To partially ameliorate the second difficulty one
could recompute the expectations for the two-card
hands which include the split card after artificially
removing another card of the split denomination from
the deck. Use these expectations as surrogates for
the result of making exactly one split into two hands.
Then appeal to the less reliable pseudo-expectations
of the first paragraph of this section to provide
a gradient with the purpose of inferring the outcomes
if splitting produced three hands, four hands, etc.
This increases computing time, but presumably results
in greater accuracy. incorporating this, along with
the interactive approximations to dealer probabilities
of Method B, results in a seven-minute run (rather
than just three minutes without the extra pair splitting
activity) which assesses the singledeck expectation
as +.000422. Compare this to the precise value of
+.000425 for composition-dependent basic strategy.
The exact computations required about 10 hours of
computer time, the vast majority of it being the evaluation
of pair splitting, assuming that the proper splitting
strategy is unknown at the outset. (That is, i "try"
to split 4s against a ten, then discover it isn't
correct, just as i try to draw to each total of hard
20 in the main part of the program, learning anew
of its futility.)
biggest problem with poker newsletters is that while
some of them contain useful poker information, others
are virtually mere banner farms. There's a third group
that are useful in limited ways: they're just "commercial"
news, advising online players about current deals
available at online cardrooms.
For four years I published a bi-weekly newsletter
called Wednesday Nite Poker, but things change on
the Internet and my old sponsor and I had a parting
of the ways, albeit friendly enough. My newsletter
should be back by the time you read this (under a
different name, but using the same format). To find
out where visit Andy Glazer. com. It's free.
Two of the sites mentioned previously, CardPlayer.com
and PokerPages.com, also offer newsletters which are
worth checking out. Beyond that, I'm not willing to
speculate six months into the future.