Card Counting: What It Actually Does, and Why It Barely Applies Online
Card counting has the best public relations of any technique in any game. Films present it as a superpower: a savant tracks the deck, the numbers align, and money follows. The reality is quieter, slower and considerably more boring - and understanding what counting actually does is genuinely useful even if you never do it.
What a count is measuring
The idea is simple. High cards - tens and aces - favour the player. Low cards favour the dealer.
Why? Because the dealer must hit until 17 and has no choices. A deck rich in tens means the dealer busts more often when they are forced to draw to a stiff hand. It also means you get more blackjacks, and a blackjack pays 3:2 while the dealer's blackjack only takes your original bet.
So counting is not memorising cards. It is tracking a single running number that estimates whether what remains is rich in high cards or low ones. The common system assigns +1 to low cards, -1 to tens and aces, 0 to the middle, and you keep a running total, dividing by the decks remaining to get a "true count".
What it actually buys you
Here is the part the films leave out. Perfect basic strategy at a good table leaves the house with an edge of roughly half a percent. Perfect counting, executed flawlessly, might swing that to a player edge of around half a percent to one percent.
That is the whole prize. Not a reversal of fortune - a narrow edge, realised only over an enormous number of hands, with swings large enough that being ahead in any given session is close to a coin flip.
And the edge only materialises if you vary your bet. Counting without changing your wager achieves almost nothing, because knowing the deck is favourable is useless if you bet the same amount regardless. The count tells you when to put more money out, and that bet variation is both where the edge comes from and what makes it conspicuous.
Why it barely applies online
For a site like this one, and for most online blackjack, counting is simply not a thing - and the reason is worth stating plainly rather than treating as a trick.
The deck is typically reshuffled every hand. A count carries information from one hand to the next; if the shoe is reset each time, there is no information to carry. The running count starts at zero every hand and stays meaningless.
That is not a restriction imposed to stop anyone. It is simply what happens when dealing is done by a shuffle function rather than a physical shoe. The randomness is re-drawn each hand, so there is no depletion to track.
Which means that here, basic strategy is not just the best available approach - it is the complete approach. There is nothing above it.
The useful idea underneath
Even where counting does not apply, the concept behind it is one of the most transferable ideas in games: as cards leave, what remains becomes more predictable.
That principle is the entire skill in the trick-taking family. In a game of Hearts, every card played is public and permanent, so by the endgame a good player is not guessing at all - the unseen cards have shrunk to a set small enough to reason about. The bidding version makes it explicit: play Spades online and your contract depends directly on counting what has gone.
And it has an exact analogue outside cards. Modern falling-block games do not deal pieces at random - they deal from a shuffled bag containing one of each. In Tetris, seeing four pieces from the current bag tells you exactly which three remain, which is counting in the purest possible form. The bag depletes; the information grows.
Blackjack in a continuously shuffled game is the rare case where that principle is switched off. Recognising which games deplete and which reset is a genuinely useful lens.
The variants where counting matters more or less
Deck penetration - how far into the shoe the dealer goes before reshuffling - matters more than the counting system. Deep penetration gives the count time to become extreme; shallow penetration kills it.
Among the variants here, Single Deck is the one where a count would move fastest, because each card removed is a large fraction of what remains. Double Deck is next. Multi-deck shoes dilute the effect, which is precisely why casinos moved to them.
Spanish 21 is an interesting case: removing the tens from the deck changes the base odds so much that the compensating bonus payouts are doing heavy lifting, and the count behaves differently because the high-card population is already altered.
Double Exposure goes the other way entirely - you see both dealer cards, so the information counting would give you is partly handed over directly, and the game compensates elsewhere.
What to do instead
If you want to actually play better here, the leverage is not in exotic technique. In rough order:
- Play the chart without deviation, especially on the three or four hands where it feels wrong.
- Never take insurance. It is the one bet that is straightforwardly bad without a count behind it.
- Know the rules of the variant you are playing. Payout structure moves the edge more than any play decision you will make.
- Treat the session length as the real variable. Variance dominates short sessions regardless of skill.
That last point is the honest one. Accepting that outcome and decision quality are different things is the hardest discipline in any game with randomness - and it is the same acceptance required when you play Yahtzee online and a correctly allocated scorecard still loses to someone who rolled better.
Where to practise
Open the strategy chart and play a session of Single Deck, then the same number of hands of Spanish 21 without adjusting your approach. The difference in results is a good demonstration of how much the ruleset, rather than the technique, determines the outcome.
The rules hub and strategy section cover the variants, the glossary names the terms, and the payout FAQ explains why 3:2 versus 6:5 matters so much. Everything on this site is play money - there is no real gambling here.
For why players who own the chart still deviate from it, see the companion piece.