MJ: CVData Optimal Departure

[MJ]

I ran around 30 simulations for wonging with KO-P experimenting with different entry and exit points. The best I could come up with is the counter wonging in at RC = -2 and exiting at -8 (assuming 6 round lag), yielding a SCORE of 43.26. This is for AC rules with 6D.

If CVData supported departing while observing a shoe (assume 6 round lag time), by about what % would SCORE improve? Clearly, EV would improve from playing a few more rounds per hour, but SCORE is not EV. While EV increases, SD will also increase right along with it. If both EV and SD increase by the same proportion, then SCORE would technically remain the same, correct?

According to the BJA3 ODP study on Table 13.9 under "Shoes Never Entered" for WiWo 4.5/6(lag) it says 57.26%. On pg 354 this row is explained, "...tells us how many shoes get ditched while we're just watching...". So, 57% of shoes are abandoned while we are simply observing. If the % is so high, it makes me wonder how my simulation WR/SCORE would have been affected if this feature was added to CVData.

Anyway, I hope this drives home the importance of including departure points/lag times while observing. I don't know how far down it is on the list, but hopefully it gets bumped up a few spots. :-)

MJ

[Don Schlesinger]

> Clearly, EV would improve from playing a few
> more rounds per hour, but SCORE is not EV.

That's correct, but, actually, you're making a correct statement, but for the wrong reason, as you'll see in a minute.

> While EV increases, SD will also increase right along with > it.

But not in the same proportion. If you play twice as many hands, your e.v. doubles, but your s.d. increases by only sqrt(2), which 1.414. Now, SCORE is the square of DI, which, in turn, is the ratio of the e.v. and the s.d. (that is, DI = e.v./s.d.), with an extra, arbitrary, factor of 100 in the numerator, to make the numbers more manageable. So, by playing more hands, in fact the DIs and SCOREs would not be the same, except for one important item, which I guess I've discussed, oh, only about a million times (see below).

> If both EV and SD increase by the same proportion,

Which they don't (see above).

> then SCORE would technically remain the same, correct?

Here's the point: Once you pick a game, SCORE remains the same no matter how fast you play, because SCORE can't be altered by playing faster or better, according to the technical definition of SCORE (see BJA3, "More on SCORE," pp. 177-183). So, above, you might discuss how your risk-adjusted win rate does, or doesn't, improve (it does), but the SCORE, technically, doesn't change.

Don

[MJ]

Thanks Don. I understand all that. But, how much do you think risk adjusted WR or C-SCORE would improve if the counter could depart while observing? In my simulation, the SCORE worked out to be 43.26 for WiWo using a 6 round lag when exiting the shoe.

Ok, one other question. I would assume kelly-factor does not affect C-SCORE. I input my bet ramp into CVData based upon what CVCX gives me. Regardless of what kelly-factor I set my CVCX, whether it be 0.25, 0.5, 1.0, etc, shouldn't the C-SCORE on CVData be unaffected? My thinking here is that the EV and SD increase/decrease proportionately as K-f changes.

MJ

[Don Schlesinger]

> Thanks Don. I understand all that. But, how much do
> you think risk adjusted WR or C-SCORE would improve if
> the counter could depart while observing?

The answers to all these questions are, "A little bit." In BJA3, I explained that, although, technically, the results from ODP while watching, as opposed to while playing, are different, they aren't different enough to get worked up over.

> Ok, one other question. I would assume kelly-factor
> does not affect C-SCORE. I input my bet ramp into
> CVData based upon what CVCX gives me. Regardless of
> what kelly-factor I set my CVCX, whether it be 0.25,
> 0.5, 1.0, etc, shouldn't the C-SCORE on CVData be
> unaffected? My thinking here is that the EV and SD
> increase/decrease proportionately as K-f changes.

Correct.

Don

[MJ]

> The answers to all these questions are, "A little
> bit." In BJA3, I explained that, although,
> technically, the results from ODP while watching, as
> opposed to while playing, are different, they aren't
> different enough to get worked up over.

I understood you to write (in BJA3) something to the affect of the ODPs while observing are different from those while actually playing, but not enough to make a real difference. This is why most Hi-Lo players use a departure point of TC = -1 regardless if they are observing or playing.

Just to make sure I understand you here, are you saying that even if a counter never departs while observing, this would make little difference to C-SCORE (maybe cost a buck or two)?

Perhaps I misunderstood what I read in BJA3(see below), but if 57% of shoes are abandoned while observing, that means we are giving ourselves a chance to observe more shoes with the likelyhood of actually entering more shoes. But with CVData, we never get to abandon those 57% of shoes where the count tanks. Now, if you say that it makes little difference to C-SCORE, then I'll take your word for it. I just want to make certain we understand each other.

"According to the BJA3 ODP study on Table 13.9 under "Shoes Never Entered" for WiWo 4.5/6(lag) it says 57.26%. On pg 354 this row is explained, "...tells us how many shoes get ditched while we're just watching...". So, 57% of shoes are abandoned while we are simply observing".

MJ