Card counting is usually explained as a snapshot: a running count exists, you divide it by the decks remaining, and the result tells you something. What that explanation leaves out is time. The count is not a mood that appears in the shoe; it is a quantity with a shape, and the shape changes as the shoe plays out.
Two things move through every shoe: the running count's spread, meaning how far from zero it typically sits, and the true count's spread, which is what actually drives decisions. The first rises and then falls. The second only grows. The interaction between them is why the depths of a shoe matter, why a counter's favorite rounds arrive late, and why the cut card's placement changes the character of a table.
To show the shape with numbers rather than adjectives, we simulated 200,000 six deck shoes counted with Hi-Lo, where 2 through 6 count plus 1, 7 through 9 count 0, and 10 through ace count -1. The method is simple: the running count is tracked card by card, and the true count is the running count divided by the exact number of decks remaining, unrounded, measured at the moment the given number of decks has been dealt. If you want the background on that division, our pieces on Hi-Lo and on running count versus true count cover it.
The numbers, one table
Here is what the 200,000 shoes looked like at six depths. The spreads are standard deviations, shown at their exact formula values, which the simulated spreads match within 0.01; the percentages are the share of simulated shoes in which the true count was at or beyond each level at that exact moment.
| Decks dealt | Running count SD | True count SD | True count +2 or higher | True count +3 or higher | True count -2 or lower |
|---|---|---|---|---|---|
| 1 | 5.78 | 1.16 | 5.0% | 0.6% | 5.0% |
| 2 | 7.31 | 1.83 | 15.2% | 5.7% | 15.2% |
| 3 | 7.76 | 2.59 | 23.9% | 13.6% | 23.8% |
| 4 | 7.31 | 3.66 | 31.5% | 22.5% | 31.7% |
| 4.5 | 6.72 | 4.48 | 35.4% | 25.0% | 35.6% |
| 5 | 5.78 | 5.78 | 39.8% | 33.2% | 39.9% |
The average running count is zero at every one of these points, because Hi-Lo is a balanced count: equal amounts of plus and minus tags across a full shoe. The table describes spread, not drift.
One quiet check on the method: the running count's spread follows an exact formula, not just a simulation. Each card's Hi-Lo tag has a variance of 40/52, because 40 of the 52 cards carry a tag of plus or minus 1 and the other 12 carry 0, and the standard deviation of the running count after k cards of a 312 card shoe works out to the square root of k x 40/52 x (312 minus k) / 311. The simulation matches that formula within 0.01 at every depth in the table, which is the kind of agreement you want before quoting either one.
Why the running count rises, then falls
Look down the first column of numbers: the running count's spread climbs from 5.78 after one deck to a peak of 7.76 after three, then shrinks back to 5.78 after five. The symmetry is exact, and the reason is structural. The dealt cards and the undealt cards always have opposite totals, and the full shoe sums to zero, so the count of the seen cards and the count of the unseen cards are mirror images. At the exact halfway point of the shoe, the two sides are as spread as they can be, one deck past halfway the remaining cards constrain the running count back toward the values still possible, and with one deck left the spread is exactly what it was after the first deck, 5.78 by the formula.
Halfway through a six deck shoe is 3 decks dealt, and that is precisely where the table peaks. Nothing about Hi-Lo's tags produces this; any balanced count on any number of decks shows the same shape, a bell in the middle and shrinkage at both ends.
Why the true count only widens
The true count column tells the opposite story: 1.16 after one deck, 2.59 after three, 5.78 after five, growing at every step. The running count's spread narrows only modestly through the late shoe, but it is being divided by fewer and fewer decks. A running count of +6 early, spread across five unseen decks, is a faint true count of about +1; the same +6 against one unseen deck is +6. Our piece on estimating decks remaining is entirely about this denominator, and the table shows why it deserves the attention.
This is also where estimation errors bite hardest. Misjudging five decks remaining by half a deck changes the denominator by a tenth. Misjudging one deck remaining by half a deck changes it by half. Late in the shoe, when the true count is widest and decisions weigh most, the same eyesight error does several times the damage it did earlier in the shoe.
