Kelly bet ramp & variance for side bets

I'm trying to find the optimal bet ramp for a side bet, but am not sure if what I am doing (same method as for regular BJ) makes sense.

Let's assume you have a side bet of suited BJ pays 75 to 1.
Off the top, odds of a suited BJ are ~ 84 to 1, so you'd be at an ~ 11% disadvantage. If you put money on this side bet off the top, variance would be ~ 68 (from a simple variance calculation). I am ignoring the BJ hand variance for now, looking only at the side bet variance.

Now for creating a betting ramp: if we look at a high count, for example when instead of the odds of a suited BJ being 84:1 they are 60:1, the same variance calculation shows variance of ~ 95.

This is more or less in line with what CVD sims throw up: is it correct to use EV with these variances for an optimal betting ramp ? Or are these variances too large to be applicable for kelly bet sizing ?

Thanks

D.

For simple odds payout (i.e. only 2 possibilities, lose 1 unit or win X units), simply determine your % advantage and divide that by the payout and bet that percentage of your bankroll. This would be full Kelly. If you want to bet half Kelly, bet half as much, quarter Kelly, bet a quarter as much, etc.

In your example, you have a 1 in 61 chance of winning your bet and it pays 75 to 1. Your advantage is 15/61=.246. Divide .246 by 75 and you get .328% so for full Kelly, you bet .328% of BR. If you BR is 10k, that would be $32.80.

For the most part you will find with sidebets, that if they are somehow vulnerable to the count or a specialized count, the edge is so volatile, that during the rare times when you have an edge, your edge is huge, so unless you have a tiny BR, bet-sizing is not much of an issue. More important is to develop a proper count and approach, be *very* certain if/when you have the edge, and then bet as much as you can and hope for the best.

Hope this helps!

Fine tuning your advantage

If you are serious about this side bet a way to determine the distribution of suits is important to factor into the equation. I am not going to worry about the math but a team mate who keeps two counts that have a simple +1 and -1 for the two black suits and the two red suits should flag most serious imbalances in the distribution of suits.

Perhaps have him flat bet at minimum and when your bet grows large enough and your count shows an advantage in BJ likelihood you bet a predetermined bet in your ramp which would have him indicate the amount of any imbalance by an increase in bet to a predetermined amounts. After which you could place side bets before the dealer deals the cards.

A count that is ace neutral with ace side count would be much more effective at predicting the actual likelihood of a blackjack than an ace reckoned count. The infamous JSTAT(2 thru 9 => +1, 10 thru K => -2) I think it is called would be the most accurate. These fine tuning adjustments should make the side bet much more profitable.

+1 to everything.

-Sonny-

Sim of derived betting ramp for sidebet

Thanks - I put together a betting ramp using this approach - EV/payout/2 for a half kelly ramp.

However, when I sim it (CVD) I get a RoR of close to 5%, instead of the <2% expected (betting ramp includes wonging, so no penalty on low counts).

SO I am not quite sure what this means - do I infer that in this case, the RoR resulting from the simmed EV and variance is not applicable (the case of long shot side bets) ?

NB when I sim this, I am also including an equivalent bet on the blackjack hand. I had assumed that the additional variance of a regular blackjack hand would have little impact on the overall high variance - is that a correct assumption ?


Thanks

D.

I'm no help with sims. I have literally never used one.

I would agree that the BJ bet shouldn't have much effect on the variance, especially because with this sidebet, when there is an edge on the sidebet, there is most certainly also a big edge on the BJ bet.

As an aside, if there is a good sidebet opportunity, I generally like to spread with the count on the BJ bet as little as possible so as not to give away my AP-ness. On the other hand, with mediocre sidebets like Lucky Ladies, I usually just play my normal game and plop out the LL bet on the rare occasions when the count is high enough.

This is a good and rather simple enhancement to the suited sidebet, ala Royal Match.
It could also assuage the strike-point to identify opps to make the bet below the pure count-related strike point. zg