MJ: Question for Don or Norm: How many sessions will a counter win?

[MJ]

I read that a skilled card counter will win, lose, and push roughly 43%, 49%, and 8% of all hands played respectively for a typical 6 deck game with good rules (DAS DA2, LS, 75%Pen). Does that mean for every 100 sessions a counter plays, he will win, lose, and break even 43, 49, and 8 sessions respectively?

This seems logical based upon the percentages. Now some might argue that if a counter loses more sessions than he wins, then what the hell is the point? I propose that although he will lose more sessions, the net sum of money won for the winning sessions will be greater than the net sum of money lost during the losing sessions. Thus, the counter will pull ahead despite losing more sessions than he wins!! Is my thinking correct?

Further, in The Counter by Kevin Blackwood, the protagonist states that a winning counter will gradually pull ahead in the long run and win 2 out of every 3 sessions played (I will double check that figure later). Anyhow, that session win rate seems ridiculous and erroneous.

MJ

[victoria]

> I read that a skilled card counter will win, lose, and
> push roughly 43%, 49%, and 8% of all hands played
> respectively for a typical 6 deck game with good rules
> (DAS DA2, LS, 75%Pen). Does that mean for every 100
> sessions a counter plays, he will win, lose, and break
> even 43, 49, and 8 sessions respectively?

> This seems logical based upon the percentages. Now
> some might argue that if a counter loses more sessions
> than he wins, then what the hell is the point? I
> propose that although he will lose more sessions, the
> net sum of money won for the winning sessions will be
> greater than the net sum of money lost during the
> losing sessions. Thus, the counter will pull ahead
> despite losing more sessions than he wins!! Is my
> thinking correct?

> Further, in The Counter by Kevin Blackwood, the
> protagonist states that a winning counter will
> gradually pull ahead in the long run and win 2 out of
> every 3 sessions played (I will double check that
> figure later). Anyhow, that session win rate seems
> ridiculous and erroneous.

> MJ

Not Don or Norm
But, even though you will lose more hands than you win, it does not correlate to session wins and loses because dealers do not get paid 3/2 on a blackjack nor can they double or split. This is important because when the count is high and you have more money on the felt, you will get more blackjacks and be more successful on your doubles and splits.
Session wins are based on dollars won and not on the amount of hands won.

Victoria

[Don Schlesinger]

> I read that a skilled card counter will win, lose, and
> push roughly 43%, 49%, and 8% of all hands played
> respectively for a typical 6 deck game with good rules
> (DAS DA2, LS, 75%Pen). Does that mean for every 100
> sessions a counter plays, he will win, lose, and break
> even 43, 49, and 8 sessions respectively?

No, not at all. You're talking about hands, not sessions. And, it makes no difference how many hands you win, if you get paid more for the winning ones than for the losing ones.

> This seems logical based upon the percentages. Now
> some might argue that if a counter loses more sessions
> than he wins,

He doesn't -- unless he's a lousy counter! :-)

> then what the hell is the point? I
> propose that although he will lose more sessions,

He won't.

> the
> net sum of money won for the winning sessions will be
> greater than the net sum of money lost during the
> losing sessions.

Well, that part is true.

> Thus, the counter will pull ahead
> despite losing more sessions than he wins!!

See above.

>Is my thinking correct?

No. See BJA3, Table 2.2, p. 21. In that table, a session is one hour of play. Obviously, a question such as yours can't be answered, if you don't define the length of a session.

> Further, in The Counter by Kevin Blackwood, the
> protagonist states that a winning counter will
> gradually pull ahead in the long run and win 2 out of
> every 3 sessions played (I will double check that
> figure later). Anyhow, that session win rate seems
> ridiculous and erroneous.

Maybe his session length is longer than one hour. Also, Kevin played a lot of single-deck, where the percentage of winning sessions, in his day, might have been higher than for shoes. But, 67% is straining things a little. :-)

Don

[Parker]

Session results are meaningless. It's all one session; all that matters is how much money is won (or lost) at the end.

The only reason we even have sessions is because the body requires things like food, rest, and bathroom breaks, and very few casinos will tolerate an aggressive counting game for hours on end.

Furthermore, session results can be manipulated to "prove" just about anything. For example, suppose we use a 2 unit stop-win and a 100 unit stop-loss. Our session win-loss rate will be outstanding - and bear little or no relationship to the net total dollar amount won or lost.

Why even concern yourself with session results?

[MJ]

> And, it makes no difference how many hands
> you win, if you get paid more for the winning ones
> than for the losing ones.

