Bet Spreads

Hi, I am brand new to the forum and it looks like the forums contain a wealth of information (i have spent a few hours browsing various threads). It all looks very excellent!!

I have a small question regarding bet variation.

Up to this point I have been playing a 1-12 spread i.e. 1,2,4,8,10,12 for TC +1,+2,+3,+4,+5,+6. I have an (arbitrary) bankroll of 600 units (taken from business profits) plus 251 units of winnings = 851.

I am thinking of changing to a 1-10 spread in the order of 1,3,6,10,10,10 - doing some calculations (using results from simulations from bjstats.com) in xls, seems that my expected unit return per hour would be about the same, but my gut feel is that the change in spread would adversely impact on my variance and thus RoR?

Background on my game:
I have been playing for just over six months:
4D or 6D, S17, DAS No surrender 70-80% pen
count using Hi-Low Method with I18 indeces
Hours played: 223 (using either $5-$60 or ($5)$10-$120 standard unit)
Status: +251 units (1.13 units/hour)
Hands played: approx 15,000 avg 70 hands/hr (often engage in heads-on play)

Would welcome comments that anyone may have, as to whether it would be smart to do this or not.

Matt

Wait a few more months...

I would wait until I had at least 100 4 hour sessions prior to changing betting strategy.

Your spread 1-10 in 4-6D games is only adequate if you avoind negative counts, in which case 1-8 will suffice.
If you do not avoid the negEV counts you will need 1-20+ spread. zg

Thank you for your answer on this zengrifter - also I very much enjoyed reading your interview on this site.

Can I confirm a couple of items re your EV statement-

- negEV refers to any situation where the casino has an advantage - i.e. the standard advantage driven by the casino's rules +/- change in advantage due to the true count?
- Avoiding negEV- It would be difficult for me to wong into shoes where i play - the casinos are relatively quiet (at the times i go) - hence i typically play from the beginning of the shoe and exit if TC is worse than -2. However, i play with only 0.5 units if the TC is below +1 - does this approach effectively turn my 1-12 spread into a 1-24 spread?

Thanks in advance,
Matt

thanks for your reply!

how many hours (on a weekly basis) do serious counter normally go for? I aim for approx 20 a week unless i am very busy with non-B.J. projects.

I'd say yes.

If so, does that mean your original 600 unit roll was actually a 1200 unit roll?

I can't really help you that much with your question. Aside that, to me, stuff like "I have been playing for just over six months:4D or 6D, S17, DAS No surrender 70-80% pen" is already 4 or more different games, lol. All I can suggest is if you change to 1-10 (or maybe 1-20 lol), maybe it's possible your original dollar unit might change to keep your ROR the same. Maybe not. Maybe your unit variance would go down but, in dollars, go up, maybe not, etc. Maybe you'll choose to bet to your new 851 unit roll or maybe make it comparable to your original 600 unit roll but with a different dollar unit obviously.

You mention winning 1.13 units per hour, what is the variance of that - like how would you know if a different spread increases it or decreases it anyway?
Likewise with ROR. Like I guess the 1.13 is apparently actual but what was expected over your 15000 hands?

I guess if you have a spreadsheet that you are using, and if it has the basics of freq, adv, SD at each TC, I could try to help with the ROR and variance stuff for different spreads for whatever game and assumptions that spsheet is portraying.

But one spsheet can't portray a 4D 70% game and an 80% 6D game or vice-versa at the same time kind of thing. You can assume they are all "equal" if you want. And, like you say, any differences may not be great enough to bother you. That's up to you to decide.

Hi Kasi, thanks for your reply! Apologies for not being more specific - re-reading my post I realise how vague i was in places!

I play at a number of casinos, some have 4D and some are 6D - the penetration varies with each dealer, but on the whole I am looking at pen scenarios of 75 & 83. I need to do some more learning regarding SD, ROR and Variance before discussing further (but will then come back to you!). Although I aced this topics at University, i now remember little about them.

