Discount Gambling

Counting CSM Blackjack (+EV)

Posted in +EV, blackjack, csm by stephenhow on July 27, 2012

A few people have asked me about the countability of blackjack dealt from a constant shuffle machine (CSM). I’m a big advocate of counting the CSM, especially for card craps, because of the ease of the windowed count. Even if the dealer collects no muck (i.e., immediately shuffles completed hands back into the CSM), you’ll still be +EV more than 8% of the time for good BJ rules. It’s a lot easier to count a CSM than a shoe. I call it counting for the ADHD crowd. All you have to do is pay attention to the last 16 cards (or the collected muck) fed into the CSM before the hand is dealt. Plus, you’ll probably never get backed off from CSM blackjack, even when wildly varying your bets.

EV vs. Windowed Count

I use my model of the ShuffleMaster 126 (source) CSM in the blackjack simulations for this post. I’ve talked in detail about this model before, in my posts on card craps. Basically, there’s a buffer of at least 16 cards in the chute (so the dealer never waits for a card), which introduces state into the system. If the dealer collects no muck, you simply use the running count of the last 16 cards fed into the shuffler. Use the simple hi-lo count (2-6 are +1, 10-A are -1). If the dealer collects a lot of muck, and feeds it all into the shuffler right before the next hand starts, then use the count of the entire muck.

For these simulations, I used 6 decks in the CSM, and typical-good H17 blackjack rules (3:2 BJ, late surrender, re-split Aces 3-times, double-after-split). My blackjack analyzer calculates the ideal EV for these rules at -0.445% for 6 decks. I ran the simulator head’s up against the dealer, and kept track of the 16-card windowed count and the subsequent hand outcome. I plotted the next-hand EV vs. the windowed hi-lo count in the graph below.

The graph shows a very linear relationship between the 16-card windowed hi-lo count and the EV of the next hand. When the running windowed count is +5 or more, the next hand from the CSM is +EV. The windowed count is ≥ 5 about 8.2% of the time.

Frequencies of 16-Card Windowed Counts
Count Frequency Approx. BJ EV
0 11.5% -0.44%
1 10.8% -0.34%
2 9.36% -0.24%
3 7.46% -0.16%
4 5.46% -0.05%
5 3.67% +0.04%
6 2.25% +0.14%
7 1.25% +0.22%
8 0.627% +0.29%
9 0.280% +0.43%
10 0.111% +0.49%


Serious card counters will tell you you can’t count a CSM. But the data above shows that a CSM goes +EV more than 8% of the time. Plus, it’s infinitely easier to count a CSM than it is to count a shoe. You can lose track of the count for a hand or two. As soon as you regain attention, you’ll know what the count is. You can probably vary your bets wildly without attracting any attention or interest from the floor. You can probably even Wong hands when the count is bad. Or less than +5.

Counting a CSM is great for the casual counter. It’s basically short-attention span counting. If you see the last 16 cards into the CSM prior to the deal are low (have a running count of 5 or better), then you’re +EV for the next hand. Even if you just see a net +3 count for the last 16 cards, you still know the next hand will be better than average. You can start/stop paying attention on a per-hand basis (unlike a shoe, where you have to wait for the next shoe if you lose the count).

At it’s simplest, CSM counting will tell you when the next hand will be better-than-average (half the time), or worse-than-average (half the time). So, if you Wong half of the time, you’ll only play the better-than-average hands (EV better than -0.45%; the above curve to the right of count=0), and miss the bad hands. That’s a quick way to reduce the house edge from 0.45% to 0.22% (only play 53% of the hands; wait until the count is ≥ 0).

46 Responses

Subscribe to comments with RSS.

  1. oyster625 said, on July 28, 2012 at 6:01 am

    Could this principle apply to CSM games like Pontoon? or Switch Blackjack?

