for a royal flush, for example.

vpFREE Administrator

···

On 31 Dec 2004 at 22:09, JShort_DE wrote:

... for a visitor that visits 2-4 times a year (I hope), does
playing a high variance game matter if the EV is also high?

#4

We've never run into a situation where the variance was
high enough to have it affect our decision making process.

loose deuces?
15/10 dollar loose deuces has a 10%ror bankroll of $40,000!

However, there probably is some point that we would reject
a play for extremely high variance ... a 101% ER on a game
that only paid for a royal flush, for example.

rf/23000=1.01, so rf=23230, var~=(23230^2)/23000=23462, kelly bankroll
= 23462/.01= 2,346,230 bets!

···

--- In vpFREE@yahoogroups.com, "vpFREE" <vpFREE@C...> wrote:

#5

Hi Jeff;
The short answer is Variance always matters, as it affects the
"shape of the game". The more detailed and relevant questions are,
(I think), for the occasional visitor you describe are these:

1. Is higher variance good or bad?
2. How MUCH does variance matter?
3. Are long term bankroll caclulations revelevant? I know you didn't
ask this question, but usually when people want to know about
varinace it has something to do with bankroll.

More short Answers.
1. It depends
2. It depends.
3. No. Well maybe.<g>

1. High variance (and obviously the term high is relative) can be a
big plus, for instance, if one wishes to have a good chance of a
profitable hot streak of jackpot hands leaving with a nice chunk of
the casino's money. The other half of that is that it also gives you
a bigger chance of losing quickly, limiting the amount of play you
can get in by draining your trip stake in a short time. Low variance
games will help people to maximize their playing time, coin-in and
comp status. And staying in the game longer also means you increase
your chances of a royal. There are endless variations of this. The
point is it depends on your goals. Adn your attitude.

2. If the visitor wishes to increase his play over time and be a
long term winner, variance becomes much less important than EV. If
one wanted to simplify it, one could say variance (both the good and
bad part) is more important in the short term and achieving the max
EV is more important for the long term. But that's too simple.
Variance has a direct affect on long-term bankroll.

3. Long term bankroll calculations are just that - they are designed
to calculate what you need to keep on playing indefinitely. The
occasional visitor just needs to know... well again that depends on
the visitor:
- One might want to know how much to bring to be pretty sure you
can keep playing for the whole trip, even if you are having a bad
streak, given the fact that you plan to play dollar Double Bonus.
- Or one might have a given stake and want to know which game to
play to make it most likely to last the longest. The reason I said
"well, maybe" is that looking at the relative long term bankroll
requirements might give you a clue about which game to choose in
this scenario. Or it might not. <g>

As in the variance examples, the trip stake variations are almost
endless. One of my orginal partners in VPP, Greg Brooks, liked to
come to town with a couple thousand bucks and play $5 Bonus Poker
until he ran out of dough or had some major win(s). Usually this
will result in a very short amount of playing time <g> but not
everyone is after extended sessions. If you lose your gambling
stake hopefully you bring some non-gambling dough and there are
plenty of other things to do in LV (I'm not sure what they are but
that's what I hear).

BTW, here's the most important thing about a trip stake - no matter
how good a game you play, or how good a player you are, don't make
any plans for that money being with you when you get home.<g>

Hope this helps a little. Frankly, I'm more confused, than when I
started to write it.<g>
Skip

JShort_DE wrote:

···

On my last trip to LV two weeks ago, I played 15/8 LD for the
first
time and hit 4 deuces; which, of course, makes me like the game
even more.

I knew that the game had a high variance, but I didn't know how
high
until I came home. So now, I wonder...

... for a visitor that visits 2-4 times a year (I hope), does
playing
a high variance game matter if the EV is also high? Since I
played
this at the LVH, I also know that there are other possibilities
there,
like JOB, PE & AA. In fact, AA has a higher EV and lower Variance
than LD. However, there are two factors that would make me want
to
play LD more...
    The MG machines that have LD are TITO and come in .25 & $1
denoms;
while the machines that have AA are coin drop and are only in .25
denoms.

Just curious what the experts think...