But yes i have already put some spreadsheets together at this point that provide for freq, adv, expectancy at each TC to work out an expected rate of return - and i have created variables for number of decks, pen, hands/hour (also linked to number of players/boxes and number of decks). However these do not provide for SD, VAR or ROR.

I made a similar spreadsheet if you're interested:

http://www.blackjackinfo.com/bb/showthread.php?t=2845

It calculates approximate EV, SD and RoR for any bet spread you enter.

-Sonny-

hey this is great - thank you!

Please don't apologize - no need for that. And I apologize for making you feel that you needed to apologize lol.

I just meant, I think lol, that one still needs a sim to at least generate those freq, adv and SD's at each TC for whatever assumptions you want - like whether you are using indexes, what counting system, pen level, game rules, etc. I don't think a spreadsheet can really do that. But once you have them, it can calculate pretty much whatever for changing spreads, like Sonny's sheet does.

But, while it can do that, I still struggle with optimal spreads and how one knows when to place max bet or intermediate bets on the way to a 1-X spread. Sims are good for that too.

Like, in Sonny's sheet, if I downloaded the right one, what if he bet $80 at TC+4 instead of $60 in the first tab? Like I get N0 would go down so, even though ROR would go up to 7.3%, so would that make it a "better", maybe I mean more optimal, spread?

Also, not that it matters, I don't understand what the 0.62% in cell B15 means. Like I get all his numbers but I get an EV of 0.868%.

Don't use cell B15 for EV, use cell B20. You can ignore cell B15 completely.

-Sonny-

so is that the cell i was asking about way back when? i forget.
so does that formula need to be in there? duuh maybe the value is used for something else in the sheet.
well what ever that's supposed to be the average ev of the tc's? but the one tab has it where all the tc's ev's aren't counted up. or something. the other tabs do add em all up.
but i'd think Kasi'd be interested in that sheet as it addresss's multiple hands.

going further with this topic...

one thing that puzzles me is the distribution of frequency across the different true counts. I understand that these are available in one of Wong's books, for now i have been running simulations using CVD with 500million rounds and then look at the True Count distribution table (for 4D/6D & 75/83 pen). I am then inserting the frequencies from this table into Sonny's spreadsheet.

I would have thought that a +1 TC occurs as often as a -1 TC, and a +2 TC as often as a -2 TC and so on. However it seems that in each instance the negative TC occurs slightly more often. What is the likley reason for this? Is it to do with rounding the TC downwards?

Matt

Yeah, that's the same cell you pointed out earlier in this thread. To be honest, it doesn't do anything and doesn't even need to be there. That spreadsheet is from 3 years ago and I don't remember why I put it here.

-Sonny-

i spent the day further working on putting together a probability model using very valuable information from this forum and from CVD and CVBJ. I think I'm definitely making some progress and I can certainly answer my own questions that I posed at the start of this thread. It was interesting to spend 2-3 days at home assessing my game rather than counting for 5 hours at the casino each day :grin:

The analysis helped me understand how various betting spreads (including multiple hands) impact on my EV, SD and ROR. I realise that in my workings I am continously 'underbetting' my intended bets as compared to the Kelly optimal wager. I have been doing this in order to minimize the ROR. The bet spreads that I have put together for 4D/6D 75/83pen combinations, produce lifetime ROR between 5 and 11% (bankroll=850 units).

Can someone comment what a typical target ROR equates to? Is 5-11% much lower than counters use?

The other point i have noticed is the large impact on overall EV from the '+7' and '+8 and above' true count categories. Advantage of 3-3.5% with 20-25 units spread over 2-3 hands, occuring 1-1.5% of the time, makes up a whopping half of the overall EV. Am I understanding this correctly?

I just wish I knew what cells I could ignore in my sheets but I just know there are alot of them lol.

While I'm here, any particular reason you used a variance of 1 in cell D14? I realize they are all estimates anyway and it doesn't make much difference. I was too embarrassed to ask the other nite since it's so meaningless.