21 TrainerFree on the App Store. Instant right or wrong feedback with a clear explanation on every hand.
Get the app
The mirror in the last column
Read the +2 column and the -2 column side by side: 5.0 and 5.0, 15.2 and 15.2, 23.9 and 23.8, 31.5 and 31.7, 39.8 and 39.9. High and low true counts are equally likely at every depth. That symmetry comes from Hi-Lo's tags, twenty cards tagged plus 1 and twenty tagged minus 1 per deck: a shoe that drifts rich in high cards is exactly as probable as one that drifts rich in low cards, so the counter's favorable moments and their unfavorable mirror images arrive in equal measure. The practical takeaway is not about betting, which this piece does not cover, but about patience: in this simulation, a true count of +3 or higher was present after one dealt deck on 0.6 percent of shoes, and after five decks on 33.2 percent.
A caution about what these numbers are. Each row is a snapshot at one moment: the state of the shoe at exactly that depth, across 200,000 shoes. They are not the share of hands played at each count, because hands are dealt throughout the shoe, not all at the 5 deck mark, and a snapshot percentage cannot be summed or averaged as if they were. Treat the table as a description of where in the shoe the count concentrates, nothing more.
Why this makes penetration the number counters care about
The favorable counts live late. That single fact, visible down the rows of the table, is what makes the cut card so consequential for a counter. A six deck shoe cut off after four decks reaches the 4 deck row, where a true count of +2 or higher sits at 31.5 percent of shoes, but never sees the 5 deck row at 39.8. Cut it off after three and even that is gone.
Our deck penetration piece makes the same point from the other direction: penetration does not make favorable moments more favorable so much as it makes them happen at all, which is why counters describe penetration in terms of frequency rather than size. The table above is the frequency argument in numbers. For how the count behaves across deck counts more generally, see our piece on why counting gets harder as decks increase, and for counts whose tags do not sum to zero over a deck, balanced versus unbalanced counts.
Practice notes, and the honest limits
If you drill counting, drill it with this shape in mind. Early shoe hands teach tag fluency; late shoe hands teach everything else, the division by a shrinking denominator and the cost of a lazy estimate. The Pro upgrade in 21 Trainer includes a true count drill for exactly that conversion, alongside the counting systems and index deviation training. Counting with your head is legal in the United States. Casinos are private businesses that may refuse service to suspected counters, and no technique removes the risk from the game; a wide true count describes the composition of the shoe, not a promise about the next card. If counting stops being worth the effort at the tables you actually face, our piece on when counting is not worth it is the honest companion read, and losing the count entirely has its own recovery habits.
Train the conversionHi-Lo, KO, Omega II and Zen with a true count drill in the Pro upgrade. Virtual chips only, and your progress stays on your device.
Download
Frequently asked questions
What happens to the running count as a shoe progresses?
Its average stays at zero in a balanced count like Hi-Lo, but its spread rises and then falls. The dealt cards and the undealt cards always have opposite totals, and the full shoe sums to zero, so the running count's spread peaks halfway through the shoe, after 3 of 6 decks in our simulation, and then shrinks back toward zero as the shoe ends.
Why does the true count spread widen late in the shoe?
Because the true count divides the running count by the decks remaining. Late in the shoe the running count's spread has narrowed only somewhat, while the denominator has shrunk a lot, so the same running counts convert to much bigger true counts. In our table the true count's standard deviation grows from 1.16 after one deck to 5.78 after five.
How often is the true count plus 2 or higher in a six deck shoe?
It depends strongly on depth. In our 200,000 shoe Hi-Lo simulation, a true count of +2 or higher was present 5.0 percent of the time after one deck dealt, 15.2 percent after two, 23.9 percent after three and 39.8 percent after five. These are snapshots at each depth, not the share of hands played at each count.
Why does deck penetration matter so much for counting?
Because the counts a counter waits for mostly appear late, and a shallow cut card ends the shoe before they can. As our deck penetration piece puts it, penetration does not make favorable moments more favorable so much as it makes them happen at all. A shoe cut off early never reaches the depths where high true counts concentrate.
Is card counting illegal?
Counting with your head is legal in the United States. Casinos are private businesses that may refuse service, back off or ban suspected counters, and no technique removes the risk from the game. Using electronic devices to count is separately illegal in some jurisdictions. Counting is a skill for reading the game, never a promise of income.