Jeff

vpFREE Links: http://members.cox.net/vpfree/Links.htm

                   Yahoo! Groups Sponsor
                       ADVERTISEMENT
                        [click here]

  ---------------------------------------------------------------
Yahoo! Groups Links

   * To visit your group on the web, go to:
     http://groups.yahoo.com/group/vpFREE/

   * To unsubscribe from this group, send an email to:
     vpFREE-unsubscribe@yahoogroups.com

   * Your use of Yahoo! Groups is subject to the Yahoo! Terms of
     Service.

#6

A couple of modest comments:

-- Experience will tell you that you want a larger trip stake when
player a higher variance game than lesser. In my experience, if I'm
playing for an extended day, I want close to twice the bankroll
playing 10/7 DB than when playing 9/6 Jacks. Over time, things smooth
out, so for 3 days I want 50% more for 10/7 DB.

-- Variance is a much greater concern over time to the player who's
really stretching their bankroll (say, allowing a bankroll of 3-5
royals) than one who has allowed a very generous bankroll (say 10
royals). The larger the bankroll, the lower the risk of ruin AND the
longer allowance for experienced return to approach expected return.

The most secure players have a very secure bankroll and play with
strong confidence that they'll ultimately realize the expected rewards
of the games they play.

···

------

How does this translate for the very casual player? I think they're
very close to as important as to a very active player.

No question the single trip stake considerations are identical. As
far as risk of ruin over time, factor in the fact that when playing a
positive game it's expected that your bankroll will grow with time.
Conseqauently, the greatest risk of ruin occurs at the outset of play,
whether ultimately playing extensively or on a limited basis.

So what I'm saying is that a Visitor should be very much concerned
with Variance.

- Harry

#7

i'm gonna disagree with your statement "variance becomes much less
important than EV"

let's say you had a choice between two games of equal EV but one of
them had a sequential royal and thus much higher variance, which game
would you choose and why?

or you had a choice between two progressives of equal EV, but one of
them had an additional 1% tied up in the progressive?

or you had a choice between a flat top game, say full pay deuces wild,
and an 8/5 job progressive that was at +1%?

it's a common misconception to think that in the "long run" variance
doesn't matter, but variance is always there

it takes 100N0 hands to get to the point where one standard deviation
(the square root of variance) is within +/-10% of EV, for fpdw
N0=var/(er-1)^2=450,000 hands, 100N0=45 million hands

···

--- In vpFREE@yahoogroups.com, Skip Hughes <skiphughes@e...> wrote:

2. If the visitor wishes to increase his play over time and be a
long term winner, variance becomes much less important than EV. If
one wanted to simplify it, one could say variance (both the good and
bad part) is more important in the short term and achieving the max
EV is more important for the long term. But that's too simple.
Variance has a direct affect on long-term bankroll.

#8

--- In vpFREE@yahoogroups.com, "Harry Porter" <harry.porter@v...>
wrote:

So what I'm saying is that a Visitor should be very much concerned
with Variance.

maybe a simplified way to think of this is that variance is more
important in the short term and ev is more important in the long term
and the crossover (where variance in a sense equals ev) is at
N0=variance/(er-1)^2 hands or 450,000 hands for fpdw, so for total
hands less than N0, variance dominates, whereas for total hands
greater than N0, ev dominates (keeping in mind that N0 itself is a
function of variance and ev and thus varies wildly by game)

#9

--- In vpFREE@yahoogroups.com, "nightoftheiguana2000"
<nightoftheiguana2000@y...> wrote:

it takes 100N0 hands to get to the point where one standard

deviation

(the square root of variance) is within +/-10% of EV, for fpdw
N0=var/(er-1)^2=450,000 hands, 100N0=45 million hands

Iggy, I do not understand. I need 45 million hands to what?
Again, you told me once, what does NO stand for?

