New iOS app shows real-time EVs as you play

[Dieter]

So I'm pretty confident in the numbers produced by my algorithms.

That's great. I'm glad you're confident in the quality of your product. That's an important part of a sales pitch.

I've used my code to calculate all my basic strategy tables and they agree with the ones found on the Wizard of Odds web site, except for some rare, minor differences

The fact that there are differences at all tells me that my level of confidence in one of the charts has to be different than my level of confidence in the other chart. If your methodology was equivalent, your results would be equivalent.

The old numbers have been backed up by formal mathematical papers for over 50 years. If you're saying that the old numbers are wrong (and yours are right), I would need a similar type of proof to have confidence in it. If your numbers don't match the old numbers, one of the sets of numbers is wrong.


Good luck with your business venture.

[tomker]

A suggestion: Being APs, I think a lot of people here actually don't want to spend the $6 to give your app a spin. But, I think if you made it available to them through some sort of private code or way to allow them to try it for free, you would get more valuable feedback and more important input than you would ever dream of.

Don

Hi Don,

Thank you for the suggestion. This got buried in a post pages ago, but I did make an offer that the first 10 people from this forum to buy the $6 upgrade in my app could email me their iTunes receipt and I would happily PayPal them the full amount that they paid. So far nobody has taken me up on the offer but it still stands.

Unfortunately Apple doesn't have a way to gift in-app purchases to other people so this system of emailing receipts and PayPaling money is the best I can do.

[tomker]

... The old numbers have been backed up by formal mathematical papers for over 50 years. If you're saying that the old numbers are wrong (and yours are right), I would need a similar type of proof to have confidence in it. If your numbers don't match the old numbers, one of the sets of numbers is wrong. ...

I never said that the "old" numbers are wrong and I clearly explained that the differences in the strategy tables are due to different assumptions about a player's hand, i.e., it comes down to questions of card composition.

What if there are two ways to make a particular hard total and the correct decision is different for each way? What are you supposed to put in the table? Do you base it on the EV delta, i.e., which decision loses the least when applied in all possible situations? But what if one card composition is more likely than the other due to the dealer's up-card? How does that figure in? Do you do some sort of weighted average of the EV deltas?

I'm happy to discuss this but I wish I could do so from a less defensive position. You can easily ask me questions about this stuff without the implications that my work is wrong.

[Dieter]

What if there are two ways to make a particular hard total and the correct decision is different for each way? What are you supposed to put in the table? Do you base it on the EV delta, i.e., which decision loses the least when applied in all possible situations? But what if one card composition is more likely than the other due to the dealer's up-card? How does that figure in? Do you do some sort of weighted average of the EV deltas?

Well, I'm not the app designer, but I would guess there would be an option for composition dependent strategy vs total dependent strategy, where appropriate. There might be an option for count-dependent basic strategy deviations, too. Merging the deviations and composition dependent strategy might be useful.

[tomker]

Well, I'm not the app designer, but I would guess there would be an option for composition dependent strategy vs total dependent strategy, where appropriate. There might be an option for count-dependent basic strategy deviations, too. Merging the deviations and composition dependent strategy might be useful.

Sure, I'm just trying to explain the (again, rare) differences between the strategy tables that my app displays and the ones on the Wizard of Odds site.

My app has an option to display the strategy table (total-dependent) for the rules that the user specifies, so something has to be displayed there.

But if you're interested in composition dependent, the bottom line of the real-time odds display shows the EVs based on the cards in the current hand so the highest EV (which is underlined) shows the correct decision based on composition-dependent strategy.

[bigplayer]

If I poll 10 possible voters and 6 of them say they're gonna vote for Mitt Romney and 4 of them say they're going to vote for Barack Obama and the poll has an estimated accuracy plus or minus 20%, that 20% is the standard error (otherwise known as the margin of error). The smaller your sample the bigger the standard error. As your sample size increases your standard error will also decrease. Because the edges at blackjack operate in a fairly narrow range (usually 1% to 1.5%) it takes many millions of rounds before standard error begins to drop to a manageable level. What is the use of running a 500,000 round simulation that says your win rate is 1.1% if the standard error is +/- 0.3%. (Actual EV somewhere between 1.4% and 0.8% 68% of the time and between 1.7% and 0.5% 95% of the time). To get Standard Error down to a tiny amount so that you know your EV estimate is accurate you need very large sample sizes. With small samples you'll be getting fairly close to the same result most of the time, but maybe 1 time in 10 or 1 time in 20 you'll get a result that's way way off in left or right field which will be enough to make the information you're generating worthless.

When I run a blackjack simulation it's usually 400,000,000 rounds. That is why the original blackjack simulations (with much smaller samples) done in the 1960's took weeks and needed to be run on the fastest IBM Mainframes only available in large universities or government labs.

