Long-run EV

[matt21]

hi everyone, hope all is well with everyone's counting!

i have been looking over my completed sessions and have also run many of my actual sessions through simulations on CV Data.

Actual Performance results:
my historical EV is currently running at +1.68 units/hour - this is based on 382 hours of live casino play (no wonging) and approx 35,000 blackjack hands (i have made some estimates of number of hands played as the hourly rate has changed over the year based on the playing conditions that i have encountered.)

Simulation results:
I have run simulations for the various playing conditions that I ahve encountered (penetration, game rules, # of players, #hands per hour, betting spread). This has resulted in a range of EV's. I have tried to play as many sessions as possible in high-EV playing conditions and minimise play in low-EV conditions. The low-EV conditions extend down to +0.5 units/hr and the high-EV conditions exceed +2.0 units/hr.

Thus both practice and theory seem to suggest that my long-run EV (for good conditions) is somewhere between 1.5 units and 2.0 units per hour.

Do these EV figures sound correct or am I off the path? Are others players encountering similar results in their simulations and/or their live casino play?

Would love to hear comments from others

 

[johndoe]

Sounds pretty reasonable for favorable games. Tough to get that much without wonging at more common games with less pen, for the most part.

 

[kewljason]

results

I've played 1146 hours so far this year. Up 1776 units, or about 1.55 units per hour, which is similar to your results. Not exactly sure of my hands per hour. I'm sure its well below 100. ( would guess 70 ish, as unfortunately I rarely get to play heads up with the dealer at my playing level and more often than not am playing with 3 or 4 other players.

 

[Kasi]

Well, I'll just throw stuff out there for the fun of it. Good possibility it's crap anyway lol.

So Matt has maybe made 1.68*382=642 units. Over 35000 hands or .018 units per hand.

Maybe Kewlj has played 80220 hands and won 1776 units or .0221 units per hand.

So maybe they are comparable per hr, which is fine even though Kewlj is making 20% more per hand.

And of course what's the point of comparing how one guy plays to another anyway? Might be totally different games, spreads etc. lol.

And, then, I think maybe Matt spreads to multiple hands alot? Assuming so, does/should he count that as 2 hands or 1 round? Just asking.

How will Matt know when his EV=1 SD after playing alot of different games but maybe only knows I won on avg so many units in so many hours?

I guess you could weight it all together somehow, 10 hours at a bad game, 30 hours at a good game, 20 more at the same game but heads up, 40 at a good but not quite as good game, 30 more same game but difeerent spread, etc.

One game at a time is all one can do - lump similar stuff together if you want maybe. If I did this cr*p, i'd go with hnads - guess i probably really mean rounds? Anyone? Anyway, I'd just try to keep apples to apples as much as you can.

I never tried to avg all my internet crap together. What would it mean anyway? 1D Crypto was different than 6D Crypto was different than 1D Micro was different than etc. So, if I played a different game, it got it's own space lol. I'd rather know if one software/game was getting way out there on the curve on the wrong side while another game/software was getting way out there on the good side than avg them together so everything looked peachy keen.

So now I can say dumb stuff like I made 40 cents a hand. But I'd much rather know I made $1.95/hd at one joint compared to 6 cents a hand at another joint.

Long way of saying, forget the past use CVDATA, ither before or after each session going forward for each different game.

What kind of log do you keep? Just curious - answer optional of course.

 

[kewljason]

I wasn't trying to "compare" my results with matt21. I was simply sharing my experiences and results in response to his question, "Do these EV figures sound correct or am I off the beaten path?", as per his request.

As for different results for different types of games, I personally have no need to break down my results that way. 95% of my play takes play between 3 casinos in atlantic city which all have exactly the same rules. 6 or 8 decks, stand soft 17, no surrender. The only difference is penetration, and I wont play less than 75% and can rarely find more than 80-83%, so even that is similar.

 

[matt21]

I've played 1146 hours so far this year. Up 1776 units, or about 1.55 units per hour, which is similar to your results. Not exactly sure of my hands per hour. I'm sure its well below 100. ( would guess 70 ish, as unfortunately I rarely get to play heads up with the dealer at my playing level and more often than not am playing with 3 or 4 other players.

hi kewljason, thanks for posting your actual results. it's very helpful for me to have someone else's results on hand for reference. And it seems we are in a very similar range in terms of our resuults. Although if you are doing 70 hands/hour then i think 1.55 is quite a performance!

Thanks again for posting these

 

[matt21]

Hi Kasi - thanks for your reply and your questions - always very appreciated - here's my thoughts

And of course what's the point of comparing how one guy plays to another anyway? Might be totally different games, spreads etc. lol.

i was just trying to get an approximate comparison - yes you are right we are playing different games rules, playing conditions and betting spreads so we would not expect to have exactly the same result even before allowing for natural standard deviation

And, then, I think maybe Matt spreads to multiple hands alot? Assuming so, does/should he count that as 2 hands or 1 round? Just asking.