DWK

#10

here's one explanation of N0:
http://www.bjmath.com/bjmath/refer/N0.htm

i can make another attempt:
there is a mean, an average return, and there is a distribution of
results about that mean value, one standard deviation below and above
the mean represents 68% of the total possible results, two standard
deviations represents 95% of the total possible results, three
standard deviations represents 99.7% of the total possible results

N0 is the number of hands at which the mean catches up to (equals) one
standard deviation and the result is that 84% of the possible results
are now positive for a positive expectation gamble or negative for a
negative expectation gamble

···

at 100N0 the standard deviation has been reduced to 10% of the mean, meaning 68% of the possible results equal the mean +/-10% --- In vpFREE@yahoogroups.com, "deuceswild1000" <deuceswild1000@y...> wrote:

--- In vpFREE@yahoogroups.com, "nightoftheiguana2000"
<nightoftheiguana2000@y...> wrote:
>
> it takes 100N0 hands to get to the point where one standard
deviation
> (the square root of variance) is within +/-10% of EV, for fpdw
> N0=var/(er-1)^2=450,000 hands, 100N0=45 million hands

Iggy, I do not understand. I need 45 million hands to what?
Again, you told me once, what does NO stand for?

DWK

#11

NOG;
Your examples assume an equal EV, so none of them are at actually
variance (sorry, couldn't resist) with my statement. That is, EV is
not a factor at all in those examples, so obviously variance is of
the greater importance (something > nothing).
Can you how me some examples where a lower EV is preferable to a
higher EV game (in the long run) because of a lower/higher variance?
Skip

nightoftheiguana2000 wrote:

···

i'm gonna disagree with your statement "variance becomes much less
important than EV"

let's say you had a choice between two games of equal EV but one
of
them had a sequential royal and thus much higher variance, which
game
would you choose and why?

or you had a choice between two progressives of equal EV, but one
of
them had an additional 1% tied up in the progressive?

or you had a choice between a flat top game, say full pay deuces
wild,
and an 8/5 job progressive that was at +1%?

it's a common misconception to think that in the "long run"
variance
doesn't matter, but variance is always there

#12

Skip Hughes wrote:

NOG;
Your examples assume an equal EV, so none of them are at actually
variance

Should read "so none of them are actually at variance"

Here's a puzzle - why it it so much easier to spot the typos after
they're posted?
Skip

#13

or you had a choice between a flat top game, say full pay deuces wild,
and an 8/5 job progressive that was at +1%?

assuming infinite time and an infinite bankroll, of course you would
go with the +1% ev

but, who has infinite time and/or an infinite bankroll?

variance is as important as ev
the sharpe ratio is one way to measure that importance:
sharpe ratio = ev/sqrt(variance)

···

--- In vpFREE@yahoogroups.com, Skip Hughes <skiphughes@e...> wrote:

NOG;
Your examples assume an equal EV, so none of them are at actually
variance (sorry, couldn't resist) with my statement. That is, EV is
not a factor at all in those examples, so obviously variance is of
the greater importance (something > nothing).
Can you how me some examples where a lower EV is preferable to a
higher EV game (in the long run) because of a lower/higher variance?
Skip

nightoftheiguana2000 wrote:

> i'm gonna disagree with your statement "variance becomes much less
> important than EV"
>
> let's say you had a choice between two games of equal EV but one
> of
> them had a sequential royal and thus much higher variance, which
> game
> would you choose and why?
>
> or you had a choice between two progressives of equal EV, but one
> of
> them had an additional 1% tied up in the progressive?
>
> or you had a choice between a flat top game, say full pay deuces
> wild,
> and an 8/5 job progressive that was at +1%?
>
> it's a common misconception to think that in the "long run"
> variance
> doesn't matter, but variance is always there

#14

Hey Night;
As I went back and looked at this again, I think you kinda
misinterpeted what I was saying. My fault I'm sure, but I did say
(in my first sentence) "Variance always matters...". Actually, I
think we are really pretty much in agreement.

nightoftheiguana2000 wrote:

i'm gonna disagree with your statement "variance becomes much less
important than EV"

  I think that progressives examples actually prove my point. There
are at least a few local pros who specialize in playing
progressives. For those folks (and any folks who play on a daily
basis), playing a $1 7/5 Bonus progressive with a 101.2% EV and a
Variance of 104%, rather than a flat top Double Bonus with 100.5% EV
(including CB) is the best choice (as long as they have the
bankroll). For an occasional visitor, many would choose to bypass
the higher EV and choose the lower variance game. I would if I was a
visitor. However, visitors with a different attitude may choose the
EV the big jackpot brings. But they are taking a far greater risk. I
know a lot of players (including locals) who never play
progressives, because of the variance.
     
The whole point was that how important variance is, depends on many
variables. The right choice depends on individual circumstances,
attitudes, etc. For a player who plays every three - six months,
those factors are a little different than for those who play every
day.

