Blackjack Insider Newsletter, March 2004, #50

As many of our fine readers now know, Henry Tamburin (editor of BJI) and myself (Dan Pronovost, production editor of BJI) teamed up to develop a new card counting system for Blackjack, called Speed Count. The result of two years of intense research and development is a new card counting system, different from anything before, that is many times easier to learn and master compared to popular existing systems. It brings the potential of blackjack card counting to the masses of gamblers who play without a positive edge, or have tried counting and failed.

Not surprisingly, there has been a lot of buzz about Speed Count. Since the system is 'hush-hush', with few details revealed, there are a lot of skeptics and doubters! Fair enough... this article gives as much information as we can on Speed Count, providing some important simulation performance metrics in comparison to other card counting systems.

From years of selling blackjack training software and working with blackjack players, I've learned that the hardest skill to master to become a proficient blackjack card counter is keeping track of the running count as cards are dealt. Adding and subtracting 'on-the-fly' sounds easy, until you're chasing a modest to fast dealer at a full table, and dealing with negative integer arithmetic! Although you may think you're accurately tracking the count, even small errors can be costly and cause the slim advantage you've gained to turn into a loss. If you're not betting properly with the advantage because you've tracked the count wrong, then you are most likely playing at a loss without knowing it.

Speed Count was designed from scratch to remove this barrier, and many more. There's no on-the-fly subtraction, no negative integer arithmetic, no multiplication, no division, no estimating decks or shoes dealt or remaining, and players track 1/2 or less information of other count systems such as High-Low or Knock-Out. By leveraging some consistent mathematical properties unique to blackjack, you can have a system that only takes a few hours to master, virtually eliminates count errors, yet delivers most of the advantage possible with popular count systems.

As such, Speed Count requires far less player attention during blackjack rounds. This allows players to concentrate more on camouflage: talking to other players, observing the pit bosses and casino environment, and generally appearing as a more average gambler. Speed Count is not as taxing on players as more advanced systems such as High-Low and Knock-Out, and lessens personal exhaustion factors after long play sessions.

Speed Count is for average blackjack players who have been leery of learning card counting, or who have tried already and failed. It's easy to learn and use, yet delivers most of the advantage of good count systems. For expert players already playing a high-end system well without errors, Speed Count will not provide further advantage, but could provide good cover since the casinos will not be familiar with it's different characteristics and playing strategy.

Speed Count is not a 'fake' system, using progression betting or any other blackjack voodoo that is commonly popularized. It is a card counting system, in that the player tracks certain concrete and non-subjective information as the hands are played. This information is used to maintain the 'speed count', which ultimately determines how you play.

Speed Count delivers about 75% percent of the performance advantage of popular count systems such as High-Low (without play indices), depending on the game and rules. We've completed exhaustive computer simulations to back up the power of Speed Count. We've even made sure we compared Speed Count fairly by adjusting the bet spreads so that the metrics of the systems were comparable (for the statisticians out there, we adjusted the bet spreads to ensure that the average bet size and standard deviation were very close).

Included below is some simulation data from Blackjack Audit for a couple popular blackjack games. We ran 100 million rounds of blackjack for each entry in the table below. Here is a summary of the two games we simulated:

• allow re-splitting of aces, no hitting split aces
• maximum of two splits (3 hands), no surrender
• American hole card rules (peak on 10 and ace), blackjack pays 3 to 2

In the tables, we show four simulation metrics for each system entry:

• Win rate: the average profit in bet units per round. This factors in all bets made in the system (splits, doubles, and insurance). Multiply by 500 ($5 unit bet size, 100 hands per hour) to get the expected win rate in dollars per hour.
• Average bet size: the average total bet size per round, in dollars. This includes all bets made in the system (splits, doubles and insurance).
• Standard deviation: this is the square root of the variance, in bet units. This value is a reflection of the risk (or variation) in profit from round to round. A higher value means the system will result in more fluctuations in your round-to-round bankroll.
• Expectation: this is average percentage profit a player can expect from each bet made with the system: players can multiply this percentage by and (and all) bets made, to get the average return they can expect to see. For our simulations, it is calculated as the total profit made across all rounds, divided by the sum of all wagers made. All bets made are included, such as splits, doubles, and insurance.

We compared Speed Count to the High-Low count system in a few different configurations:

• The bet spreads in 'Speed Count - Regular' are set to provide similar performance metrics to common High-Low bet spreads (see below).
• · 'Speed Count - Regular' does not use any insurance play indices (insurance is always declined).