Then your saying %win/loss rates for hands played are irrelevant so long as you win more money on the winning hands than you lose on the losing hands? That is all well and good, but there is the concept of certainty equivalent which must be taken into consideration. What if you only won 1 hand out of 100 and lost the other 99 (of course the earnings on that 1 hand exceed the losses on the other 99)?
This game in theory has a positive EV but the variance is brutal!!!

> He doesn't -- unless he's a lousy counter! :-)

As Victoria pointed out, the payoffs on hands played are not even money due to BJs, splits, insurance, double downs, etc.
For the sake of argument, what if all payoffs in BJ were even money? Now would the %win/loss rates for sessions played resemble the %win, loss, and break even rates for hands played? If so, then logic would dictate that the actual %win/loss rates for sessions played when the payoffs are NOT even money would be somewhere in the ball park of 43%, 49%, 8% (but more favorable to the player).

> No. See BJA3, Table 2.2, p. 21. In that table, a
> session is one hour of play. Obviously, a question
> such as yours can't be answered, if you don't define
> the length of a session.

BJA3 is the next book on my list. :-)

On one hand I see your point, on the other hand I don't.
I think you are implying that the longer the session, the greater the %win/loss rate for all sessions played. IE, the %win/loss rate for 100 1-Hour sessions will be less than the %win/loss rate for 100 5-Hour sessions.

OTOH, length of the session should not really matter. %Win/loss rates for sessions played should on average loosely resemble %win/loss/push rates for hands played (for reasons given above). Hence, whether it be a 1 hour session or a 10 hour session, if you play enough sessions in the long run the %win/loss rate in either case should approach the %win/loss/push rates for hands played.

> Maybe his session length is longer than one hour.
> Also, Kevin played a lot of single-deck, where the
> percentage of winning sessions, in his day, might have
> been higher than for shoes. But, 67% is straining
> things a little. :-)

I checked the book and Kevin was playing SD games. On page 87, Blackwood states "If a card counter played correctly, he would win about two out of every three days".

MJ

[MJ]

I agree with everything you wrote Parker. But, we are only human after all and we like to assess our play from session to session. Also, if you have some idea of %win/loss for sessions, then you have another way to determine if you are on the right track beside standard deviation. If you only win 2 out of every 10 sessions, you are doing something wrong and do not even need knowledge of SD to tell you that.
I find it comforting to know in theory what % of sessions played I will win and then compare to my actual session results. Hope that makes sense.

MJ

> Session results are meaningless. It's all one session;
> all that matters is how much money is won (or lost) at
> the end.

> The only reason we even have sessions is because the
> body requires things like food, rest, and bathroom
> breaks, and very few casinos will tolerate an
> aggressive counting game for hours on end.

> Furthermore, session results can be manipulated to
> "prove" just about anything. For example,
> suppose we use a 2 unit stop-win and a 100 unit
> stop-loss. Our session win-loss rate will be
> outstanding - and bear little or no relationship to
> the net total dollar amount won or lost.

> Why even concern yourself with session results?

[David Spence]

> As Victoria pointed out, the payoffs on hands played
> are not even money due to BJs, splits, insurance,
> double downs, etc.
> For the sake of argument, what if all payoffs in BJ
> were even money? Now would the %win/loss rates for
> sessions played resemble the %win, loss, and break
> even rates for hands played?

Not at all. As the length of a session increases, the session win rate approaches 100% (if ever so slowly), even if the win rate on an individual hand is only slightly more than 50%. Think about it this way: the casino has only a slight edge (1.4%) on the pass line at craps. In this case, the casino only wins slightly more than 50% of pass line bets. Nevertheless, if you consider a session to be one year's play at the pass line, the casino's win rate is virtually 100%. That is, the pass line almost never has a losing year for the casino. The session win rate of nearly 100% in this case is MUCH higher than the individual bet win rate of just over 50%.

Back to blackjack. Even you only have a marginal edge, and the chance of winning a single hand is close to 50%, this does NOT translate into the chance of having a winning session being close to 50%. A small advantage, whether it's from winning more hands than losing (what you've stipulated for the sake of argument), or from having larger bets on winning hands (the truth), can lead to SESSION win rates of significantly greater than 50%.

Put another way, the more time you allow your edge to grind away at the house, the more likely it is that you'll emerge a winner. For an advantage player: short sessions = win rates just above 50%, longer sessions = ever increasing win rates. Many factors are at work, of course, but a session length of 500 hours may produce a win rate of roughly 80%, even if the individual hand win rate is close to 50%.

[Don Schlesinger]

> Then your [you're] saying %win/loss rates for hands played are
> irrelevant so long as you win more money on the
> winning hands than you lose on the losing hands?

Yes, absolutely.