Skip

#15

--- In vpFREE@yahoogroups.com, "nightoftheiguana2000"
<nightoftheiguana2000@y...> wrote:

or you had a choice between a flat top game, say full pay deuces

wild,

and an 8/5 job progressive that was at +1%?

assuming infinite time and an infinite bankroll, of course you

would

go with the +1% ev

but, who has infinite time and/or an infinite bankroll?

Well, I'll probably agree with that one (although the bankroll
required is definitely finite), since the difference in return edge
is only about 24% (1 - .76), but the difference in Variance is
almost 500% (134 - 28). Given a huge difference in variance and a
very small difference in return, even every day players would likely
choose the lower variance game. But the deck is stacked in that
comparison, if you's pardon the expression.

How about a comparison of flat-top 10/7/5 DB at 100.52%/29 and
16/13 "Downtown Deuces" at 100.92%/56? At about twice the EV edge
and and twice the variance, I think most daily players would easily
prefer the 16/13 DDW (all other things being equal). I know I sure
would and it would be no contest. For an occasional visitor, it
might not be quite so easy a decision.

In dollar and higher games the EV differences are often larger
(since there are almost no higher percentage flat-tops than 10/7 -
with some occasional exceptions of course - and most regular dollar+
progressive players are looking for a pretty significant EV (at
least 101+%).

  I certainly don't agree that variance and EV are of equal
importance. In practical applications it depends on the
circumstances and goals.
Skip

#16

How about a comparison of flat-top 10/7/5 DB at 100.52%/29 and
16/13 "Downtown Deuces" at 100.92%/56? At about twice the EV edge
and and twice the variance, I think most daily players would easily
prefer the 16/13 DDW (all other things being equal). I know I sure
would and it would be no contest. For an occasional visitor, it
might not be quite so easy a decision.

.52%var29:
Kelly Bankroll (~11%ror)=var/ev=29/.52%= 5577 bets
N0=var/ev^2=29/.52%^2= 1.1 million hands
Sharpe Ratio=ev/sqrt(var)=.52%/sqrt(29)= .1%

.92%var56:
Kelly Bankroll (~11%ror)=var/ev=56/.92%= 6087 bets
N0=var/ev^2=56/.92%^2= 661,626 hands
Sharpe Ratio=ev/sqrt(var)=.92%/sqrt(56)= .12%

the N0 for .92%var56 is a lot more favorable and the sharpe ratio
indicates it is a better bet although it does require a bit more
bankroll or more risk for the same bankroll

I certainly don't agree that variance and EV are of equal
importance. In practical applications it depends on the
circumstances and goals.

i agree, i'd go with the ev/sqrt(var) ratio which means variance is of
less importance than ev, a 4x difference in variance would be covered
by a 2x difference in ev

but, if you wanted to play to a fixed risk, then variance and ev are
of equal importance because bankroll is roughly proportional to
var/ev, a 2x difference in variance needs a 2x difference in ev for
the same risk and bankroll

···

--- In vpFREE@yahoogroups.com, "Skip Hughes" <skiphughes@e...> wrote:

#17

By definition gamblers are "risk loving" Not "risk averse" otherwise
they would not choose to enter into this hobby, so the use of sharp
ratio, which is of limited value in financial analysis is borderline
perverse in gambling anayalsis. Use of utility theory would be much
more applicable...gamblers actually find utility in high variance
bets...most people find JOB a low variance game rather boring and
perfer games with higher variance. Utility theory is a way to
reconcile people's "rational" choice of seeking games with higher
variance even with comparable ev...it's fun...they enjoy it...not a
wise investment strategy but perfectly fine for recreational purposes.

I would also question the over reliance on CLT approximation to the
normal distribution in calc. risk of ruin. CLT tends to fall apart in
the extremes of the distribution, it does work rather well around the
mean. I would not put much confidence in values 3 and 4 Standard
deviations out.
The possible outcomes on an individual hand of vp is very positively
skewed...-5 to 4,000...this is going to weaken the approx. to normal.
The ev calc. are good, I'm much more weary of the variance
calculations that are bandied about.

