bet spread for baccarat dragon 7 counting?

Hello,

I enjoy counting the dragon 7, I think its fun because it gives me something to concentrate on, and I like baccarat because the people who are serious about predicting the player/banker streaks are so serious but yet so wrong.

I was wondering what would happen if instead of flat betting, what would happen if I counted with an unbalanced count (http://discountgambling.net/category/baccarat/) and then instead of flat betting I would bet 1 unit per count above zero (so if RC is 5, i would bet 5 units, if RC is 12 I bet 12, etc) my average bet would be 5 units so my units would have to be re-sized, but if I did this would it lower my variance at all? how about my EV?

I dont know how to program or how I would run the simulation myself, has anyone else thought of this? is it a good idea?

Quote: Bondy3

Hello,

I enjoy counting the dragon 7, I think its fun because it gives me something to concentrate on, and I like baccarat because the people who are serious about predicting the player/banker streaks are so serious but yet so wrong.

I was wondering what would happen if instead of flat betting, what would happen if I counted with an unbalanced count (http://discountgambling.net/category/baccarat/) and then instead of flat betting I would bet 1 unit per count above zero (so if RC is 5, i would bet 5 units, if RC is 12 I bet 12, etc) my average bet would be 5 units so my units would have to be re-sized, but if I did this would it lower my variance at all? how about my EV?

I dont know how to program or how I would run the simulation myself, has anyone else thought of this? is it a good idea?

I don't think that this would help that much. You are going to hit it pretty rarely no matter what... even with a slightly higher edge, it is still unlikely to hit. So, your variance is always going to be sky-high. I'd just flat-bet a fixed amount whenever I had an edge.

If you can find a baccarat table that charges commission (ie, regular baccarat, not EZ) and has the "regular" dragon side bet (not the dragon 7), I'd suggest playing that instead. Your edge is a little lower but the variance is a lot lower, so you can bet more with that lower edge (and thus make more money). Note that the count is different for that bet; you can find it on Stephen How's page as well.

Quote: AxiomOfChoice

I don't think that this would help that much. You are going to hit it pretty rarely no matter what... even with a slightly higher edge, it is still unlikely to hit. So, your variance is always going to be sky-high. I'd just flat-bet a fixed amount whenever I had an edge.

If you can find a baccarat table that charges commission (ie, regular baccarat, not EZ) and has the "regular" dragon side bet (not the dragon 7), I'd suggest playing that instead. Your edge is a little lower but the variance is a lot lower, so you can bet more with that lower edge (and thus make more money). Note that the count is different for that bet; you can find it on Stephen How's page as well.

a little lower? the edge will be 5 times lower.. like instead of winning 1 unit every other shoe I will win 1 unit every 10 shoes :(

I understand the variance being sky high on the dragon 7 side bet (you must play 625 shoes to have a 90% chance of being ahead if flat betting) I was wondering how many shoes you have to play to be 90% chance of being ahead of you spread like I mentioned in the first post

Quote: Bondy3

a little lower? the edge will be 5 times lower.. like instead of winning 1 unit every other shoe I will win 1 unit every 10 shoes :(

I understand the variance being sky high on the dragon 7 side bet (you must play 625 shoes to have a 90% chance of being ahead if flat betting) I was wondering how many shoes you have to play to be 90% chance of being ahead of you spread like I mentioned in the first post

First, the edge is about 1/9 of a bet per shoe, not 1/10 (11.2% according to Stephen How).

Second, and more importantly, you should not be concerned solely with edge. You should be concerned with edge / variance, since that determines how much money you win if you are betting according to Kelly (which you should be).

Variance of dragon 7 is about 40. Variance of player dragon is around 6.1. So, while the edge is 4.5x as high in the dragon 7, the variance is 6.55x as high, so you make less money. Your units in player dragon should be about 6.55x as big as they are in dragon 7, so you are happy to win only 2/9 as many units. For example, if your bankroll is big enough to bet $100 on profitable dragon 7 hands (earning $52/shoe), that same bankroll is big enough to bet $655 on each profitable player dragon situation, earning $73.36/shoe.

The increase is significant; over 40% more money with the same bankroll.

Quote: AxiomOfChoice

First, the edge is about 1/9 of a bet per shoe, not 1/10 (11.2% according to Stephen How).

Second, and more importantly, you should not be concerned solely with edge. You should be concerned with edge / variance, since that determines how much money you win if you are betting according to Kelly (which you should be).

