cvcx optimal betting schedule

[Charlie Mosby]

I've got a couple of questions (and more!) about using cvcx to determine the optimal betting spread and unit size.


Some perhaps unnecessary details are thrown in, but I have highlighted the real questions.........just look for the larger fonts when you scroll down




Thanks!




0. --------- running sims myself -----------
This is just a hypothetical question....I imagine that one would need to simulate an entire array of parameters (for a given set of rules) to be able to provide a functionality that is similar to that of using the archive, right?


Can someone briefly describe what the process is like?










1. Main question: using the archive


I'm using the archive for hi-lo created in 2007, at the bottom of the list of the 'included sims'


------ 1-1: CE/WR --------
I've read BJA3 and other sources about the basics of CE and I know that in general the higher the CE the better, but I haven't seen much discussions on the internet about CE/WR. It reads in the help that "when CE/WR is less than 0.5 the risk is higher than what you want."


I played around with the numbers a bit and found that even with acceptable RoR (like 0.5~3%) and score (like 70~80) the CE//WR ratio can be quite low, like 0.1~0.3 (sometimes even negative?!). Very often when the betting unit is bigger or the spread is larger, CE will be higher, but that is more than offset by the wager size, rendering a low CE/WR.


I'm wondering how one should really take this CE/WR into account (besides RoR and score) when determining the optimal betting spread.
(For the casino rules and bankroll I play with, I can barely get a CE/WR over 0.5 and have spent many hours trying to strike a balance. Wonging in at tc = 1 or tc = 1.5 doesn't make much of a difference in getting a higher CE/WR.....well, there is considerable difference, but my CE/WR are still low whatsoever.)










------ 1-2: max bet --------
The optimal bet size is f = adv/var, but at some point we need to cap it (end the ramp) and we call this the max bet.
Traditionally people said divide your entire bankroll by 150 (or 50, or 100, whatever) to make your max bet, but that was back when people didn't have tools to obtain optimal schedule (and people did things like dividing the bankroll by 400 to get the unit bet size).

So now, officially, using cvcx, given a set of rules (S17 or H17, RSA or not, etc), what should my max bet be? According to the RoR or CE/WR I feel comfortable with?



(of course, the traditional way is not very far off, which produces a spread for my bankroll a RoR of just about 10% if play-all.... btw I know this is WAY too high from a more serious professional point of view....but hey, Don talks about 13.5% all the time in his book!)






------ 1-3: table rule --------
The feature that allows the user to start playing two hands after the specified tc is very nice, however, the common table rule that requires you to play at least 2 times the table minimum for each spot really interferes with it......is there any way to add this constraint? (when using the archive in cvcx viewer)

[Norm]

1-1: Generally, you want CE/WR to be 0.5. If you let CVCX do it's thing and don't add unusual constraints, it will be 0.5. Or as close as it can get using the usual constraints (e.g. bankroll).
1-2: Actually, f = adv/var, is an oversimplification. But close. In any case, if you let CVCX do its thing, it will tell you the max bet for the given spread, bankroll, risk, rules, penetration, etc.
1-3: No.

[Charlie Mosby]

[QUOTE=Norm;12704]1-1: Generally, you want CE/WR to be 0.5. If you let CVCX do it's thing and don't add unusual constraints, it will be 0.5. Or as close as it can get using the usual constraints (e.g. bankroll).
1-2: Actually, f = adv/var, is an oversimplification. But close. In any case, if you let CVCX do its thing, it will tell you the max bet for the given spread, bankroll, risk, rules, penetration, etc.

1-2-1
By the oversimplification, are you talking about the denominator being actually the 2nd moment instead of the variance?

1-2-2
sorry let me rephrase:

I don't specify the RoR but always specify manually the unit(minimum) bet size, which will be the table minimum or 2 times table minimum for playing two hands.


Along with the spread (the maximum multiple of the unit bet size) specified, this will fix my max bet prior to using cvcx.

For a given pair of unit size and spread, the ramp in the middle will be determined by cvcx.


My question is, what kind of choice of {unit, spread} is optimal, given the bankroll and the game rules? How should I utilize the resultant RoR and CE/WR combination to make a decision? Note that when Wonging, a spread of 1-6 and 1-16 makes very little difference in the score, however, there got be some way to decide which is better, right? (ignoring possible heat issues with 1-16)

(the real question is, where to cap the ramp? Since as a low roller I always start with table minimum even when Wonging, I just need to determine the spread size then let cvcx work other numbers out. )


Thanks.

[Norm]

1-2-1: Optimal bet is KEB*EV/Var where KEB is the Kelly equivalent bank.
1-2-2: Normally, CVCX will determine the optimal min and max. If you force the min, to obtain the optimal spread, just change the spinner and look for the highest SCORE.

[Charlie Mosby]

1-2-2: Normally, CVCX will determine the optimal min and max. If you force the min, to obtain the optimal spread, just change the spinner and look for the highest SCORE.

But the score basically grows without bounds with larger and larger spread...... so I'll just stop at the resultant RoR that hits the boundary of my comfort zone?


Also, what about CE/WR?
Some slightly higher scores have very low CE/WR, like 0.07 or 0.12, sometime even negative!! (but as you said it should remain at 0.5)

I'm wondering if it was a bug. I know for some combination of numbers you'll get negative CE, but hear me out......I just downloaded the update for cvcx (update575), and nowforexactly the same game rules and parameters, I get very different CE from before, while other outcome (ave. bet size, win rate, etc) are all the same.

Let me sent you the screen shots.

[Norm]

But the score basically grows without bounds with larger and larger spread...... so I'll just stop at the resultant RoR that hits the boundary of my comfort zone?


Also, what about CE/WR?
Some slightly higher scores have very low CE/WR, like 0.07 or 0.12, sometime even negative!! (but as you said it should remain at 0.5)

I'm wondering if it was a bug. I know for some combination of numbers you'll get negative CE, but hear me out......I just downloaded the update for cvcx (update575), and nowforexactly the same game rules and parameters, I get very different CE from before, while other outcome (ave. bet size, win rate, etc) are all the same.

Let me sent you the screen shots.

Sorry, don't know what you mean. Clearly CE/WR cannot remain at 0.5 if you set restraints. CVCX with no restraints calculates optimal betting according to your risk and other settings. When you start setting constraints, it does the best that it can to maximize SCORE.

[Norm]

for exactly the same game rules and parameters, I get very different CE from before, while other outcome (ave. bet size, win rate, etc) are all the same.

When you turn on Manually Adjust Min Bet, the Kelly Factor box is removed since you can’t set a KF if you are forcing a min bet. CE is based on your Kelly Factor. But, there is none specified. I assume the last KF that was set. So, although those two screens look like CE is the only difference, the non-visible, assumed KF is also different.

I added CE/WR for stat-types. SCORE is more important. CE probably shouldn't be displayed at all without a KF.