    • stephenhow said, on July 28, 2012 at 9:46 am

      Yes, you can apply a windowed count to any game dealt out of a CSM. The effectiveness of the count depends on the strengths of the EORs (effect of card removal) for the particular game. I’ll definitely look into Blackjack Switch; that’ll be interesting!

  2. J said, on August 23, 2012 at 1:59 am

    Hi, you alluded to the fact that serious card counters will tell you that you can’t count a CSM. That’s definitely the conventional wisdom. How sure are you about your assumptions re: windowed count? Might you be wrong? If you are right, why isn’t this info more widely known?

    Sorry to sound skeptical but it feels like you must be missing something, somewhere.

    • stephenhow said, on August 23, 2012 at 7:11 am

      Serious counters will concede a CSM is countable to a degree, but they’ll say it’s not worth counting. My numbers are correct, and the windowed count works as stated. I also argue that it’s easier to count a CSM, because you can vary your bet wildly ($5 to $100, a 20x swing would get no attention; Wong-ing would get no attention), and you can lose the count temporarily.

      I’m correct about all of this. I think it’s good to know about this effect, but I don’t think it’s very profitable in the scheme of things.

      The effect is stronger for 4-deck CSM Pontoon, which I’ll publish later.

  3. Josphe said, on September 18, 2012 at 12:32 pm

    I’d agree that’s a way to loss less at least, pretty good for leisure play.
    Just wondering if the card just inserted have the same probability to reappear at the end of the buffer as the other cards or not. I think that depends on the mechanic, IIRC one of the selling point of the ShuffleMaster is that the shuffle is not perfectly random, so I support the answer is no? Just a thought, might take that into consideration. But I guess it’s too hard to put that into the equation even if it’s significant.

    • stephenhow said, on September 18, 2012 at 1:06 pm

      Right, the last cards fed into the CSM have *zero* probability of coming out in the N cards, where N is the minimum buffer depth of the chute (I use N=16). It’s hard to make an ideal (perfectly random) CSM, because it would slow down the dealing process. The N-card buffer in the card chute speeds up the dealing process, but affects the randomness of the cards out of the shoe (relative to the recent cards dealt). That’s the point of this post. I modeled the CSM with an N-card buffer in the chute, and showed how to use a N-window count against it.

  4. flixo21 said, on September 25, 2012 at 5:58 pm

    I know some casinos with very good rules (0,11% edge off the top) but using one2six shufflemaster
    I want to play it but I didn’t understand everything of your post.
    What is “buffer” and “muck”
    it depend of the casino, they put back the discard sometime directely after each round, sometime after 2 or 3 round. My question is, when to begin counting hilo and when to go back to “zero”
    for example ; if the count is +5 at the end of the first round
    I bet for a RC of +5 or a TC of 0,5-1 in the 2nd round (the dealer put back the discard at the end of the first round) and count again from zero for the 2nd round ?
    How do I do if the dealer decide to put back the discard ‘during’ a round and not at the end …

    Thanks you in advance

    • stephenhow said, on September 26, 2012 at 9:43 am

      The buffer is a minimum number of cards maintained in the front of the chute where the cards egress the one2six. The muck is the accumulated discards before it is placed back in the shuffler. For a CSM, you should use the running count of the muck plus the last N (buffer depth) cards fed into the shuffler. If there is no muck (it was just fed back into the shuffler), then only use the count of the last N cards.

      • flixo said, on April 23, 2013 at 5:18 pm

        I understand, but when the dealer put back the discard (muck) into the csm there is a time to wait in order to have these cards reshuffle. how to know when I have to reset my RC ? (when the muck is completely in the csm)?

    • Wilson Fang said, on June 20, 2013 at 1:42 am

      If only forfeited for original bet but not all bet for split or double down while house blackjack…the house edge is around 0.11%…Venetian blackjack in Macau…

  5. flixo21 said, on April 23, 2013 at 5:22 pm

    The cards putted back into the csm make time to be completely reshuffled in to csm, how to know when to reset EXACTLY my RC in the case with a muck of 1 deck etc… ?