Yes I spread to 2 hands for +2, +3, +4, and +5 and to 3 hands for +6, +7 and +8 (or higher). CV data allows for this in the analysis - you can incorporate multi-hand betting strategies in the simulations which is great. However this is one of my problems. In my own probability model i count a multi-hand round as one hand rather than several hands. My figure of 35000 hands may actually be more than that.

How will Matt know when his EV=1 SD after playing alot of different games but maybe only knows I won on avg so many units in so many hours?

I guess you could weight it all together somehow, 10 hours at a bad game, 30 hours at a good game, 20 more at the same game but heads up, 40 at a good but not quite as good game, 30 more same game but difeerent spread, etc.

re EV=1 SD -> are you referring to N0?

I individually track the EV for each session so that I know how good or bad the playing conditions are/were - and to then help me to repeat the good ones and avoid the bad ones. But I dont actually add the EV's from the different scenarios together to work out a historical EV.

What kind of log do you keep? Just curious - answer optional of course.

My log is getting quite extensive - I track the following for each session:
Duration, begin/end time, pen, result, std bet size, #of players, # of shoes per hour - then I also calculate and log EV for these conditions using CVD, and the EV from my own probability model

It would be great to hear more about people's actual results in terms of units/hour - assuming they have played for at least a couple of hundred hours.

 

[kewljason]

hi kewljason, thanks for posting your actual results. it's very helpful for me to have someone else's results on hand for reference. And it seems we are in a very similar range in terms of our resuults. Although if you are doing 70 hands/hour then i think 1.55 is quite a performance!

In fairness, I should point out that 70 hands/hour was just a very rough estimate/guess on my part, based on the fact that I rarely get to play heads up with the dealer where you can play quickly and really bump up that hands per hour number. I am very strick about wonging out of negative shoes, so most shoes that I start I do not finish, which means I am rarely sitting and waiting during shuffles. So as long as the casino I am in has a reasonable number of tables open, I think I am able to play more hands per hour than if I just sat at a semi full table playing for the entire hour, which in atlantic city would be about 3 shoes or 60 hands (dealer dependant of course) Maybe I am getting in more hands than I think. I will try to keep better track of hands played.

 

[Kasi]

As for different results for different types of games, I personally have no need to break down my results that way. 95% of my play takes play between 3 casinos in atlantic city which all have exactly the same rules. 6 or 8 decks, stand soft 17, no surrender. The only difference is penetration, and I wont play less than 75% and can rarely find more than 80-83%, so even that is similar.

So much the better. Makes life so much easier playing same game in same way more or less anyway

In that case, I'd guess maybe you reached your N0 a long time ago. Or at least anyway coming close by now? :grin:

Just a guess - don't get mad. I have no idea how you bet lol. Just seems plausible to me you've reached "long run".

Also, fwiw, believe it or not, I'm pretty sure you know, my feeling is that anyway, not that what I think matters an iota because it absolutely, absolutely, doesn't, or have a real good idea of, whether you have or not reached N0 by now, how close you may be, etc.

Basically, in a nutshell, able to compare EV and SD per round and be comfortable with where you are 1776 hours later

 

[Kasi]

Yes I spread to 2 hands for +2, +3, +4, and +5 and to 3 hands for +6, +7 and +8 (or higher). CV data allows for this in the analysis - you can incorporate multi-hand betting strategies in the simulations which is great. However this is one of my problems. In my own probability model i count a multi-hand round as one hand rather than several hands. My figure of 35000 hands may actually be more than that.

Don't feel all alone. It's one of my problems too lol. At least as far as SD goes, I sent you a sheet on how I'd do it. Maybe it's more that you were betting 1 unit at certain times but using 2.5 units as "unit" lol than "hands" vs "rounds".

Maybe just posting a sample CVDATA sim would allow for input by people that understand it better than I do and you could compare its results to the results of your model.

For some reason, I like to use initial bet advantage, initial avg bet, variance compared to initial avg bet, etc. I get confused with total bet adv for calcing EV or win rate/hr. Although I guess TBA*total wagered would equal IBA*initial avg bet. I'd definitely use initial avg bet variance and IBA for ROR calcs. I don't even know why lol. It just makes more sense to me to know the risk of placing your initial bet.

I even think CVDATA may have some feature, not sure where I got this idea, where you can tell it instead of betting 2/hds at $50, what if I bet $25 on 1 and $75 on the other. So, if so, make sure that that's not what it's doing for you in its calcs.

re EV=1 SD -> are you referring to N0?

Yep lol.

Don't leave home without it

I individually track the EV for each session so that I know how good or bad the playing conditions are/were - and to then help me to repeat the good ones and avoid the bad ones. But I dont actually add the EV's from the different scenarios together to work out a historical EV.

Even if you did, what's the point? Just because you may have a higher avg EV doesn't mean it's a better game to play. What matters is best use of your money to produce highest hourly growth given a risk. In other words, what's wrong with picking game with highest SCORE?

If you want highest EV, bet all your money in some huge +count every time.