Variance of dragon 7 is about 40. Variance of player dragon is around 6.1. So, while the edge is 4.5x as high in the dragon 7, the variance is 6.55x as high, so you make less money. Your units in player dragon should be about 6.55x as big as they are in dragon 7, so you are happy to win only 2/9 as many units. For example, if your bankroll is big enough to bet $100 on profitable dragon 7 hands (earning $52/shoe), that same bankroll is big enough to bet $655 on each profitable player dragon situation, earning $73.36/shoe.

The increase is significant; over 40% more money with the same bankroll.

I understand what you are saying, but my question isnt about which bet is best, (I can easily switch to the other dragon side bet) the question I have is about the amount I should bet, Should I stick to flat betting or should I ramp it up based on the Count? if I ramp it based on the count would there be a noticeable decrease in variance?

Quote: Bondy3

I understand what you are saying, but my question isnt about which bet is best, (I can easily switch to the other dragon side bet) the question I have is about the amount I should bet, Should I stick to flat betting or should I ramp it up based on the Count? if I ramp it based on the count would there be a noticeable decrease in variance?

You should definitely ramp it up based on the count, if you can do so easily. Doesn't Stephen How graph the count vs edge somewhere? If not you will have to simulate it yourself. Bet BR * edge / variance each hand. To make it simple, I'd just use 40 for variance (variance actually moves with the count as well but you probably want to keep things simple -- I'd just always use 40)

This also assumes that you can get away without making any bets on the main game. If you can't, things get more complicated.

My point was that if the other game is available, your gain from switching will far outweigh any gain you get from changing your betting structure.

Quote: AxiomOfChoice

You should definitely ramp it up based on the count, if you can do so easily. Doesn't Stephen How graph the count vs edge somewhere? If not you will have to simulate it yourself. Bet BR * edge / variance each hand. To make it simple, I'd just use 40 for variance (variance actually moves with the count as well but you probably want to keep things simple -- I'd just always use 40)

This also assumes that you can get away without making any bets on the main game. If you can't, things get more complicated.

My point was that if the other game is available, your gain from switching will far outweigh any gain you get from changing your betting structure.

I understand what you meant by the gain from switching outweighs the gain from the chain in betting structure.

He doesn't show EV per count, but he does show EV vs frequency for both bets
http://stephenhow.files.wordpress.com/2011/10/evdragon1.gif

when looking at the shoe simulator I see the count sometimes get as high as +20 (rarely)
http://imadegen.com/dragon7ub/

I would model it myself but I dont know how to code :(

Quote: Bondy3

I understand what you meant by the gain from switching outweighs the gain from the chain in betting structure.

He doesn't show EV per count, but he does show EV vs frequency for both bets
http://stephenhow.files.wordpress.com/2011/10/evdragon1.gif

when looking at the shoe simulator I see the count sometimes get as high as +20 (rarely)
http://imadegen.com/dragon7ub/

I would model it myself but I dont know how to code :(

You're unlikely to get someone to do it for free. I could do it but it seems like a lot of work for not much gain. Maybe ask Stephen? If he still has his data lying around, he might already have this information, but never have posted it. Also Eliot Jacobson sometimes does analysis like this; I'm not sure if he'd be interested or not.

You could also look at the EOR numbers that he posted and try to estimate something. This is probably easier with a balanced count and true-count conversion, but his unbalanced count is pretty simple from what I remember (it just assumes an average number of cards per hand; it's easy to convert to a balanced count)

Finally, you might be able to look at the information available to you (distribution of counts, break-even point, and average EV for bets above the break-even point) and figure something out. Again, you will need to do a true-count conversion, but from there you can probably assume an almost linear relationship between count and edge, and then just fit the data so that the average EV is correct.

It sounds like it is more work than it's worth, but I might be wrong about that.

Quote: AxiomOfChoice

You're unlikely to get someone to do it for free. I could do it but it seems like a lot of work for not much gain. Maybe ask Stephen? If he still has his data lying around, he might already have this information, but never have posted it. Also Eliot Jacobson sometimes does analysis like this; I'm not sure if he'd be interested or not.

You could also look at the EOR numbers that he posted and try to estimate something. This is probably easier with a balanced count and true-count conversion, but his unbalanced count is pretty simple from what I remember (it just assumes an average number of cards per hand; it's easy to convert to a balanced count)

Finally, you might be able to look at the information available to you (distribution of counts, break-even point, and average EV for bets above the break-even point) and figure something out. Again, you will need to do a true-count conversion, but from there you can probably assume an almost linear relationship between count and edge, and then just fit the data so that the average EV is correct.

It sounds like it is more work than it's worth, but I might be wrong about that.

I looked at the EOR on the wizard of odds, as far as I can tell from using the unbalanced count for every RC I see it amounts to about 0.2% change in house edge,