by the Green Baize Vampire

The consensus appears to be amongst the experts that side-counts and multi-level counts provide an unneccessarily high degree of error for the comparitively small degree of gain.

In the rge21 reference library Arnold Synder points out that a gain of .1 % from a side-count may be an effective increase in win-rate of 10% for the card-counter who plays with an edge of 1%.

Moreover, the optimal bettor will have a higher win rate with an edge of 1.1% than the .1 increase implies because he can bet slightly more money with the same risk of ruin.

But,I thought, more importantly, because he wins faster, a player with a 1.1% edge will win more money after the same fixed number of hands than the player with a 1% edge. He will then have an added advantage of a larger bankroll to bet proportionally from. Additionally, his bankroll will get larger and larger proportionally to that of the 1% player.

I tested my understanding of this by writing a simulation where player A had a 1% advantage and player B had a 1.05% advantage (a rough approximation of two counters in good shoe game,one using a side-count. one without).

Each player bet in accordance with pure Kelly, ie they resized their bets with every wager. They each played a 100000 games.

The results surprised me.At the end of the sim (I ran it several times) Player B had more than 4 times as much money as Player A. When I gave player B a 1.2% edge he won several hundred times faster than A. You can get approximately .2% extra with an ace-count in deeply-dealt single-deck games.

Now, I appreciate my simple "biased coin-toss" program does'nt approximate real blackjack very well. And it's not possible for humans to resize their bets precisely at every stage. But clearly this is a significant difference.

Personally I would go to almost any lengths to win 4 times as much money during my lifetime (100000 hands being a reasonable number of hands a counter might play in a lifetime).

It was pointed out to me that both players would bump against the table limits before the conclusion of each sim, so I began again, setting the target for each player to double their bankroll six times (a reasonable long-term ambition for a card-counter). I discovered Player B achieved this goal 25% faster than player A, due to the tiny .05% extra advantage.

So, do I recommend the use of higher-level count systems?

Of course, it is not quite that simple. The traditional justification for using a simpler system is that it reduces error, which is not one I subscribe to, because human error results as much from tedium as it does difficulty (to be on the edge of your ability results in the fewest errors).

However, there are other important considerations. Are you fatigued more easily playing a higher-level system? Do you need to spend more time practicing and less playing, does your playing speed go down? These are all factors which can reduce your potential win rate. You have to answer these questions honestly and determine what is best for you.