A blackjack training game with dynamic Basic Strategy charts

Hey everyone,

I’ve been working on a blackjack training game called **Blackjack Training**, and it’s finally finished and available on Steam.

The project started during my studies and has grown quite a bit over the years. The goal was to build something specifically for practicing blackjack strategy and card counting rather than simply playing a casino game.

It includes:

* Basic Strategy training
* Hi-Lo deviations
* True Count
* Instant feedback when you make an incorrect decision
* Configurable casino rules and deck numbers
* Surrender and other rule variations
* Statistics and performance tracking
* A 3D casino environment

One of the things I wanted to do differently is make the *Basic Strategy chart dynamic with card-counting deviations*.

The chart doesn't simply show a static Basic Strategy decision. When playing with deviations enabled, the recommended action changes according to the current True Count and the selected rules.

For example, with a 16 vs. 10, the chart can show "Hit" at a True Count below 0 or "Stand" at 0 or higher, rather than always displaying the basic-strategy decision.

So the chart itself reflects the deviation you're supposed to make at the current count.

The game page is now online on Steam:

/app/4934240/Blackjack_Training/

Since this community is specifically interested in blackjack, I thought some of you might find it interesting.

Thanks for taking a look!

That dynamic strategy chart sounds like the most useful part to me. A lot of blackjack trainers are fine for memorizing basic strategy, but once you start working with true count deviations you usually end up bouncing between the game and a separate chart.

Yeah, that was actually one of the main things I wanted to get right.

Instead of having to constantly switch between the game and a separate deviation chart, the chart updates based on the current True Count and the selected rules.

And if you make a wrong decision, the game immediately tells you what the correct decision should have been, so you can keep practicing without having to stop and look up the index separately.

Quote: Haudi

For example, with a 16 vs. 10, the chart can show "Hit" at a True Count below 0 or "Stand" at 0 or higher, rather than always displaying the basic-strategy decision.

Regarding the hit/stand decision for this particular hand of 16 vs. 10, it should be “hit” when a hi-Lo True count is less than or equal to 0 but “stand” when it is greater than zero?

Another thing is why bother using Hi-Lo? You just use computer-perfect decisions for every situation, just like Wizard’s blackjack calculator.

Computer perfect is impossible to use in real life play.

It is possible with an iPhone app. Just like the video poker app in another thread.

Also, using the Hi-Lo true count to make the hit-stand decision is too rough for winning the hand 16 vs. 10. Players should use the running count for this hand.

I haven't tried the app, but perhaps when an index play changes the BS play, the app could inform the player. For example, on 16 vs. 10 at a TC of +3, the chart could say "Stand. Hit for TC<= 0" or some other way to notify the user that this is an Index play.

Dog Hand

Yes, exactly. For 16 vs. 10, the decision is to hit at a True Count of 0 or below, and stand when the True Count is above 0.

As for why I use Hi-Lo: the game is specifically meant to train card counting, so I wanted to use an actual counting system rather than just giving the mathematically perfect decision every time. The point is to practice the process of counting, keeping track of the True Count, and then applying the appropriate deviations.

You can certainly use computer-perfect decisions, but that wouldn't really train you to count cards.

That makes sense. I was using the True Count because the game supports different numbers of decks, so I wanted the deviations to scale with the remaining decks rather than relying on a fixed running count.

Would you recommend using the running count specifically for 16 vs. 10 regardless of the number of decks, or would the index need to change depending on the number of decks and rules?

I see what you mean. For regular Basic Strategy training, the feedback currently just shows when your decision is wrong and what the correct Basic Strategy decision would have been.

I do have this kind of feedback for the Insurance bet, though. It tells you whether you should take Insurance or not based on the current count. I could definitely see the same approach working for index plays as well.

Quote: Haudi

That makes sense. I was using the True Count because the game supports different numbers of decks, so I wanted the deviations to scale with the remaining decks rather than relying on a fixed running count.

Would you recommend using the running count specifically for 16 vs. 10 regardless of the number of decks, or would the index need to change depending on the number of decks and rules?
link to original post

I would recommend using the running count (RC) for 16 vs. 10, because true count (TC) numbers are not accurate enough. For example, RC=+3, remaining decks=4, so TC=+0.75. Do you round or truncate it to an integer TC number? That’s a difference of one between +1 and 0.

Also, this becomes a math puzzle here. When TC=0 exactly, is it hit or stand? This hand of 16 vs. 10 is very composition dependent too.

That’s a good point. I currently use the True Count for the deviations, and I do round it to an integer for the decision thresholds.

I can see how that becomes problematic for 16 vs. 10, especially around the 0 index and with different numbers of decks remaining. The composition-dependent aspect is also something I need to take into account.

I’ll definitely look into using the appropriate Running Count index for 16 vs. 10 rather than treating it like a standard TC-based deviation. Thanks for pointing that out!

The use of the Hi-Lo Count for making an index play on 16 vs.10 is very inefficient because it counts the removal of a six(6) as the same as the removal of a five(5). Here are the Effect of Removal Coefficients on EV for the HIT/Stand decision for the "16 vs. 10"; given a 8D, H17, DAS game.

The 16v10 decision is the most important and most frequently occurring "close call" decision in a No Surrender BJ game. It is very unfortunate that the counts designed for changing one's bet size will lead you to make the wrong decision on '16vT' whenever a relatively high fraction of 6's are counted.

Having the chart adjust to the current situation could make practice more realistic. It also helps players understand why a particular move is recommended instead of simply memorizing a table. The key would be keeping the interface simple enough to use quickly.

Exactly, that was one of the main ideas behind it. I wanted the chart to be useful while actually playing without getting in the way.

You can toggle it in and out with the Tab key, so it’s there when you need it and out of the way when you don’t. When it’s open, you can still clearly see your hand and the table, so it doesn’t interfere with the gameplay.

I have a screenshot of it on the store page.

The statistics should also help players see which situations they keep getting wrong. Having different rules and deck settings is a nice touch for more realistic training. It looks like a useful way to practice without simply relying on a static chart.