paladingamingllc wrote:
Concur, but with respect to Jazbo's theorem regarding
variance/covariance of multiline poker (covar is about .1 per line
according to him) I'd like to make a couple of points.
Jazbo's n-play webpage
http://jazbo.com/videopoker/nplay.html
is, of course, the source of my statements. (Getting a comprehensive
handle on the n-play math and calculation of the related co-variance
is another backlogged goal 
I assume your parenthetical is meant to say that co-var runs about 10%
of variance (correlation coeff).
In my experience, the covariance seems to take over at around 20
lines, not 10.
My 10-line comment is based upon Jazbo's equation where, when n=11,
the contribution of each to total variance is roughly equal (which is
a direct consequence of the 10% relationship noted above).
However, variance is a single point descriptor of how results
fluctuate around the mean and, as often is noted, the vp distribution
curve isn't normal, particularly in the short-term. I can well see
where the skewness of the curve, combined with the altered result
expectation in any given period that excludes a royal hit, would alter
the point (number of lines played) at which covariance becomes the
dominant factor.
When you remove the huge contribution of the royal to both variance
and covariance, I'd expect the ratio between the resulting variables
to shift dramatically. Given my 10-play experience, I can easily see
where that point would be a reasonably higher value.
Bear in mind that over time (likely well within 1 mm 10-play deals
with most games) the distribution curve begins approximating a normal
distribution (as the overall occurance of royals begins to be
much more predictable). At that point, covariance would become more
significant relative to variance ... and I'd look for it to follow the
math and dominate for play in excess of 11-lines for most games.
I look at playing 100-line more as a time/space
problem, yes, a quantum physics issue, rather than a
mathematical/statistical one. You just can't press the buttons fast
enough to get more than 600 hands an hour. I play single line
probably about 2x faster. Getting in a large number of deals, not
necessarily hands, per hour reduces ones variance as well. Not
variance by hand, but variance by hour.
Absolutely. I tend to refer to the speed effect on session bankroll
variance as being $-variance/hr. (var x $bet/hr.)
Finally, as the Joker games are popular in AC, because the Joker is
so much more valuable than any other card in the deck, Joker and DJ
games have a much higher covariance than, say, Jacks. I think Jazbo
would agree with this.
Jazbo publishes a list of var/covar values for various games. Given
that AC-JW5K doesn't become interesting until the 5K progressive meter
advances ER to 100%, he lists values for a 100% and 103% play.
The values are:
100%: 124.2/9.0
103%: 200.6/13.5
As the 5K meter advances, covar actually diminishes relative to
variance. But, once again, when you remove the jackpot I look for
covariance to be all the stronger vs. variance. Since the prog isn't
found in anything higher than 3-play, I'd still look for variance to
dominate in the short run.
As far as DJ, the var/covar numbers are appreciably smaller than those
of a game such as DB or AA (which are relatively modest themselves).
In the Deuces games, there are four of them, which would help bring
the covar back to normal.
Interestingly, Deuces has the highest ratio of covar/var of the games
tabulated by Jazbo (DB runs a close 2nd). That's consistent with my
3-play UD prog experience at NYNY -- my initial introduction to that
play was quite rougher than I'd anticipated (though clearly tamer than
my 3-play DB experience).
Consider that one AC casino offers a 99.99 version of 5 Joker, and
consider how low the variance is on that game, it's like Pick'em, I
believe.
In the short run, I'd suggest like flipping coins by comparison (with
a steady meter decline between the occasional strong hits once or
twice an hour that can be as disheartening as 6 days of rain
Of course, with one of the weakest cb/comp/ratings to be found in AC,
it's largely a $1 novelty play.
- Harry
