vpFREE2 Forums

Bob Dancer's LVA - 16 DEC 2014

#21

NOTI wrote: <<You might be tempted to play something that is a likely loss in the short
term if you're well ahead for the current tax year, but this is dangerous
because you could end up paying taxes on noise or variance, which is
negative EV.>>

Cogno asked: "I'm not getting this. Can you give an example?"

It's only an issue if you go into the red for a tax year and that loss was only due to variance, which presumably would be the cause if you had an edge. The implication is that next tax year you might get luckier, again due to variance, but end up having to pay taxes on that luck. Variance is typically symmetrical or close to it, so if variance causes a loss one tax year but a gain the next, you are paying taxes on the variance, which is an additional negative EV. You want to pay taxes on EV, not on the variance, variance is a false gain, but can be taxable.

Example 1: your gamble for a tax year amounts to an even gamble but with a deviation of $200. So, for example, you could lose $200 one year and win $200 the next, but you would pay taxes on a $200 false gain, not your true gain of 0. You thought you had an even gamble but you don't, instead you are paying taxes on an illusionary gain.

Example 2: your gamble for a tax year amounts to an edge of +$100 but with a deviation of $200. So, for example, you could lose $100 one year and win $300 the next, but you would pay taxes on a $300 false gain, not your true gain of $200. You are paying taxes on your edge and your variance.

[Non-text portions of this message have been removed]

#22

OK, I misunderstood. There are no adverse tax consequences to a
high-variance play the last day of the tax year that does not have the
possibility of moving you into a different tax bracket (or a net loss).

···

-----Original Message-----
From: vpFREE@yahoogroups.com [mailto:vpF…@…com]
Sent: Thursday, December 18, 2014 1:56 PM
To: vpFREE@yahoogroups.com
Subject: RE: [vpFREE] Bob Dancer's LVA - 16 DEC 2014

NOTI wrote: <<You might be tempted to play something that is a likely loss
in the short term if you're well ahead for the current tax year, but this
is dangerous because you could end up paying taxes on noise or variance,
which is negative EV.>>

Cogno asked: "I'm not getting this. Can you give an example?"

It's only an issue if you go into the red for a tax year and that loss was
only due to variance, which presumably would be the cause if you had an
edge. The implication is that next tax year you might get luckier, again due
to variance, but end up having to pay taxes on that luck. Variance is
typically symmetrical or close to it, so if variance causes a loss one tax
year but a gain the next, you are paying taxes on the variance, which is an
additional negative EV. You want to pay taxes on EV, not on the variance,
variance is a false gain, but can be taxable.

Example 1: your gamble for a tax year amounts to an even gamble but with a
deviation of $200. So, for example, you could lose $200 one year and win
$200 the next, but you would pay taxes on a $200 false gain, not your true
gain of 0. You thought you had an even gamble but you don't, instead you are
paying taxes on an illusionary gain.

Example 2: your gamble for a tax year amounts to an edge of +$100 but with a
deviation of $200. So, for example, you could lose $100 one year and win
$300 the next, but you would pay taxes on a $300 false gain, not your true
gain of $200. You are paying taxes on your edge and your variance.

[Non-text portions of this message have been removed]

------------------------------------
Posted by: nightoftheiguana2000@yahoo.com
------------------------------------

vpFREE Links: http://www.west-point.org/users/usma1955/20228/V/Links.htm

------------------------------------

Yahoo Groups Links

#23

Bob wrote: "I'm way ahead this year. I lost last year."

I assume you had an edge in both years, so your loss last year was due to variance. Variance is close to symmetrical, so it's probably expressing itself in a positive way in your results this year, but you will pay a tax on it. Paying tax on variance is a negative EV because there is no legal carry forward of gambling losses. You could have done some planning to avoid that negative EV. Essentially you made a strategy error last year (that error was taking on more variance than you had edge) and that error is expressing itself in addtional tax you will pay this year.

