Card Counting, Explained

After reading this you will understand how the Hi-Lo count works, how to turn a running count into a bet size, and why even a genuine 1% edge can lose money for months at a time.

Card counting is not memorising which cards have left the shoe. It is a single running tally that tracks whether the remaining cards favour you or the dealer. When the shoe is rich in tens and aces, you get more blackjacks (paid 3:2), your double downs land more often, and the dealer busts more when they must hit a stiff hand. You bet more in those moments and less otherwise. That is the whole edge.

Here is the hook. A player flat-betting perfect basic strategy at a good table faces a house edge near 0.5%. A counter using the same strategy plus a 1-to-15 bet spread can flip that into a player edge of roughly 1%. The swing comes almost entirely from betting big when the count is high, not from playing hands differently.

What the Hi-Lo count measures

Hi-Lo assigns a value to each card as it appears:

Low cards (2, 3, 4, 5, 6): Each counts +1. When these leave, the shoe gets better for you.
Neutral cards (7, 8, 9): Each counts 0. They barely move the edge.
High cards (10, J, Q, K, A): Each counts -1. When these leave, the shoe gets worse.

You add these up as cards are dealt. This total is the running count. A full 52-card deck contains twenty +1 cards and twenty -1 cards, so a complete deck sums to exactly zero. That balance is what makes the system work: a positive count means the low cards came out early, leaving a deck disproportionately full of high cards.

The count only tells you about cards you have not yet seen. A running count of +8 after one deck of a six-deck shoe is far more meaningful than +8 with five decks still to play, because the same eight-card surplus is concentrated into fewer remaining cards.

From running count to true count

The running count alone does not size your bet. You must divide by how many decks remain. That gives the true count, and it is the number that actually predicts your edge.

\text{TC} = \frac{\text{running count}}{\text{decks remaining}}

Here TC is the true count, the running count is your mental tally, and decks remaining is estimated from the discard tray: look at how thick the stack of played cards is and subtract from the shoe size. Each true point is worth about 0.5% of edge. Starting from a 0.5% house disadvantage, you break even near a true count of +1 and gain roughly half a percent for every point above that.

\text{edge} \approx 0.5\% \times (\text{TC} - 1)

So at \text{TC} = 3 your edge is about 0.5% × 2 = 1%. At \text{TC} = 5 it is about 2%. This linear rule is an approximation, but it is close enough to drive real betting decisions.

Sizing your bets with a spread

A bet spread is the ratio between your smallest and largest bet. A 1-to-15 spread means your top bet is fifteen times your bottom bet. You bet the minimum when the count is neutral or negative, and you scale up as the true count rises. A common schedule with a $10 unit:

Example bet ramp for a 1-to-15 spread, $10 unit
True count	Units	Bet	Approx edge
≤ +1	1	$10	−0.5% to 0%
+2	4	$40	+0.5%
+3	8	$80	+1.0%
+4	12	$120	+1.5%
≥ +5	15	$150	+2.0%+

You wager small money into your losing situations and large money into your winning ones. That asymmetry is where the edge lives. Widening the spread raises your edge but also raises variance and makes you conspicuous to the pit.

Deviations: the Illustrious 18

Basic strategy is computed for a neutral deck. When the count is extreme, a few plays change. The best-known set is the Illustrious 18, the eighteen index plays that capture most of the value from strategy deviation. The single most valuable one: take insurance when the true count is +3 or higher. Normally insurance is a bad bet, but a ten-rich deck makes the dealer's hole card more likely to be a ten.

Learn the bet spread first. Perfect basic strategy plus a good spread earns most of the available edge. The Illustrious 18 add roughly 0.1 to 0.2 percentage points on top. Worth having, but not where the money starts.

A worked example from the demo

Simulating a 1-to-15 spread over a 1% edge

Run the bankroll simulator with the defaults: a six-deck shoe, a 1-to-15 spread on a $10 unit, a $3,000 bankroll, and roughly 100 hands per hour. Assume an average player edge of 1% with a standard deviation per hand of about 3 times your average bet.

Your average bet across all counts is around $25 (most hands are at the $10 minimum, a few are large).
Hourly expected value: 100 hands × $25 × 1% = $25 per hour.
Hourly standard deviation: roughly √100 × $25 × 3 ≈ $750. Your hourly swing dwarfs your hourly profit by about 30 to 1.
Over 100 hours: expected profit $2,500, standard deviation √100 × $750 = $7,500.

Read that last line carefully. After 100 hours of correct play, your expected gain is $2,500, but a one-standard-deviation band runs from a $5,000 loss to a $10,000 gain. Winning is the long-run tendency, not the typical week.

The expected bankroll rises steadily to $5,500, but the true outcome scatters widely around this line. Real equity curves cross below the starting point often.

Reading the results and risk of ruin

Risk of ruin is the probability your bankroll hits zero before your edge rescues you. It falls fast as your bankroll grows relative to your bet size. As a rough guide, a full-time counter wants a bankroll near 1,000 times their average bet to keep risk of ruin under about 5%.

\text{RoR} \approx e^{-2 \cdot \text{edge} \cdot \text{bankroll} / \sigma^2}

Here edge and \sigma (standard deviation) are both measured in the same bet units, and bankroll is in those units too. The exponential form is why doubling your bankroll roughly squares your survival: it does not halve your risk, it slashes it.

Risk of ruin drops sharply as the bankroll grows. At 1,000 average bets it is near 2%, at 300 it is near 30%.

Common mistakes

The biggest error is under-bankrolling. A 1% edge with a wide spread routinely swings thousands of dollars. If you play $150 top bets on a $1,500 bankroll, your risk of ruin is close to a coin flip, edge or no edge.

Other frequent mistakes: forgetting to divide by decks remaining and betting off the raw running count, which overbets early in the shoe; playing at a table where the dealer shuffles with two decks still in the tray, which destroys deep-count value; and widening the spread so aggressively that the pit backs you off within an hour. Counting is legal, but casinos are private businesses and can refuse your action.

Related tools

To set the baseline the count fights against, use the Blackjack House Edge tool to compute the exact house edge from your table rules, then the Blackjack Strategy & EV Explorer for the correct basic strategy chart. For an honest look at what non-countable games cost, the Expected Loss Calculator and the Slot Session Calculator show the drain over a session. If you want to compare against games where no legal edge exists, see Roulette Odds, Craps Odds and Lottery Odds.

Frequently asked questions

How much can you actually make counting cards?

Edges run from about 0.5% to 1.5% depending on rules, penetration and spread. On the worked example above, that was about $25 per hour with $750 hourly swings. It is a slow grind, not a fast income, and heat from the casino usually caps how long you can keep it going.

Do you have to memorise every card played?

No. You only keep one number, the running count, updated by +1, 0 or -1 per card. The skill is doing it fast and quietly while playing normally.

Is card counting illegal?

No. Using your own mind to track cards is legal. Casinos may still shuffle early, restrict your bets, or ask you to leave. Counting devices are a different matter and are illegal in many places.

Why does a winning system lose for months?

Because variance dominates the short run. In the 100-hour example the standard deviation ($7,500) was three times the expected profit ($2,500). A losing stretch is not a sign the math is wrong, it is the expected texture of a 1% edge.

Does a bigger spread always help?

It raises your edge and your hourly EV, but it also raises variance and risk of ruin, and it makes you obvious to the pit. A 1-to-15 spread is a practical balance for a six-deck game; wider spreads need both a bigger bankroll and more discretion.

Try it yourself in the tool