How Card Counting Works: A Plain-English Guide to Hi-Lo and Wong Halves

Card counting is one of the most talked-about and least understood ideas in casino games. Films make it look like a superpower; in reality it is a disciplined bookkeeping exercise grounded in simple probability. This guide explains what card counting actually is, how a running count and a true count differ, how the popular Hi-Lo system compares with the more precise Wong Halves system used in the NeuroMash trainer, and why counting is a far slower and more fragile edge than most people assume. It is written for education and curiosity — to help you understand the mathematics of blackjack — not as encouragement to gamble.

What card counting is (and is not)

Most casino games are independent trials. A roulette wheel has no memory: each spin is unaffected by the last. Blackjack is different because the cards are dealt from a finite shoe and, until the shoe is reshuffled, cards that have been played are gone. That means the composition of the remaining cards changes from hand to hand, and with it the player's expected result.

Some cards help the player when they remain in the shoe. Tens and aces make blackjacks (which usually pay 3:2) more likely, help double-downs succeed, and make the dealer, who must draw to 17, more likely to bust. Small cards such as 4s, 5s and 6s do the opposite: they help the dealer turn weak totals into made hands. Card counting is simply a way of keeping a rough, mental tally of whether the cards still to come are richer or poorer in high cards than a fresh shoe would be.

Counting does not mean memorising every card, it does not predict the next card, and it does not guarantee winning any particular hand, session, or even year. It shifts long-run averages by a small amount — often around one percent in favourable conditions — and only for a player who also plays near-perfect basic strategy.

Basic strategy comes first

Before counting matters at all, a player must know basic strategy: the mathematically best decision (hit, stand, double, split or surrender) for every combination of player hand and dealer up-card, given the table's rules. Basic strategy alone reduces the house edge in a typical shoe game to roughly half a percent. Counting is layered on top of it. Someone who counts perfectly but plays hands by instinct will still lose more than a non-counter who plays basic strategy correctly. That is why the NeuroMash site pairs the counting trainer with a basic strategy solver that calculates the chart for the specific rules you choose.

The running count

Every counting system assigns a value, called a tag, to each card rank. As cards are revealed, you add the tags together. The cumulative total since the last shuffle is the running count. In balanced systems (both Hi-Lo and Wong Halves are balanced) all the tags in a full deck sum to zero, so a freshly shuffled shoe starts at zero and, if you counted every card correctly, the count returns to zero at the end.

A positive running count means more low cards than average have been seen, so the remaining cards are relatively rich in tens and aces. A negative running count means the opposite.

The true count

A running count of +6 means something very different with one deck left than with five decks left. Six extra high cards spread across 260 remaining cards barely changes the odds; concentrated into 52 cards it changes them a great deal. To compare situations fairly, counters convert the running count into a true count, a rate per deck:

True count = running count ÷ decks remaining
Example: running count +6 with about 3 decks left gives a true count of +2.

The true count is what drives decisions. Betting levels and the small number of strategy deviations counters use are all keyed to the true count, not the running count. Learning to do this division quickly and approximately, in your head, while also tracking new cards, is one of the hardest practical skills involved.

Estimating the decks remaining

Nobody counts the exact number of cards left. Instead, players glance at the discard tray and estimate how many decks have been used, then subtract from the total shoe size. Most people estimate to the nearest half deck, and many to the nearest whole deck in a six- or eight-deck shoe. Practising with a real stack of cards — seeing what one, two and three decks look like in a tray — is the usual way to build this skill. Small estimation errors matter less than you might think: dividing +6 by 2.5 instead of 3 yields a true count of 2.4 rather than 2, which rarely changes a decision. Consistency matters more than precision.

The Hi-Lo system

Hi-Lo, popularised in the 1960s and still the most widely taught system, uses only three values. It is a level-one system, meaning every tag is −1, 0 or +1.

Its strength is simplicity. Pairs of cards such as a 5 and a king cancel out, so experienced counters process cards in pairs and keep up with fast games. The cost is precision: Hi-Lo treats a 2 the same as a 5, even though removing a 5 from the shoe helps the player considerably more than removing a 2.