[Non-text portions of this message have been removed]

#24

Cogno wrote: "OK, I misunderstood. There are no adverse tax consequences to a
high-variance play the last day of the tax year that does not have the
possibility of moving you into a different tax bracket (or a net loss)."

Right. If a play causes a change in tax rate, such as not getting a credit for a loss, then you have to consider that change and recalculate your true net EV. Your true net EV is the one that counts for the Kelly criterion. Or another way of thinking of it, for Kelly, the money that goes into your gambling bankroll is the money that counts, other diversions are simply leaks.

[Non-text portions of this message have been removed]

#25

Given an infinite sample size, the net average result is greater than the net standard deviation for a positive per hand EV. Of course it would seem unreasonable to assume an infinite sample size, what with the unlikelihood of casinos in the afterlife and what not. Fortunately, one can use math to solve for the sample size at which the average result is equal to the standard deviation. And that formula is variance/edge/edge which has also been called the Nzero point. So, for example, FPDW for one hand has an average result of +.76% and a standard deviation of 5.08 (sd=sqrt(var)). That's in units of bets, so if you were playing for 5 quarters, the average would be .0076 x $1.25 = $.01 and the standard deviation would be 5.08 x $1.25 = $6.31 . So you can clearly see that at one hand the standard deviation just swamps that tiny average result. But, fortunately, the Nzero is a reasonable (compared to infinite) 25.84/.0076/.0076= 447,368 hands. The question arises as to how long you have to get to the Nzero point, is it your entire gambling lifetime? I would say no, the IRS wants you to report your gambling results on an annual basis and they don't allow carry forward of losses, so you'd better get to Nzero before the end of the tax year or you'll be paying a penalty if your result is a loss due to variance. Alternately if you are already ahead this tax year, you have a buffer, as the tax problem is only a problem is you show a loss for the tax year. But the problem is very real, the penalty is approximately your marginal tax rate on a winning year times your loss on a losing year. You pay the penalty on the winning year so it is somewhat hidden.

[Non-text portions of this message have been removed]

#26

Thank you NOTI. For me this post was one of the most informative I've ever read on vpFREE.

My math skills are rusty and never really extended broader or deeper than mid-level calculus and algebra. Your explanation was easily understood. It also gave substance to a nagging feeling on my part that high variance has a hidden cost separate from bankroll considerations.

Well done!

G'luck all,
Gamb00ler

nightoftheiguana2000 wrote :

Given an infinite sample size, the net average result is greater than the net standard deviation for a positive per hand EV. Of course it would seem unreasonable to assume an infinite sample size, what with the unlikelihood of casinos in the afterlife and what not. Fortunately, one can use math to solve for the sample size at which the average result is equal to the standard deviation. And that formula is variance/edge/edge which has also been called the Nzero point. So, for example, FPDW for one hand has an average result of +.76% and a standard deviation of 5.08 (sd=sqrt(var)). That's in units of bets, so if you were playing for 5 quarters, the average would be .0076 x $1.25 = $.01 and the standard deviation would be 5.08 x $1.25 = $6.31 . So you can clearly see that at one hand the standard deviation just swamps that tiny average result. But, fortunately, the Nzero is a reasonable (compared to infinite) 25.84/.0076/.0076= 447,368 hands. The question arises as to how long you have to get to the Nzero point, is it your entire gambling lifetime? I would say no, the IRS wants you to report your gambling results on an annual basis and they don't allow carry forward of losses, so you'd better get to Nzero before the end of the tax year or you'll be paying a penalty if your result is a loss due to variance. Alternately if you are already ahead this tax year, you have a buffer, as the tax problem is only a problem is you show a loss for the tax year. But the problem is very real, the penalty is approximately your marginal tax rate on a winning year times your loss on a losing year. You pay the penalty on the winning year so it is somewhat hidden.

[Non-text portions of this message have been removed]