What's PE even mean? I know what it stands for, but what does a .6 mean? Higher is good, but does it mean 60% you're making the right deviation--and 40% u should use BS? If that's the case, in a shoe game, then how much BC should you give up to gain PE? If I can improve my PE by .05 and my IC by .05, should I give up .02 in BC?
In most modern games, betting efficiency is what matters. PE matters when you're using small spreads in pitch games because you're afraid of getting backed off. If you're getting 2/3 of your gain from PE, you're not spreading enough.
The Cash Cow.
Spreading 1-4 on an S17 DAS 70-75% DD game (the best game you'll commonly find), even with Hi Opt 2, is not that good of a game. I'm getting SCORES in the 50s, depending on number of indices used.Originally Posted by Tthree
Also, if you're playing those kinds of games in Vegas, they will bust your ass. Even with a 1-4 spread. I don't know any serious pros who play a lot of Vegas DD for serious stakes because of the heat.
Last edited by moo321; 02-04-2013 at 07:02 AM.
The Cash Cow.
Back in the the old days, I had the idea to write code to calculate PE.
First think of something simple, such as insurance correlation, or betting correlation. What you need there is a set of effects of removal (usually normalised
to one deck) for A through T, plus ev, as per Griffin's book. Now to get PE, you have to work your way up the tree of dealer cards and player cards.
So you need EORS for 16-T, then 15-T(where for linearity it was assumed BS would be used for 16-T. All the way up to doubles, das or no das, splitting etc..
Calculating these things exactly seemed impossible so I just used monte carlo methods pull one card out, play the hand, put it back, play the hand. Right up the tree. Also I needed my own simulator (which I had) to get the overall percentage contribution for each dealer/player combination.
Then to put the whole ship load together, you calculated the covariance for your count vs the EOR's for all the playing decisions, and then added them together
weighted my the matrix of the frequency that each play (from the simulator) and then normalised again to get the final playing correlation.
As a byproduct, I could plug the EOR's in as seperate counts into my simulator (ie perfect linear play) and just for comparison purposes a standard 6d/1.5 DAS DOA game, cant remember the spread, but Brh-1 gave N0=25000, the 'perfect' player had N0=15000.
PS:for GZ The crunching for the EORS used to take days on Pentium-1's 300MhZ - that was the time I dropped off my horny GF to go home and check things were running right. What was I thinking???
The Playing Efficiency (PE) of a system represents how well the system identifies the correct play for a given set of playing conditions (rules, number of decks, etc.) compared to how well a computer might do it while tracking all card values and using precise probability calculations. The computer might still lose, but it's making the correct decision based on what it knows. So a PE of 1.0 means the system does as well as a computer might ideally perform. A PE of 0.5 means it performs half as well.
I found the discussion in Arnold Snyder's "The Blackjack Formula" (1982) easy to understand.
Last edited by Intermediate; 02-04-2013 at 07:58 AM.
I didn't think BS would have a PE=0 because some of the BS plays might still be the same play that would be dictated knowing all the cards that had been played along with all the probabilities. That's just my intuition/guess. May have to go back to Griffen's work to get that answer.
In "The Blackjack Formula", Snyder refers to "Playing Advantage" (PA) which is "the percentage of profit which may be realized by altering play from Basic Strategy. Any time one is playing basic strategy, even if the count is being used for bet variation, PA=0." Is that what you are thinking about Tthree?
Well, lots of people stick to the shoes for this reason. Outside of Vegas, there may be some DD, but on average, Vegas DD is far too sweaty to play, especially rated. Most people that are in Griffin are there because they played DD rated in Vegas.
The Cash Cow.
It should be well noted that the notions of PE and BC were developed over 30 years ago when computers were really slow. And they were always meant to act as guidelines not as a law of nature. Statements like "I have added a 7 side count and my PE increased to 0.74" and "I side-counted the aces to increase my BC to 0.99" are absolutely ridiculous to say the least. The only safe-proof way to test the performance of a counting system for a given set of rules and a given penetration is to RUN A SIMULATION (using a good simulator) you can do 500 million hands in less than 10 minutes on a average computer of if you are too lazy to do that, check CVCX online or Modern Blackjack
Chance favors the prepared mind
Too right - if you relied on BC/PE/IC figures you would assume UstonSS would beat Halves with Brh-I coming a poor third. It was only by simulation that the damage of counting the '9' as a negative card on insurance and PE turned the tables on all that.
Ditto AOII vs HiOpt-II
Brh.
So have we decided what PE tells us in relation to BS? If my PE is less than .5, does that mean using basic strategy all the time is better? One could interpret it to mean: when the count indicates you should deviate from BS, less than 50% of the time that is true...so the closer to .5, the less valuable are indices. A .48 BC seems like it would perform worse than flat-betting w/perfect strategy, since you'd only be correctly identifying good situations 48% of the time, thus losing more money than you would by flat-betting. Or like someone said above, would BS be a 0 and does a PE of .5 mean it performs 50% better than BS?
I see a lot of discussion about pitch. I only play shoes. So for me, BC is paramount. But it would still be nice to know what PE is actually telling me. I still only know what it stands for and that higher is better. The other two stats are pretty straightforward.
Last edited by Boz; 02-04-2013 at 11:47 AM.
Those are in order of their BC. And both ushtonss and Halves have weak PE and IC figures, and both count the 9. This stuff I already know. In a multi-deck game, UshtonSS seems like a good choice if you can handle it. What new information did simulation give you that would lead someone like me to rely less on these numbers?
Posting Permissions
-*-
Reply With Quote
Bookmarks