> That
> is all well and good, but there is the concept of
> certainty equivalent which must be taken into
> consideration. What if you only won 1 hand out of 100
> and lost the other 99 (of course the earnings on that
> 1 hand exceed the losses on the other 99)?
> This game in theory has a positive EV but the variance
> is brutal!!!

Well, that isn't the case for BJ; the average squared result of a hand is about 1.15, and the manner in which you optimally bet your money reflects this variance. When you develop a roulette system, your optimal Kelly wager will be different! :-)

> As Victoria pointed out, the payoffs on hands played
> are not even money due to BJs, splits, insurance,
> double downs, etc.

We know that already! :-)

> For the sake of argument, what if all payoffs in BJ
> were even money? Now would the %win/loss rates for
> sessions played resemble the %win, loss, and break
> even rates for hands played?

Not if you were a card counter and bet more when you had the edge than when you didn't. But, without the 3 to 2 on naturals, you can't have the edge.

> If so, then logic would
> dictate that the actual %win/loss rates for sessions
> played when the payoffs are NOT even money would be
> somewhere in the ball park of 43%, 49%, 8% (but more
> favorable to the player).

I'm not following your point.

> BJA3 is the next book on my list. :-)

Better late than never. :-)

> On one hand I see your point, on the other hand I
> don't.

I'm sorry.

> I think you are implying that the longer the session,
> the greater the %win/loss rate for all sessions
> played.

Yes, of course. I would hope that if your session were 10 hours, instead of one hour, you would have a higher winning percentage for the 10-hour ones. Otherwise, something is terribly wrong with your play.

IE, the %win/loss rate for 100 1-Hour sessions
> will be less than the %win/loss rate for 100 5-Hour
> sessions.

You're not stating it correctly. Your win rate, for playing the game, doesn't vary with the length of the session. But, your percentage of winning sessions varies, according to how you define the length of a session. So, suppose I win at the rate of 1% of all money bet. That can't change, on average, whether I play for an hour or a year. It is what it is. But, if I play 100 one-hour session, I will win about 52-53% of them. If I play 100 10-hour sessions, I will win about 59% of them. Do you understand the difference?

> OTOH, length of the session should not really matter.

See above.

> %Win/loss rates for sessions played should on average
> loosely resemble %win/loss/push rates for hands played
> (for reasons given above).

You're stuck on a totally meaningless concept that you badly need to drop.

> Hence, whether it be a 1
> hour session or a 10 hour session, if you play enough
> sessions in the long run the %win/loss rate in either
> case should approach the %win/loss/push rates for
> hands played.

Nope.

> I checked the book and Kevin was playing SD games. On
> page 87, Blackwood states "If a card counter
> played correctly, he would win about two out of every
> three days".

Well, a day isn't a session, now is it?? :-)

Don

[Don Schlesinger]

> I agree with everything you wrote Parker. But, we are
> only human after all and we like to assess our play
> from session to session. Also, if you have some idea
> of %win/loss for sessions, then you have another way
> to determine if you are on the right track beside
> standard deviation. If you only win 2 out of every 10
> sessions, you are doing something wrong and do not
> even need knowledge of SD to tell you that.
> I find it comforting to know in theory what % of
> sessions played I will win and then compare to my
> actual session results. Hope that makes sense.

It makes sense under these (impossible) conditions only:

1. Every game you play has exactly the same number of decks, rules, people at the table, and speed of the dealer (in other words, hands per hour).
2. You play precisely the same amount of minutes, for every session, without ever deviating.
3. Similar to the above, you never try to "manipulate" a session result, by quitting prematurely, to preserve a win, or playing on, past your regular time, to try to dig out of a hole and change a loss into a win.

Good luck with all of the above! :-)

Don

[MJ]

[johnnienolan]

> It makes sense under these (impossible) conditions
> only:

> 1. Every game you play has exactly the same number of
> decks, rules, people at the table, and speed of the
> dealer (in other words, hands per hour).
> 2. You play precisely the same amount of minutes, for
> every session, without ever deviating.
> 3. Similar to the above, you never try to
> "manipulate" a session result, by quitting
> prematurely, to preserve a win, or playing on, past
> your regular time, to try to dig out of a hole and
> change a loss into a win.

> Good luck with all of the above! :-)

> Don

The conditions above are easy enough against a computer game. You can try and force 'longer' winning streaks by stretching a session out in order to sustain a streak. That's great if you still aren't a true believer yet. But to prepare yourself for the swings, a fixed time is a much better preparation, especially if you are still capable of going on any kind of tilt. To the extent that losing at the start of a session usually provides you with some cover in real play, stretching a session out can make sense.