The Wong Halves system

Wong Halves, introduced by Stanford Wong, uses fractional values that track the true effect of each card more closely. It is a level-three system because its tags span from −1 to +1.5 in half steps.

Notice the differences from Hi-Lo. The 5, the card whose removal most helps the player, gets the biggest value. The 7 counts as a small positive and the 9 as a small negative. As a result, Wong Halves has a higher betting correlation — it measures how favourable the shoe is for betting purposes more accurately than Hi-Lo.

The fractions are awkward to add in a hurry, so many players double every value (making them +1, +2, +2, +3, +2, +1, 0, −1, −2, −2) to count in whole numbers, then halve the result before converting to a true count. The NeuroMash trainer lets you practise the native half values so you become comfortable with them directly.

Hi-Lo or Wong Halves: which should you learn?

For most learners Hi-Lo is easier and loses only a little accuracy. Wong Halves rewards people who enjoy mental arithmetic and want the most precise picture of the shoe. The extra accuracy only pays off if you can apply it without mistakes; a perfectly kept Hi-Lo count is worth more than an error-prone Wong Halves count. Either way, the concepts in this guide — running count, true count, deck estimation — are identical.

Bet spreads, explained conceptually

Knowing the shoe is favourable is only useful if something changes in response. The main lever is the size of the wager. In theory, a counter bets a minimum amount when the true count is neutral or negative and a larger amount when it is positive. The ratio between the smallest and largest bet is called the bet spread.

Conceptually, the trade-off is straightforward. A wider spread captures more of the advantage in good shoes but draws far more attention and greatly increases short-term swings. A narrow spread is quieter and steadier but produces an edge so small it can vanish entirely under slightly worse table rules. The theory of proportional betting (often discussed under the name Kelly criterion) suggests sizing bets relative to the edge and the size of the bankroll so that a run of losses cannot wipe a player out. In practice the variance in blackjack is enormous compared with the edge, which is why we don't recommend any particular betting scheme here, and the trainer's bet suggestions are illustrations of the idea, not advice.

Counters also sometimes alter playing decisions at certain true counts — for example taking insurance only when the true count is high, or standing on 16 against a dealer 10 when the shoe is rich in tens. These index plays are a refinement; the bulk of the value comes from betting and from basic strategy itself.

Why the edge is smaller than people think

• Rules matter. Games paying 6:5 for blackjack instead of 3:2, dealers hitting soft 17, and restrictions on doubling can erase most of a counter's advantage.
• Penetration matters. If the shoe is reshuffled after only half the cards are dealt, high true counts rarely occur.
• Continuous shuffling machines return cards to the shoe constantly, making counting useless.
• Variance is huge. Even a genuine advantage can take hundreds of thousands of hands to show up reliably. Long losing streaks are normal.
• Human error. Miscounts, fatigue, distractions and slow true-count conversion quietly eat the edge.

Is card counting legal?

In most jurisdictions, including the United States, Canada and the United Kingdom, counting cards using only your own mind is legal. It is not cheating: you are simply paying attention to cards that are dealt face up. Using a device, app or another person's signals to help count at a live table, however, is illegal in many places.

That said, casinos are private businesses. In many regions they may refuse service, ask a suspected counter to stop playing blackjack, restrict bet sizes, shuffle more often, or ban a player from the property. Rules differ by country, state and province, so anyone curious should check local laws. NeuroMash takes no position on visiting casinos and does not link to or promote any gambling operator, online or offline.

Why learn it at all? The educational value

Card counting is an excellent, concrete introduction to ideas that appear throughout mathematics, statistics and finance: conditional probability, expected value, sampling without replacement, variance, and the gap between a small statistical edge and a reliable result. Practising a count also exercises working memory, attention and fast mental arithmetic — which is what the NeuroMash trainer is designed for. You can benefit from all of this without ever placing a bet.

How to practise with the NeuroMash trainer

1. Learn the Wong Halves values from the table above until you can name each tag instantly.
2. Use the auto-deal mode on the trainer at a slow speed and keep the running count in your head.
3. Answer the running and true count quizzes; increase speed only once you are consistently correct.
4. Review the basic strategy chart for your chosen rules so that playing decisions become automatic.