--- In vpFREE@yahoogroups.com, "nightoftheiguana2000"
<nightoftheiguana2000@y...> wrote:

variance is as important as ev
the sharpe ratio is one way to measure that importance:
sharpe ratio = ev/sqrt(variance)

> NOG;
> Your examples assume an equal EV, so none of them are at actually
> variance (sorry, couldn't resist) with my statement. That is, EV

is

> not a factor at all in those examples, so obviously variance is of
> the greater importance (something > nothing).
> Can you how me some examples where a lower EV is preferable to a
> higher EV game (in the long run) because of a lower/higher

variance?

> Skip
>
> nightoftheiguana2000 wrote:
>
> > i'm gonna disagree with your statement "variance becomes much

less

> > important than EV"
> >
> > let's say you had a choice between two games of equal EV but one
> > of
> > them had a sequential royal and thus much higher variance, which
> > game
> > would you choose and why?
> >
> > or you had a choice between two progressives of equal EV, but

one

···

--- In vpFREE@yahoogroups.com, Skip Hughes <skiphughes@e...> wrote:
> > of
> > them had an additional 1% tied up in the progressive?
> >
> > or you had a choice between a flat top game, say full pay deuces
> > wild,
> > and an 8/5 job progressive that was at +1%?
> >
> > it's a common misconception to think that in the "long run"
> > variance
> > doesn't matter, but variance is always there

#18

--- In vpFREE@yahoogroups.com, "jaydavidson118" <jaydavidson118@y...>
wrote:

I would also question the over reliance on CLT approximation to the
normal distribution in calc. risk of ruin.

i don't believe the formula for R(1):
SUM[Pi x R(1)^Wi]
uses clt

once you solve for R(1) (can be done with a spreadsheet),
ror=R(1)^bankroll

#19

Any ROR calc would have to take into account game variance... all
variance calcculations that I have seen relied on approximations to
normal distribution, which are based on clt...the normal approx. tend
to weaken greatly at the extremes of the distribution, which is where
ror calcs. are made...calcs based on the central part of the
distribution would be reasonabley accurate.
--- In vpFREE@yahoogroups.com, "nightoftheiguana2000"
<nightoftheiguana2000@y...> wrote:

--- In vpFREE@yahoogroups.com, "jaydavidson118"

<jaydavidson118@y...>

wrote:
> I would also question the over reliance on CLT approximation to

the

···

> normal distribution in calc. risk of ruin.

i don't believe the formula for R(1):
SUM[Pi x R(1)^Wi]
uses clt

once you solve for R(1) (can be done with a spreadsheet),
ror=R(1)^bankroll

#20

the formula cited below, credited to Evgeny Sorokin but found by
others including Jazbo Burns, does not use variance
it is my understanding that it is an exact solution but i have not
attempted to follow the derivation
it is possible the derivation is online somewhere
for full pay deuces wild the R(1) number is 0.999346831403995 and
represents the risk of losing a bankroll of one bet, to get the risk
of losing other bankrolls: R(bankroll)=R(1)^bankroll
Cindy Lui's calculator uses the R(1) number to solve ror:
http://www.gamblingtools.net/vp/vpanalyzer.html
another property of the R(1) number is that it can be used to find the
min-ror strategy, by adjusting the payoffs with this formula:
(1-R(1)^W)/(1-R(1)) and using something like vpsm to find the strategy

--- In vpFREE@yahoogroups.com, "jaydavidson118" <jaydavidson118@y...>
wrote:

Any ROR calc would have to take into account game variance... all
variance calcculations that I have seen relied on approximations to
normal distribution, which are based on clt...the normal approx.

tend

to weaken greatly at the extremes of the distribution, which is

where

···

ror calcs. are made...calcs based on the central part of the
distribution would be reasonabley accurate.
--- In vpFREE@yahoogroups.com, "nightoftheiguana2000"
<nightoftheiguana2000@y...> wrote:
>
> --- In vpFREE@yahoogroups.com, "jaydavidson118"
<jaydavidson118@y...>
> wrote:
> > I would also question the over reliance on CLT approximation to
the
> > normal distribution in calc. risk of ruin.
>
> i don't believe the formula for R(1):
> SUM[Pi x R(1)^Wi]
> uses clt
>
> once you solve for R(1) (can be done with a spreadsheet),
> ror=R(1)^bankroll