I have had quite a few requests for information about my past work (many years ago) as a blackjack player, card counter, and creator of software to "help" with blackjack strategy while playing.
I started card counting in high school (early 1970's) because I had become fascinated with probability theory. Edward Thorp's "Beat The Dealer" provided a mathematically consistent and successful card counting system. At that time, Las Vegas was not concerned with card counters - the very few people who did it mostly were not successful. As in daytrading (what I'm doing now - see here), you could learn the theory but unless you executed it almost perfectly (minimal mistakes, minimal "breaking the rules"), you would lose money (or at best, average about break even with Basic Strategy). I did a lot of reading and research over several years, and practiced a lot.
Living in Colorado, I worked as a blackjack dealer for several years as a "fun casino." Fortunately, this place was VERY strict about maintaining reality. So, although it was for fun-money ("Bustop Bux"), we had a pit boss (I was the pit for about a year), 10 blackjack tables, and rules enforcement - cheating was not allowed, either by the player or the dealer (fired a couple for that!), and the pace of the game was good. This was an ideal place for me to continue card counting practice - counting cards while dealing, training others to count cards, etc. Since it was not for real money, we encouraged card counting - many people came to play and practice before their Vegas trips.
While dealing blackjack, I became acquainted with a few other professional card counters who were also gadget-heads. We came with an idea of creating a computer that would fit in your pocket that would help to automate the play of blackjack by tracking all of the cards, and making betting and play decisions, with less human error for incorrect play (reducing human error to incorrect inputting and reading of output). Some details about this HP-41C based blackjack computer are below. We worked for several months on it (I wrote the software), and eventually went to Vegas (team of 6, if I remember correctly). We had a great time - collected some of the "big" chips (yellow and purple were $500 and $1000 at the Tropicana, I think) - and had a few adventures.
My favorite "close call" was when a teammate and I were at the same table in Westward Ho (next to Circus Circus). Our computers were in our shirt pockets (with a jacket over), upside down so that they could be removed quickly, and would automatically disconnect from the "wiring harness" that went to the toe switches and output LEDs in the watch. For some reason, his wires had gotten bunched up, and were visible alongside the computer. The dealer noticed this and started nodding her head at him, trying to signal him that he was "exposed." At this point in time, Vegas was becoming aware of blackjack computer play, but had not yet made it illegal. However, they were NOT happy with people who used computers ("break your knees" comes to mind, but that might just be romanticizing the culture... or not!). The pit had somehow become alerted to my teammate's computer (eye in the sky, or possibly noticing the dealer signalling? we never found out), and I had seen the wires, and pointedly suggested that he "take a walk". He got up from the table, pocketed the chips, and walked out the door, just as I saw a couple of bouncer-types heading his way. He made it - and the casino did not pursue him - maybe they weren't sure enough. However, I remained at the table, and asked the dealer "what happened?" The dealer nicely explained to me how "some people are using blackjack computers" - and that they had been warned to be on the lookout. I played for another five or ten minutes, then left.
Oh, also - I cleaned out the "brown" chips at a table at the Maxim once. These were $25 - most casinos use green for $25. They had to come refill the chip tray! They did not get suspicious because I didn't have a big pile of chips in front of me. Whenever I would get a little bit ahead, I'd pocket three or four chips. I'm not sure if I was just good at that, or - more likely - that the casinos were accustomed to people pocketing chips now and then, and they might not have noticed ALL of my pocketing.
I don't have any publishable pictures of my old blackjack computer, sorry. It was actually an internally modified HP 41-C "calculator" (actually a programmable handheld computer in a calculator form factor). You can probably find pictures of that calculator - externally it looks the same.
Internally, we added an EPROM which contained 8K of card counting capabilities and input/output routines. This EPROM was larger than one of the plug-in modules in the end of the HP 41 - so we shoehorned it into the battery pack in the back of the calculator. Of course, that means we didn't have room for the battery... hehe... so... We made a new battery pack in the end of the HP 41 - taking up two of the plug-in memory slots.
Also, under the new battery pack in the end (the HP 41 has 4 "slots" for add-on memory and program packs), we installed a contact strip. This contact strip connected the calculator to the "output" (which was four LED's installed inside a watch with small fiber-optic fibers to make the LED's visible in the wristband of the watch, but only at a very tight angle) and to the "input" (which was four toe switches - two per boot - one over each big toe, one under each big toe). We installed this contact strip so that we could pull the calculator out of our shirt pocket, disconnecting it from the input and output - in case "security" got suspicious, and asked to see the calculator.
We built a wiring harness that went inside our clothes - the boots had small plugs to plug into the wiring harness, as did the watch.
Softwarewise (my stuff!): The HP 41-C supported four plug-in memory/program packs. The rom-packs were usually 4-K programs each. I wrote two full 4-K modules for our BJ computer: The primary one ("TurboMax") was the "playing" rom which took input of cards seen, etc., and provided output of bet amount, play strategy, etc. The secondary rom-pack ("UltraMax") added blackjack play-strategy calculations - so you could evaluate a counting system directly, including correlation calculations, playing efficiency, etc. UltraMax included Arnold Snyder's "Blackjack Formula" - which was why it was never made publicly available. TurboMax was actually available publicly (AFTER we couldn't use it anymore since the Las Vegas laws had been written to prevent "blackjack computers") as an eprom set for the HP 41-C.
Here is a quick overview of TurboMax (BTW - I wrote TurboMax and UltraMax using HP 41-C microcode - similar to assembly language):
Inputs were done with four toe switches - each signal being a sequence of two toe switches on or off. By doing this, we could signal a value from 0 to 15 (hexadecimal 0 to F). Input of cards was a single value (1 for ace, 2 for 2, ... 9 for 9, and hex 'A' for ten) since that was the most common input needed. Hex 'B' indicated "begin a new hand - no shuffle," and Hex 'C' requested a playing-decision (hit/stand/double/split) to be output to the LEDs in the watch. The other values (0, and hex 'D' thru 'F') were used to signal other game activities with a two-hex-digit code. Hex '0' was used to set rules variations for TurboMax, so that it would use the right counting strategy. For example, Hex '0' followed by Hex '7' told TurboMax to set up for "standard" Reno/Tahoe rules, Hex 'F' followed by Hex '1' thru '8' told TurboMax how many decks to play with. (etc)
Outputs were done on 4 LEDs - and interpreted in hexadecimal from 0 to 15. After a "new hand" signal from the toe switches, the LEDs showed the recommended amount to bet in units (1 thru 16) (in most cases, you wouldn't swing your bet THAT much, because the pit would get suspicious), and during play, the LEDs showed hit/stand, double, split, and surrender. Also, some of the input codes would allow you to query TurboMax and display the results on the LEDs (Hex 'E' '0' combination showed the unseen cards count, Hex 'E' '1' showed how many aces left, etc.).
TurboMax used 16 different card counting systems simultaneously - using the "best" system for each particular play. It used Wong Halves, Braun +/-, one of Griffin's systems, Uston APC, Revere APC 73, Tulcea Adjusted Improved, Snyder Zencount, HiOpt 1, DHM/Gordon, Revere +/-, Uston +/-, and four special systems I worked out (using Peter Griffin's book) for specific plays (such as playing 12 against a 2-6). I worked out these four special systems because I could get a better win correlation than any of the "regular" count systems - and these were for "critical" plays. (The other three special systems were for 16 vs. 8-T, 11 vs. 7-9, and 15 vs 9-T).
It's been a LONG time since I've discussed this - kind of fun to reminisce about it :-)