My log is getting quite extensive - I track the following for each sessionuration, begin/end time, pen, result, std bet size, #of players, # of shoes per hour - then I also calculate and log EV for these conditions using CVD, and the EV from my own probability model

Does the EV ever differ from CVD compared to prob model? What's "std bet size" mean"? Like a specific $ ramp? Does "duration" mean hours or rounds played? I'm not even sure why I'd care about shoes per hour. You don't even mention whether SD ever figures into a session's results. Does it?

It would be great to hear more about people's actual results in terms of units/hour - assuming they have played for at least a couple of hundred hours. [/QUOTE]

It simply doesn't matter what has happened to anybody else. Nobody, not a BS player or a perfect card-counter even with a computer at his side, can control variance. It's God's will whether the coin lands heads or tails.

Knowing expected units/hr is utterly meaningless. I could make 10 units an hour or 40 units an hour. The former could have high risk compared to roll, the latter, low risk. Your expected units per round is what it is. Having a large roll won't change that. It'll only change your risk of losing all your roll.

 

[matt21]

In fairness, I should point out that 70 hands/hour was just a very rough estimate/guess on my part, based on the fact that I rarely get to play heads up with the dealer where you can play quickly and really bump up that hands per hour number. I am very strick about wonging out of negative shoes, so most shoes that I start I do not finish, which means I am rarely sitting and waiting during shuffles. So as long as the casino I am in has a reasonable number of tables open, I think I am able to play more hands per hour than if I just sat at a semi full table playing for the entire hour, which in atlantic city would be about 3 shoes or 60 hands (dealer dependant of course) Maybe I am getting in more hands than I think. I will try to keep better track of hands played.

If you are wonging a lot, then i would expect you to make more per hand than me because i am playing a lot more hands but often at lower counts. So your figure of 60 may well be correct. Although if i imagine you back-counting then i would think you would be playing less than 60? But if you are playing from the start and then just wonging out, and then switching to another table that's starting, then 60 seems right.

Thanks again for sharing this info - very helpful!

Good luck!

 

[matt21]

Hi Kasi, thanks for more comments and your emails - got them and started looking at them

Even if you did, what's the point? Just because you may have a higher avg EV doesn't mean it's a better game to play. What matters is best use of your money to produce highest hourly growth given a risk. In other words, what's wrong with picking game with highest SCORE?

If you want highest EV, bet all your money in some huge +count every time.

Does the EV ever differ from CVD compared to prob model? What's "std bet size" mean"? Like a specific $ ramp? Does "duration" mean hours or rounds played? I'm not even sure why I'd care about shoes per hour. You don't even mention whether SD ever figures into a session's results. Does it?

What is SCORE?
My model and CVD do not produce the same EV - but i am looking for them to produce results in somewhat a similar range.
By std bet size i am referring to what my standard unit. Currently, my standard unit is $25 and this unit determines my $ ramp. At lower counts i bet less than the standard unit (e.g. $10) - i do that instead of wonging out.
Duration means hours. SD does not figure into my session's results. Should it?

It simply doesn't matter what has happened to anybody else. Nobody, not a BS player or a perfect card-counter even with a computer at his side, can control variance. It's God's will whether the coin lands heads or tails.

True, but i do find it useful to compare results. Once people play for longer periods of time, then results should start to become similar. A chart of my hourly EV is quite interesting. In my case, it started flattening out after around 150 hours.

Knowing expected units/hr is utterly meaningless. I could make 10 units an hour or 40 units an hour. The former could have high risk compared to roll, the latter, low risk. Your expected units per round is what it is. Having a large roll won't change that. It'll only change your risk of losing all your roll.

I find EV/hour very useful assuming that ROR is below an acceptable level. (you would also need to consider ROR if you are using EV/round) Number of shoes per hour influences number of hands per hour, whihc in turn impacts EV/hour. That's why i keep track of these statistics. If certain playing conditions give me an hourly EV of 0.7 units, and others give me an EV of 1.7 units, and both have a ROR less than 6% then obviously the latter is a better choice for me.

 

[Kasi]

What is SCORE?
My model and CVD do not produce the same EV - but i am looking for them to produce results in somewhat a similar range.
By std bet size i am referring to what my standard unit. Currently, my standard unit is $25 and this unit determines my $ ramp. At lower counts i bet less than the standard unit (e.g. $10) - i do that instead of wonging out.
Duration means hours. SD does not figure into my session's results. Should it?

Hi Matt,
I'm sometimes a little slow but I'm slowly catching on to your model lol. Using min bets as fractions of unit kind of threw me for a while lol. Also I'm just so used to an hour meaning 100 hands lol.

SCORE is just something that assumes certain things like a fixed $10K roll, and tries to measure what game is a "better" use of one's money given optimal risk and reward and, I think, playing or seeing 100 rounds an hour.

I don't know why your model wouldn't get the same answer as CVD if you used the same freq, var, adv, spread etc. If you don't it's probably only because of your underlying assumptions like how many hands you may play in an hour or be able to spread to so much in the higher counts with so many players or stuff like that.

No, you're right SD doesn't have to figure into a session's results if all you want to know is EV. It would effect risk to roll, trip, session, lifetime,liklihood of actual results for so long, etc as I know you know.