    • stephenhow said, on April 23, 2013 at 5:32 pm

      In the above post, I describe the use of a *windowed* running count, that’s never reset. However, it’s often much more convenient (and practical) to reset the count after a round of hands, particularly when the muck = last round of hands. I’ll post soon on when conditions work for non-windowed (reset after each round) count with a CSM.

      • Wilson Fang said, on June 19, 2013 at 11:08 pm

        If the dealer waited for the CSM to shuffle, this counting theory doesn’t work….pls leave the table n stop to play…..cheers

      • James said, on August 29, 2013 at 5:52 am

        Steve, It’s not practical to use *windowed* running count that’s never reset. For example :-

        Round 1, RC = +2, no of cards used =5,
        Round 2, RC = +1, no of cards used =6,
        Round 3, RC = +3, no of cards used =5,

        RC for last 16 cards = +6

        Round 4, RC = -2, no of cards used =5,

        RC for last 16 cards = 1+3-2 = +2 !

        So, before round 5, you must deduct the no of card used in round 1 and the releveant RC, IS IT PRACTICAL !?

        Please post on when conditions work for non-windowed (reset after each round) count with a CSM.

      • James said, on August 29, 2013 at 6:02 am

        In order to maintain windowed count 16 cards, you may consider 4 players + dealer( average cards used/round = 5 x 2.7 = 15.5 cards !). Then you can reset RC after each round ?

      • James said, on August 29, 2013 at 6:05 am

        sorry, should be 5 players + dealer ! no cards used/round = 6 x 2.7 = 16.2

    • Wilson said, on September 2, 2013 at 10:43 pm

      Dear Steve, James asked a good question. He only counted when windowed cards = 16 by moving average. To simply the process, i refresh my memory and count on every round by playing 2 boxes. Is it more accurate by using James methods??? pls advise

  6. flixo21 said, on April 24, 2013 at 4:55 am

    Thanks you for your anwser …

    Some casinos just wait 1 round before putting back the discard into the csm, some others don’t wait and put back during the round, but also I know some casinos where they wait 2 decks even more before putting it back. I’m pretty sure it can be profitable, espescially with good rules ? but I don’t know exactly how to count it, when to reset RC etc …

    Is there any counting system better than other for counting a CSM ? or Does it exist some different strategy for CSM ? I ask you that because I know some pros who now play on CSM with succes and I don’t know exactly what they are doing …

    Thanks you very much

  7. Wilson Fang said, on May 12, 2013 at 7:31 pm

    i used a 5m simulator by using 7 boxes data, stand onsoft 17, double on 9, 10,11, surrender allow, split for 3 times, no peek at blackjack, double allwoed after split, A split once, and CSM machines (house edge = -0.5%, -ve meant advantage for casino) averagely around 28-30 cards info to do the card counting,

    2,3,4,5,6 = +1,
    7,8,9 = 0
    A,T,J,Q,K = -1

    Bet 1x, count <+5
    Bet 2x, count = +5
    Bet 3x, count = +10
    BEt 4x, count = +15

    I prepared 1000 unit capital to fight with Casino, after 1000 hours played in Casino during weekend for 1 year ++, I made a small profit, around +0.1% from my total bet…I believe it works

    • stephenhow said, on May 14, 2013 at 12:23 pm

      Thanks for the confirmation on the countability of CSM blackjack. I’m probably going to upload a CSM BJ simulation kit to Github when I get around to it. There’s a few people out there that care about counting CSM blackjack, and want to evaluate custom strategies.

      • Wilson Fang said, on June 19, 2013 at 11:03 pm

        Sorry for late reply. Dear stephenhow….yes I implemented few new strategies…can make 0.5% house edge for players….just basic strategy, surrender at 14,15,16 vs house face card, 16 vs house 9 card, takeover our players’ bet when they made mistake in surrender… in Macau and phillipine casino….100 x 2 boxes….40 hands per hours….continue your good jobs on this web although u r lonely for nobody knows to read your efforts…..