• The wonging and hand spreading values were set to have similar corresponding distributions as with 'Speed Count - Aggressive' (i.e. the same 6% to 9% occurrence rate as with Speed Count Aggressive, for two and six decks). But it should be noted that the addition of wonging and hand spreading creates significant performance differences in the respective systems, since the skipped and spread hands proportions are only approximately the same.

The last two rows of the table shows the comparative performance of Speed Count to High-Low (rows A and B). Depending on the game, Speed Count is about 75% as effective as High-Low without play indices. Speed Count performs very well in particular with double deck games.

We need to address another criticism raised: that we did not use an optimal bet spread in our comparison, since our method meant 'fudging' the bet ramp (but not the bet range, which was kept the same). To address this point, we need to explain a few more mathematical aspects of blackjack for our novice readers…

All card counting systems that actually work, whether they are balanced like High-Low or unbalanced like Knock-Out, have a bet pivot point. This is a value for the tracking metric under which the count system is saying the casino has the statistical advantage, and over which the player has the advantage. The simplicity of card counting is that all functional count systems simply have you bet more above the pivot, and less below it.

Simulations and math have shown that this solid foundation of card counting is valid and correct: either the house or the player has the edge depending on the distribution of cards remaining in the shoe or pack, and count system to varying degrees attempt to measure this advantage, and bet accordingly.

The ideal bet pivot value is highly dependent on the count system, and may also vary with the game, rules, bet spread, and other factors. In the above High-Low computer analysis, we used the following bet spreads:

These bet spreads (1 to 4 for double deck, and 1 to 8 for six deck) were chosen since they represent rational spreads that casinos will tolerate, and are used in practice by many advantage players. Again, our analysis was intended to be understandable by novice players. We did not want to confuse the issue with bet spreads that were not reasonable or attainable in practice. The table above is based roughly on the spreads published in Wong's Professional Blackjack (pg. 18, benchmark rules), and other popular references to High-Low such as Blackjack: Take the Money and Run by Henry Tamburin.

The table above implies a bet pivot at a true count of one for the six deck game, and zero for the double deck game. But are these in fact the correct bet pivots for High-Low? We will not be maximizing our advantage if we increase our bet below the bet pivot. Interestingly, it turns out, the actual best bet pivot for High-Low is between zero and one for the double deck game, raising the question of whether betting 2 units 'off the top' is in fact best.

Novice players often see this problem, since they know you don't have an advantage at the start of a shoe or pack, yet the above bet spread calls for betting two units. The problem is that High-Low requires division and rounding (or truncation) to determine the true count, since calculating the exact true count in a casino in your head is not practical. So, unless you are a human calculator, you have to choose either zero or one as the bet pivot and use this as the true count at which you increase from your minimum bet.

We ran simulations in the double deck game with High-Low (no play indices) and a bet spread that jumps to two units at a true count of one (instead of zero), and it does turn out to perform better. We didn't do that in our original published comparison, but let's do it now:

Yes. Immediately after announcing the Golden Touch Blackjack course, we started the process of getting independent reviews of Speed Count. Each reviewer was given a reviewer's kit, that included a complete description of Speed Count, a background into it's development and mathematics, and all of our simulation data. Here is what some of our reviewers have said. Reviewers were free to say or write whatever they wanted about Speed Count, but required by agreement to not disclose any functional or operational details about the system.

We asked Dr. Don Catlin, an independent gaming industry consultant and professor of Mathematics at University of Massachusetts, to review Speed Count. Mr. Catlin used his own blackjack simulator software to test Speed Count and verify our performance data. Here are his comments:

Comments by Dr. Don Catlin:

"Speed Count is the easiest card counting system I have encountered. I designed and ran a simulation of the Regular Speed Count using a 6 deck Blackjack game with doubling after splitting, dealer hits the soft 17, late surrender, split up to three times (total of four hands), Aces are split once and receive one card each. My result was that the players edge per game was 1.014% and was 0.33% per unit wagered; this was based on 1 billion hands."

Dr. Catlin has a B.S. in Mechanical Engineering and an M.A. in Mathematics, both from Penn State University. He has have a Ph.D. in Mathematics from the University of Florida. He was a Professor of Mathematics and Statistics at the University of Massachusetts from 1965 to 2001. Since 1995 he has been involved in gaming analysis and writes the monthly mathematics column for the web site www.scoblete.com. In September of last year he published a soft cover book entitled The Lottery Book, The Truth Behind the Numbers.

Comments by Michael Shackleford:

This issue of Blackjack Insider brought to you by...