[tomker]

... With small samples you'll be getting fairly close to the same result most of the time, but maybe 1 time in 10 or 1 time in 20 you'll get a result that's way way off in left or right field which will be enough to make the information you're generating worthless.

I think a good analogy in this case would be with cell phone cameras vs. DSLRs.

Basically you're saying that you can't take a picture with an iPhone camera and expect it to be up to the quality standards of a cover photo on an issue of National Geographic, so the iPhone camera is worthless and you shouldn't even bother.

But that's not really true. The iPhone is small and convenient and you usually have it with you and you can use it to take perfectly good (if not professional level) photos.

Same with blackjack. If I was writing a book about blackjack, I would not do the calculations using my $6 cell phone app. But e.g. if I'm at a casino and I'm curious about the EVs for some hand, I think it's perfectly useful to be able to pull out my phone, type in some numbers and within a couple of seconds see EVs that are accurate to ~1% and usually better.

I was at a casino a few days ago and a friend had a pair of 2s vs. a 2, and he asked me what the advantage was of splitting vs. hitting. Within about 20 seconds I was able to pull my phone out, unlock it, type the hand in my app, and tell him that it's about a 4% advantage. I think this is fun information to be able to calculate and it'd be worth $6 to me if I didn't make the app myself.

Maybe you are such an expert at blackjack that you know the EVs of every decision for every hand for every set of rules for every count, but I have to believe most human beings don't have those thousands of EVs memorized.

[Norm]

I think a good analogy in this case would be with cell phone cameras vs. DSLRs.

Sorry, but that is a poor analogy. Cell phone cameras can take pics that are extremely clear. Blackjack play has a very slim advantage. Being off a tad is really bad news.

if I'm at a casino and I'm curious about the EVs for some hand, I think it's perfectly useful to be able to pull out my phone, type in some numbers and within a couple of seconds see EVs that are accurate to ~1% and usually better.

Hope you carry enough money for bail.

[Norm]

I apologize if the last comment sounded snide. But, using a device in a casino in Nevada has a penalty of up to five years imprisonment. If you wish to provide tools to players, you need to investigate the field a bit better first.

Maybe I should rename this site "Read before you Write."

[tomker]

Sorry, but that is a poor analogy. Cell phone cameras can take pics that are extremely clear. Blackjack play has a very slim advantage. Being off a tad is really bad news.

I think you're overstating how bad it is for an EV to be off by 1%.

For the majority of decisions you encounter when playing blackjack, the EV difference between the best and 2nd best decision is going to be relatively huge compared to 1%... so a 1% error is only going to make a 1% difference in a small fraction of cases. So that's 1% times a small percentage. Then, the inaccuracy may be in the direction that causes you to make the correct decision, not just incorrect. So that's 1% * small number * half.

It would be interesting to run a simulation where decisions are made based on EVs calculated using 100k iterations vs. millions of iterations. I would be pretty startled if the difference was over 0.01%.

[tomker]

I apologize if the last comment sounded snide. But, using a device in a casino in Nevada has a penalty of up to five years imprisonment. If you wish to provide tools to players, you need to investigate the field a bit better first.

Maybe I should rename this site "Read before you Write."

You didn't sound snide in your first post but you sure do now. "Read before you write" is basically the definition of snide.

In fact, I can very obviously provide a tool to players without better investigating the field, since I have clearly already done so.

I thought you were going to stop "contributing" to this thread?

[tomker]

I apologize if the last comment sounded snide. But, using a device in a casino in Nevada has a penalty of up to five years imprisonment. If you wish to provide tools to players, you need to investigate the field a bit better first.

Maybe I should rename this site "Read before you Write."

BTW... looked up NRS 465.075.

I think it's pretty obvious that the law is referring to calculating probabilities for a specific instance of blackjack and not blackjack in general.

Otherwise, it would be illegal to refer to one of those little plastic cards with basic strategy printed on it, and people do that all the time.

[ADMIN EDIT: The Nevada device law specifically outlaws, with a prison penalty, the use of devices in a casino that calculate odds. This does NOT apply to plastic cards.]

[Norm]

You didn't sound snide in your first post but you sure do now. "Read before you write" is basically the definition of snide.

Quite right. And, I don't apologize for that one as members here have to constantly deal with new members that post without bothering to read first. We try to be helpful. But, everyone's time is limited..

In fact, I can very obviously provide a tool to players without better investigating the field, since I have clearly already done so.

No, you have provided a tool that you suggested was the first of its kind, when it wasn't, and suggested it was useful, when I really feel otherwise. And, you suggested that it would be useful in a manner that could put you behind bars -- as has been very heavily pointed out in the news after another app was released (and then removed by Apple).

I thought you were going to stop "contributing" to this thread?

I gave a 92% chance that you would respnd with that exact statement. And the scare quotes suggest that my post was useless. Look, when someone advises that people commit felonies, I have a duty to respond.