      • Daniel said, on July 29, 2013 at 6:39 am

        I am eager to know it too. There is no way for me to play shoe games but csm games in a frequent basis.

      • mackdad said, on October 11, 2013 at 12:41 pm

        That sim would be cool to see. Last time I played against CSM, I found myself having luck just by sitting out hands after I saw the count go -5 or more. There’s no rule that you have to play the next hand when other people are at the table…or is there?

  8. Wilson Fang said, on May 12, 2013 at 7:41 pm

    somemore, u can takeover other players surrender’s bet. says surrender on 12 or 13 vs dealer’s face card, or surrender on 14,15 vs dealer’s 9 or didnt double down on 9,10,11 (please get approval becos 1 card allowed to be drawn)

    • Wilson Fang said, on June 19, 2013 at 11:14 pm

      Believe me, randomness is an order…..pls read the book by Nassim Talem, never fooled by randomness…..i like his sharp observation…..cheers

  9. James said, on May 19, 2013 at 10:15 pm

    Hi Steve,

    It’s been a while since I’ve talked with you. Hope you are doing well with your Lunar Poker.

    If the dealer put the cards into the CSM after every round, say after the 1st round, the total no of cards dealt = 9 cards and cards dealt in 2nd round = 12 cards, should we use 16 cards buffer or 21(9 + 12) cards windowed count ?

    Do you received my email about the “special” basic strategy which will give the player an edge of +0.032% ?


    • Wilson Fang said, on June 19, 2013 at 11:36 pm

      I didnt notice that how to have player edge of +0.032%….I will compute lunar poker tonight…..but house edge for baccarat insurance when banker 6 vs player 5, banker 6 vs player 4, and banker 6 vs player 0(both 2 cards). Players are encouraged to buy insurance, if pay 3 to 1…..

      • James said, on March 4, 2014 at 2:45 am

        wilson, you can do simulation ? How to contact you personally ?

      • Wilson said, on April 18, 2014 at 7:34 pm

        James, sorry for late reply, I never provide simulation service for anyone….I counted blackjack for hobby, my ultimate goal is beat the house edge for fun and used for beers-money to drink freely at night, cheers!! I’m not a pro gamblers…You may look for wrong guy…

        FYI, As maths aspects, I tested Stephen How’s idea for around 2 years. I concluded his method working in long run, I played for 12,000++ hours in 24 months ( 70 HPH, around 840,000 games, average weekly hours = 4.4 hours a week) by using his methods, I made player’s edge for (E.V. = +0.8%) so far….

        - blackjack, pays 3 to 2 (1.5)
        - early surrender allowed (14 vs 10, 15 vs 10, 16 vs 10, 16 vs 9)
        - double-down (9 vs 3,4,5,6,10 vs 2-9,11 vs 2-10)
        - double-down for any soft-hand, 7,8 vs 3,4,5,6; 5,6 vs, 4,5,6; 3,4 vs 5,6
        - Split on A once; one card allowed
        - split twice, up to 3 hands
        - double-down allowed after splitting
        - Original bet only on split or double-down, if encounter on banker’s blackjack.
        - No insurance bet and No even money forever
        - CSM usage.(Zero mucks)
        - prepare 100-200 units on Wager Capital, place minimum wager (1 unit) when TC is +5 or less than +5

        House edge of blackjack is 0.38%-0.48%

        Note: Due to 6 decks in CSM,

        if RC (on 16 cards) = +2, TC = 2/5.75 = +0.35.
        if RC (on 16 cards) = +6, TC = 6/5.75 = +1.04

