MJ: CVCX Question: Why does KO-P destroy Hi-Lo with a departure adjustment?

[MJ]

I compared KO-P and Hi-Lo (using the canned sims that came with the software) in different situations. SCORE was about even in all situations except for when a departure adjustment was used.

Rules: 6D, DAS, DA2, LS 75% Pen
Departure Adjustment: 20 hands
BR- $20k
Kelly: 0.50

With both systems the wong in point was at whichever point maximized SCORE. The results are as follows:

Hi-Lo
Wong in point: TC = +1
SCORE: $89.99

KO-P
Wong in point: RC = -5
SCORE: $149.90

Thats a HUGE difference! Can somebody explain this?

On a smaller note, I noticed something else unusual with KO-P. Look at the SCORES for the following wong in points:

RC = -7 SCORE = $140.76
RC = -6 SCORE = $147.62
RC = -5 SCORE = $149.90
RC = -4 SCORE = $147.35
RC = -3 SCORE = $149.97
RC = -2 SCORE = $148.39

Look at what is occuring between -5 and -3. Why does the SCORE decrease at -4 and then increase at -3? After -3 it continuously decreases as you might expect. Very erratic behavior.

MJ

[Norm Wattenberger]

> Thats a HUGE difference! Can somebody explain this?

The departure's are not at equiv. points. No reason they should be comparable. Look at the Help for Departure Adjustment.

> Look at what is occurring between -5 and -3. Why does
> the SCORE decrease at -4 and then increase at -3?
> After -3 it continuously decreases as you might
> expect. Very erratic behavior.

Not surprising. When calculating backcounting SCOREs over a range of counts, percentage of hands played enters an already complex equation. With the large range of counts in an unbalanced strategy; I'm not surprised that the curve wobbles. I wouldn't expect this with HiLo. It's also far more likely to occur if you use any of the Simplify options as CVCX has to come up with a unit size and simplified betting structure that approximates the requested risk.

[MJ]

> The departure's are not at equiv. points. No reason
> they should be comparable. Look at the Help for
> Departure Adjustment.

Thank you for the response. Ok, I looked at the Help section.

"When selected, an estimate is made of the SCORE and Win Rate as if you exited the table when the count has dropped to the first row of counts displayed (normally -1)".

Does that mean for KO-P I exit the table at RC = -9, which is the first row of counts displayed?

It would still seem that using a departure strategy for KO as outlined by CVCX is far more profitable than that for Hi-Lo. Can you suggest a departure point for Hi-Lo that will yield an equivalent SCORE as that for KO? If not, then KO is the clear winner here.

> Not surprising. When calculating backcounting SCOREs
> over a range of counts, percentage of hands played
> enters an already complex equation. With the large
> range of counts in an unbalanced strategy; I'm not
> surprised that the curve wobbles. I wouldn't expect
> this with HiLo. It's also far more likely to occur if
> you use any of the Simplify options as CVCX has to
> come up with a unit size and simplified betting
> structure that approximates the requested risk.

Does that mean that the SCORE really wobbles when wonging with unbalanced strategies or is it just a minor shortcoming of the software given the complex equations in use?

MJ

[MJ]

> The departure's are not at equiv. points. No reason
> they should be comparable. Look at the Help for
> Departure Adjustment.

Thank you for the response. Ok, I looked at the Help section.

"When selected, an estimate is made of the SCORE and Win Rate as if you exited the table when the count has dropped to the first row of counts displayed (normally -1)".

Does that mean for KO-P I exit the table at RC = -9, which is the first row of counts displayed?

It would still seem that using a departure strategy for KO as outlined by CVCX is far more profitable than that for Hi-Lo. Can you suggest a departure point for Hi-Lo that will yield an equivalent SCORE as that for KO? If not, then KO is the clear winner here.

> Not surprising. When calculating backcounting SCOREs
> over a range of counts, percentage of hands played
> enters an already complex equation. With the large
> range of counts in an unbalanced strategy; I'm not
> surprised that the curve wobbles. I wouldn't expect
> this with HiLo. It's also far more likely to occur if
> you use any of the Simplify options as CVCX has to
> come up with a unit size and simplified betting
> structure that approximates the requested risk.

Does that mean that the SCORE really wobbles when wonging with unbalanced strategies or is it just a minor shortcoming of the software given the complex equations in use?

MJ

[Norm Wattenberger]

> Does that mean for KO-P I exit the table at RC = -9,
> which is the first row of counts displayed?

Yes

> It would still seem that using a departure strategy
> for KO as outlined by CVCX is far more profitable than
> that for Hi-Lo. Can you suggest a departure point for
> Hi-Lo that will yield an equivalent SCORE as that for
> KO? If not, then KO is the clear winner here.

That's happenstance. You are trying to equate KO RCs with HiLo TCs. You can look at the advantages for each count for HiLo and KO using CVCX and find the approximate equivalent. But it may vary by penetration.

> Does that mean that the SCORE really wobbles when
> wonging with unbalanced strategies or is it just a
> minor shortcoming of the software given the complex
> equations in use?

The software must come up with a ramp given the constraints that you set and the constraints of integer counts since that's what people use. The degree of 'wobble' depends on the level of constraint. Smaller bankroll, tougher simplification rules, smaller Kelly fraction all add to the constraints. The software will give you the best answer given the constraints that you set.

[MJ]

> Yes

> That's happenstance. You are trying to equate KO RCs
> with HiLo TCs. You can look at the advantages for each
> count for HiLo and KO using CVCX and find the
> approximate equivalent.

Why bother? I thought the optimal departure point for Hi-Lo is TC = -1. In other words, no other departure point can yield a higher SCORE. Even if we wanted to use a different departure point, CVCX precludes this possibility by using the first row of counts by default so how can we test it?

> The software must come up with a ramp given the
> constraints that you set and the constraints of
> integer counts since that's what people use. The
> degree of 'wobble' depends on the level of constraint.
> Smaller bankroll, tougher simplification rules,
> smaller Kelly fraction all add to the constraints. The
> software will give you the best answer given the
> constraints that you set.

Here is another odd situation. A while back, I was using CVCX to figure out an optimal bet schedule for Hi-Lo. At TC = +6 the bet decreased, and then increased again at TC = +7!
Rounding was used and possibly double tags. Any idea why this would occur?

BTW, there are a couple run time errors on CVBJ when using the simplify bet schedule. I'll post them later.

MJ

[MJ]

[Norm Wattenberger]

> Why bother? I thought the optimal departure point for
> Hi-Lo is TC = -1. In other words, no other departure
> point can yield a higher SCORE. Even if we wanted to
> use a different departure point, CVCX precludes this
> possibility by using the first row of counts by
> default so how can we test it?

By running another sim with a different starting row. But CVCX wasn't really designed for this level of detail for Wonging. CVCX is a post-sim calculator. The comparisons you are talking about involve modifications that affect TC frequencies in various ways depending on the number of other players and exactly how you wong in and out. CVData is required for this level of detail.

> Here is another odd situation. A while back, I was
> using CVCX to figure out an optimal bet schedule for
> Hi-Lo. At TC = +6 the bet decreased, and then
> increased again at TC = +7!
> Rounding was used and possibly double tags. Any idea
> why this would occur?

This can occur if there is a very low frequency for one of the counts. It depends on exactly how you have TC calculations set up. It is common with single deck.

> BTW, there are a couple run time errors on CVBJ when
> using the simplify bet schedule. I'll post them later.

That surprises me. None have been reported. If you get an RT error; please send me details of the settings.