Confusion over index plays: CVBJ vs Gamemaster

#1
I must admit that I'm a bit confused over the index plays given by Gamemaster in his blackjack school.

The plays I'm using are those from CVBJ: Wong Complete High-Low. Yet while some of these are the same as those given by Gamemaster, many differ - by a considerable amount.

Take A2 vs 4: Wong gives +3 but GM gives +7 (which incidentally is Wongs index play for A2 vs 3).

Again, looking at 9,9 vs 7, Wong gives +3 whereas GM gives +6.

There are many other examples which I won't list.

Can anyone comment on this seeming discrepancy? Are the Wong indexes "old"? Luckily the I18's seem to match, but I am moving past the I18 and learning new ones.

I must admit that I've discounted the GM ones purely because in one part of his course he says that counts > 6 hardly occur on average yet he bombards the newbie with index playes of +7 to +10, which is contradictory in my mind.

Thanks in advance
 

Meistro

Well-Known Member
#2
Also what about 9 vs 7? Some places say 3, some places say 6. I assume the difference is either in how the index plays are generated (rounded or truncated?) or some are risk adverse some are expectation maximization.
 

UK-21

Well-Known Member
#4
When I was reading through the GM online course (a while back now) I remember reading something to the effect that the index plays included were very risk adverse, and had been moved up a click or two on the computer generated ones published in other books. I suppose one should view the differences not so much with a view that they are wrong, more with an eye on your appetite for risk.

Assume those published in Wong's PBJ reflect the correct mathematical decisions, and the GM ones as adjusted for risk - at the end of the day, doubling on A,2v4 is a pretty high variance play, and the win expectation is based more on the dealer busting than A,2 being turned into a winning hand (19+?) combined with a relatively large window of opportunity for the dealer to win - think of all of those uncounted 7,8,9s that could put an appearance in to spoil the day. At +7, there's a much higher probability of the dealer busting out.
 

UK-21

Well-Known Member
#6
Several of the regulars here have and/or use them.

I did find an article somewhere (written by Grosjean?) with an explanation of applying RA indexes, and how to calculate them. Great if you're a maths head and want to get that involved.

Personally I've just moved some of mine up a click (or two) when compared to those that result from the pure mathematical expectation - mainly those plays that depend on the dealer busting out to provide a win rather than those that rely on drawing a card to turn a winning hand.

I use Hi-Lo, and the "risk" factor always seems to be heavily led by the unknown number of 7,8,9s still available to be played - against a dealer 5 or 6 any of those will give a strong possibility of leaving the dealer with a losing hand at higher counts. With a dealer 4 a 7 turns the dealer's hand into a potential 19,20,21, a dealer 3 can be made the same with a 7 or an 8 etc etc.

One of the reasons I personally don't double max bets against a dealer 2 or 3 despite the fact tht the mathematical expectation of doing so is greater than just hitting (over the longer term of course). The call all depends on how much of the folding stuff you have and your personal appetite for risk of course. For somone who is well capitalised, the issue isn't really an issue.

The other approach is to use a (level 2?) count where the 7,8,9s are all taken into account and assigned a value - so the 7,8,9 FUD factor is allowed for within the maths underpinning the index plays. They are of course in Hi-Lo generated indices, but I would think the variance in the results of playing hands against dealers' 2,3,4s would be greater if comparing two like samples?
 

ycming

Well-Known Member
#7
There is a small section of RA indicies in BJA.

Personally I am not in favour of RA, if I want a lower risk factor, i would move my unit size down instead of not capitalizing the EV!

BS was calculated based on maximising EV .... may be people will start asking for a RA BS? What does that happen if the BS changes ? House edge increases ......

Remember we work on such small advantage make good use of every little one! Hence follow wong's indicies IMO :)

Ming
 

QFIT

Well-Known Member
#8
ycming said:
There is a small section of RA indicies in BJA.

Personally I am not in favour of RA, if I want a lower risk factor, i would move my unit size down instead of not capitalizing the EV!

BS was calculated based on maximising EV .... may be people will start asking for a RA BS? What does that happen if the BS changes ? House edge increases ......

Remember we work on such small advantage make good use of every little one! Hence follow wong's indicies IMO :)

Ming
This is a common misconception. Correctly used, RA indexes increase win rate, not decrease. However, I would ignore the indexes in Gamemaster.
 

assume_R

Well-Known Member
#9
QFIT said:
This is a common misconception. Correctly used, RA indexes increase win rate, not decrease.
Why exactly is this the case? I was under the impression RA indices slightly decrease EV but the good thing is that they decrease the standard deviation.

Then wouldn't win rate decrease slightly if EV was decreased? (obviously the benefit would be a lower RoR at the cost of a slightly decreased win rate)
 

assume_R

Well-Known Member
#13
So would that mean that for RA indices to yield a higher win rate, one could increase his bet without increasing his RoR? Or did I misinterpret the information from your book slightly?
 

ycming

Well-Known Member
#14
A bit confused:

"Older strategies generally used expectation-maximizing indexes. (There are exceptions.)!"

In terms of mathematics, if the older strategies is calcualted with the aim of maximising expectation, then you would gained the most EV?

"It is a common misconception that risk-averse indexes are used to reduce overall risk at the cost of lower results. In fact, RA indexes reduce the risk of specific decisions, which reduces the variance, which allows you to slightly increase your overall betting levels. This provides a slight overall improvement in results."

How would you define slight ? increase winrate by $1 / hour or? Is the marginal increases worth the change of strategy ?

Would there exist a RA BS? As the BS is calculated with the aim of maximising expectation also. e.g. changes in softhand double?

Ming
 

QFIT

Well-Known Member
#15
EV-maximizing strategies provide the highest EV. But, not the highest SCORE or win rate. Since RA indexes reduce variance, risk of ruin is reduced. So, with RA indexes, you can bet very slightly more than with EV-max indexes, and yet have the same risk of ruin. This improves the win rate. The improvement is minor -- 2% give or take depending on circumstances.

The effect isn't large enough to effect BS.
 

QFIT

Well-Known Member
#17
psyduck said:
Aren't EV and win rate the same thing?
No, win rate depends on the unit size. EV (defined as a percentage) ignores unit size. RA indexes allow a slightly greater unit size with no additional risk.
 

KenSmith

Administrator
Staff member
#18
I have to admit that while preparing the School lessons for posting here, I didn't really pay any attention at all to the index numbers being presented.

If someone wants to verify them and contact the gamemaster on the ones that appear to need correcting, I'd love to do so. Since it's his work, I won't make any changes without going through him, whether obvious or not.
 

ycming

Well-Known Member
#20
QFIT said:
EV-maximizing strategies provide the highest EV. But, not the highest SCORE or win rate. Since RA indexes reduce variance, risk of ruin is reduced. So, with RA indexes, you can bet very slightly more than with EV-max indexes, and yet have the same risk of ruin. This improves the win rate. The improvement is minor -- 2% give or take depending on circumstances.

The effect isn't large enough to effect BS.
So if my win rate is £10/hour, using RA indicies i will gain £10.20/hour ?

Thanks
Ming
 
Top