Poker Poker Poker!: [vpFREE] Re: Long term vs short term again

greeklandjohnny wrote:
> This topic pops up from time to time. Part of the reason we keep
> discussing this is that there is no consensus on what 'long term'
> really means.

I understand John's point, but I'd approach it a little differently:
"long term", as an isolated term, has little meaning. It's only in
the context of the probability that actual results will approach
expected results within some defined threshold of variance (as in "+/=
1%") that there's any true definition.

As he goes on to indicate, that tolerance has to be specified (as do
the game specifics being evaluated -- type, denomination, etc.) John
mentions his contribution to the VP FAQ on the subject -- it's an
intelligent offering: http://members.cox.net/vpfree/FAQ_LT.htm

One of the key weaknesses of the "long term" concept that he points
out is that even having played to the long term (under a given defined
threshold scenario), your exposure to an actual loss in dollar terms
generally proves to be rather immense.

He cites, by example, the calculations that were sourced from Tom Ski
in the general FAQ on this topic (it's possible that John's numbers
were calculated independently): Playing 10/7 DB, for a 95% confidence
that results will fall within 1% of expectation, a little over 1
million hands of play are required. For a tolerance of only a 0.1%
variance, over 10 millions hands must be played.

Under either long term definition (a tolerance for a 1% deviation, or
the more restrictive .1%), a shortfall of over 50,000 coins from
expected ending cash balance is within that tolerance. Whether
quarters, dollars, or whatever, that's a poor sense of comfort.

------

Surprisingly, that's seldom the angle from which critics of "advantage
play" approach their arguments. They argue that traditional
definitions of the "long term" simply aren't realistically achieved in
play.

They, of course, rely upon a nebulous concept of "long term", for with
many low to medium variance games (such as Jacks, DB, Deuces) a
long-term defined as a 1% variance tolerance is well within the realm
of an active player during their lifetime of play.

But whether the criticism is based upon the length of the "long term",
or the magnitude of the actual variance a player is exposed to once
the "long term" is reached, there's a very effective comeback in
support of the advantage play approach that should dismiss any
concern on these counts in the mind of someone approaching the subject
rationally.

I suggest that few, if any, players contemplating advantage play are
terribly concerned with how closely results approximate expected
results. They simply want reasonable assurance that over the long
haul they won't be subject to undue losses -- and ideally achieve
better than breakeven in their play.

"nightoftheiguana" (aka NOTI or "iggy") introduced the concept of "NO"
some time ago in this forum. NO represents the number of hands that
must be played for a player to have strong expectation (84%) of
positive results. This is a measure of the "long term" that is much
more practical than one defined by a variance from expected results.
And for the player who assiduously strives for a significant advantage
in their play (say, a 10/7 DB who plays with a minimum of .5% in
cb/bonus, or a 9/6 Jacks player that seeks a minimum threshold of .8%
in cb/bonus for play), such a "long term" can be surprisingly short.

As noted in the FAQ on the subject, NO is calculated as: var/((er-1+cb)^2)

So, for 10/7 DB + .5% (with a variance of 28), NO is 625 thou hands.
For 9/6 Jacks + .8%, NO is 800 thou hands.

To put these numbers in perspective, an active Jacks player who plays
800 hands per hour, 5 hours per week, would achieve NO expectation
within 4 years. The same would be true of a couple who play 10-12
hours each monthly.

And consider the FPDW player w/ .2 cb: NO = a mere 252 thou hands.

I find these numbers place the "long term" into a realm that serious
players should consider far from an abstract concept.

- Harry