        * 5.75 = 23/4, easy to approximate

        1 * 4/23 = +0.17
        2 * 4/23 = +0.35
        3 * 4/23 = +0.52
        4 * 4/23 = +0.70
        5 * 4/23 = +0.87 (I start wagering bigger bet here)
        6 * 4/23 = +1.04
        7 * 4/23 = +1.22
        8 * 4/23 = +1.39
        9 * 4/23= +1.57
        10*4/23= +1.74

        ** Use 0.17 multiplied with n number of RC


        To me, TC of +1 is equivalent to +0.5% house edge

        ** After reading this forum since 2012, get inspiration from Stephen’s idea, I started buying all BJ books (Theory of blackjack and Extra Stuff by Peter Griffin, Professional blackjack by Stanford Wong, beat the dealer by Erward Thorp and Gambling 102 by Michael Shackleford, founder) from amazon, read all probability and gambling books, purchase numerous simulator software ( and Proficient blackjack) and test the idea.

        Don’t be greedy, nobody able to win the casinos all the time, sometimes u may lose in 19 games consecutively…this game must be played in long run (N = 500-1000 hands) and set a max lost limit of a day (says, not more than 50 units a trip)!! Bear in mind, $1000 only made $8 in long run. You need to win 6.25 trips if you lost 50 units, and lastly this calcsheet is for fun purpose…can’t make as business (casino will bar u from playing, don’t waste time), cheers!!

      • James said, on April 18, 2014 at 9:30 pm

        Thanks, Wilson.

        My bj game here with 0.168% house edge. I use REKO system and count 16 cards, So far, the results is acceptable.

  10. Chanter said, on August 1, 2013 at 7:54 am

    It appears that 16 cards is the minimum buffer size, but in fact judging from my data set, the cards just reinserted aren’t readily available for play beyond that level either. Does any one know the exact card level at which cards become available for play? Seems like this is a stronger play than Stephen thought.

    • James said, on August 29, 2013 at 6:45 am

      My source said the buffer is about 20 cards

  11. michelle said, on September 5, 2013 at 9:13 am

    james,I thing no need to reset Rs after each round(4 card 5 card or six card)you just need count 16 card(16 card =1 round)you must be 2 consecutive round(1 round =16 card)get 1 True count,than bet the big wager,but you get two round -1 and then get two round +1,you do not bet the big wager.for example(one round -1 and then two round +1 ,bet the big wager.two round -1 and then four round +1,bet the big wager)

  12. michelle said, on September 15, 2013 at 7:06 am

    just bet the big wager when you get two consecutive(1 round use 16 card) round with TC+1 or more,because you never know how many card surplus on the count two consecutive round is guarantee certainly have Tc+1 or more than +1(As james say the cards is complete)but you will crazy to wait two consecutive Tc+……These day i juse play small miniman with strategy and observation,only get minority Tc+…it also does not matter,the most tragic thing,yesterday i findings the depth of the chute only 10 cards…..oh my god ……Steve!!!Where are you!!!We miss you!

  13. michelle said, on September 15, 2013 at 7:28 am

    sorry,one round use about 16 cards

  14. HannuV said, on October 9, 2013 at 8:53 am

    I’d be very interested if you could upload the program or source code, I’ve got a game in town that has 0,16% house edge off the top and is dealt from Shufflemasters 126 so would really love to try this out :).

  15. mackdad said, on October 17, 2013 at 1:29 pm

    This system could impact various side bets too.

  16. flixo said, on December 22, 2013 at 5:32 pm

    I have slight off the top advantage :

    dealer wait to deal the second round and put back the discard just before asking for additionnal cards to the players…
    My idea was to play 7 boxes to get about 2 rounds latency : 7x 1/5 of my max wager at 0 and put 7 x max wager at TC1 (7xmin table if negative)

    To do it correct I’m questionning about few things :

    example : first round I put 7×30$ and my hilo RC go to +8 , so for the 2nd round TC= +1 and I put 7×150$.
    When the dealer deal on the 2nd round the 15 cards he then put back the discard in the machin, at this moment when do I have to reset my RC ? I have to keep a new RC of the 15 cards of the second round+ additionnal card taken from the players and use this count to bet for the third round ? etc …. etc …

    Are you sure news one2six have a buffer of 16 cards in the chute ? I heard of 9 cards instead ? and how to use this addionnal information to my betting strategy above ?