I was talking about win rate as a percent lol. So I got win of 0.0155 units per hand/1.787 avg bet units per hand=0.00867 avg advantage for the player.

Like it looks like, I assumed anyway, your frequencies and advantages must have come from a sim but then I wondered why you just estimated the variance since a sim would have given you that too? Or maybe it's just intended as an instructional example. And it's a good one!

You should change your name to sageelephant since you forget nothing

I realize Sonny used estimates for the covariance when playing 2 hands.

My main question on that is how does one go about calculating covariance from variance? Is there a way? Every once in a while I run across something I don't really understand but makes me think that there is a way

Or do sims (generic term for CVCX or CVDATA lol) just calculate that for you and include it in the total variance if you specify multiple hands?

Even if they do, how does the sim figure it out?

I loved the fact Katrina gave us the covariance at each TC in her book.

I have no idea lol but it seems to. Maybe because the game is a -EV overall so it stands to reason you spend more time in neg counts?

But, you're right, the frequencies and advantages will change depending whether you tell the sim you calculate True Counts by rounding, flooring or truncating. I guess telling it whether you estimate your TC by whole decks remaining, or nearest half deck, or even exact cards remaining could change those frequencies and advantages a little too.

I guess the corollary to that is that index numbers might also change so you have to know which assumption was used for the index numbers you are learning.

Like Wong, compared to a long time ago, increased all his neg index numbers by 1 when he chose to change the assumption on how he arrived at TC's.

Like, if you truncate, you have a huge range from -0.999 to +0.999 where the TC would be zero.

Well now you're getting into areas i don't fully understand but I think it's a good thing you do not bet full Kelly or "optimally" to maximize the growth of your bankroll. While the lifetime 13.5% ROR associated with it may not sound so bad, what's not immediately obvious, from the way I understand it anyway, you have very large chances of losing large portions of your bankroll.
I get the feeling playing at full-Kelly is maybe not generally recommended.

Like it's well-known a full-Kelly better will double his roll 2/3 of the time before halving it. A 1/3 Kelly better will double his roll 97% of the time before halving it. But all that may involve alot of unrealistic assumptions in the first place.

I have no idea what ROR real counters play at - each to his own I guess.
You can always maybe re-size bets to roll at any point in time to manipulate the ROR from that point forward.

Anyway, the main thing to me in my ivory tower lol, is you know what to expect from your roll, game, betting plan etc before you play, maybe know if your unit size changes if suddenly you get a good pen, etc and, then, above all else, when you get back home that day, you have some reasonable way to assess what happened.

This is easier to do if you bet according to a plan you developed ahead of time. It requires discipline to stick to a plan so later you can measure the reuslts.

oh yeah - if you are using Sonny's sheet and spreading to 3 or more hands don't forget to change the variance accordingly - I think his 1.83 estimate would only apply to spreading to 2 hands.

Anyway, I think you are well on your way to all this.

Did you say somewhere you have CVDATA or did I misunderstand? I don't have it but wouldn't that give you ROR and EV and SD all those things you originally said you couldn't figure out? Or is it only CVCX that gives that?

Hey, if you get incredibly bored, check out the Weekend Warrior thread, play a theoretical game with low ROR, a large spread, and measure your results against that sim if you want. Somewhere in there is a sheet similar to Sonny's. Using CVBJ or Ken's game here or at card-counter.com.

Same as Sonny's sheet

It's just the same thing as Sonny's sheet, I hope lol, but calcs a few more things like avg bet units, some trip roll stuff, goal reach stuff with no time limit, N0, SD ranges, whatever, in case anyone finds it relevant.

You can change the red cells after you input whatever in the freq, adv, SD columns. The spread is units - change roll dollars and roll $unit below in it.

What this version doesn't do is spreading to multiple hands and might not work for anything other than "play-all" scenarios.

Oh yeah - this one uses SD for input whereas Sonny's uses variance.