    Thanks stephen!

    • Daniel said, on December 27, 2013 at 3:20 am

      I can confirm that there is 16 cards in the chute of one2six.
      Once I saw the dealer pulling the cards 1 by 1 out of the machine when it stopped working properly and need to be replaced with another one. After 16 cards there is no card in the chute.

      • flixo said, on December 28, 2013 at 7:35 am

        thanks you for your answer, can somebody confirm the way to count my game.

      • Stuart said, on December 29, 2013 at 8:59 am

        Hi Flixo,

        Quote= “first round I put 7×30$ and my hilo RC go to +8 , so for the 2nd round TC= +1 and I put 7×150$.
        When the dealer deal on the 2nd round the 15 cards he then put back the discard in the machine, at this moment when do I have to reset my RC ? I have to keep a new RC of the 15 cards of the second round+ additional card taken from the players and use this count to bet for the third round ? etc …. etc”

        You may email me at Would like be in touch with someone doing the same I do. Great site though.. thanks to Daniel!

        Myself I just count the first round than pick up the muck count and reset it right there. If the RC = +5 or bigger I make big bet I manipulate the boxes sometimes(2.7x cards dealt per player and use 6 boxes inclusive the dealer = 16.2 cards on average). If 7 boxes are played I use the entre muck count 16 cards+ and use that as the new count. I am resetting every round with a full table. If the 2nd round is an advantage bet than I will play that round and use the count from that round only. You could use 2 boxes to than play 2 rounds and if the count hit plus 5 or bigger play 5 boxes making it exactly 16.2 cards on average. Hope to hear from you.

      • James said, on April 4, 2014 at 1:22 am

        Hi Daniel,

        Assumed your 16 cards in the chute is correct.

        Here are my questions :-

        Say after 1st round, the total cards used for that round is 9 cards, left 7 cards in the chute. Is the chute fill up immediately with additional cards to make the total to become 16 cards again before delaer put back the used card(9 cards) to the machine ?


        after finish dealt all 16 cards then the chute only fill with a new block of 16 cards at once ?


      • Daniel said, on April 7, 2014 at 5:32 am

        Frankly I have no idea about this…

  17. Stuart said, on December 29, 2013 at 8:40 am

    Quote= ” 17 blackjack rules (3:2 BJ, late surrender, re-split Aces 3-times, double-after-split). My blackjack analyzer calculates the ideal EV for these rules at -0.445% for 6 decks. I deal with the following rules. 6 deck, early surrender, only hard doubling allowed( 9-10-11) , 3:2 BJ Do I have more or less than .445% here with this rules.

  18. willis said, on March 31, 2014 at 5:58 pm

    Hi Stephen how.are U still maintaining this blog?u did a good job I tried and been very successful for 11 session in a row. The last 5 session been up and down.Meaning I ended with more crappy hands (hard 13-16) while dealer face up is between (7-A). My question,is this simply a variance (the 0 22% house edge) OR is it rather the CSM mechanism been altered, is it possible?

  19. Tisso said, on April 12, 2014 at 6:45 am

    Stephen, hats off 2u! My further analysis shows that you CAN do something to attain EV > 0 slightly, when unsusp sitting out half the games.

    Few questions:
    *How does house advantage change per muck size. Eg this analysis was n=16. What does n vs EV look like for a given count ( which begs a 3D graph n vs EV vs count)? I’d expect dEV/dn >0.
    *If muck is collected, can we add-on the previous muck count (if <=16 cards)? Tying back to question1.
    *What is late surrender, and does dealer stand on A6 in the analysis above.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s


Get every new post delivered to your Inbox.

Join 87 other followers

%d bloggers like this: