# Your First Hundred Practice Hands: What You Will Actually Be Dealt

What a hundred random blackjack practice hands actually contain: about 38 stiff hands, 9 or 10 soft hands and about 5 non-ten pairs, and why a 100 hand accuracy score wobbles by several points.

Source: https://21trainer.app/articles/first-hundred-practice-hands/
Updated: October 13, 2026
Publisher: 21 Trainer, makers of Blackjack Strategy: 21 Trainer, an educational blackjack trainer for iPhone and Android (App Store: https://apps.apple.com/app/blackjack-strategy-21-trainer/id6759527695, Google Play: https://play.google.com/store/apps/details?id=app.blackjacktrainer)

> **Educational content, not gambling advice.** 21 Trainer and this article teach blackjack strategy and card counting as skills, using virtual chips only. No strategy or counting system guarantees winnings, and nothing here encourages real-money play. If gambling is a problem for you or someone you know, call 1-800-522-4700 (National Council on Problem Gambling).

A hundred hands is a common first practice goal. It is a round number, it fits in a couple of sittings, and it feels like real experience. Many learners reach it, glance at their accuracy score, and draw a conclusion on the spot.

The conclusion is usually premature, because a hundred hands teaches less than it feels like: the deal is lopsided. A shuffled shoe does not deal the chart evenly: it repeats the same handful of situations and can skip entire regions through no fault of yours.

That is true whether the shoe is a deck of cards at home or a trainer dealing hands at random. The numbers below are arithmetic on our piece on [how often each starting hand comes up](https://21trainer.app/articles/starting-hand-frequencies/), worked out for a full six deck shoe.

## The mix you will actually be dealt

Group the two-card starting hands by category and count what a hundred of them look like:

| Hand category | Share of hands | In 100 hands |
| --- | --- | --- |
| Hard 12 through 16 (stiffs) | 38.51% | about 38 or 39 |
| Hard 17 through 20 | 25.40% | about 25 |
| Hard 9 through 11 (doubling totals) | 12.44% | about 12 |
| Hard 4 through 8 | 8.83% | about 9 |
| Soft hands, A,2 through A,9 | 9.50% | about 9 or 10 |
| Blackjack | 4.75% | about 5 |
| A,A | 0.57% | maybe 1 |

The shares sum to 100 percent, and their shape is the story. About 64 of your first hundred hands are stiff totals and pat hands, the heavily repeated hard rows of the chart. Nine or ten are soft hands, spread thinly across eight different soft totals, since each one from A,2 through A,9 arrives the same 1.19 percent of the time. The four or five blackjacks need no playing decision at all.

The pairs that create split decisions, 2,2 through 9,9 plus A,A, amount to 5.12 percent of hands, about five per hundred, and those five are scattered across nine different pairs. (The 2,2 through 9,9 pairs are already counted inside the hard rows above.) So a learner finishes a hundred hands having seen each stiff total about seven to nine times and most pair rows once or not at all, while the overall score blends them into one number.

## The hands that may not show up at all

Rarity has a second effect: some hands plausibly never arrive. A specific pair such as 8,8 is 0.57 percent of starting hands, so the probability of seeing none in a hundred hands is 0.9943 raised to the power 100, which is 56.5 percent. More likely than not, your first hundred hands contains no 8,8 at all. Widen to 300 hands and the chance of still having seen none is about 18 percent; at 500 hands it is still close to 6 percent.

It takes about 122 hands before the chance of seeing a given pair once reaches 50 percent, and about 403 hands before it reaches 90 percent. A specific soft hand is commoner but still thin: A,7 arrives 1.19 percent of the time, so about 30 percent of hundred hand sessions contain no A,7 at all, and even 300 hand sessions miss it about 2.8 percent of the time.

So a first hundred hands can feel complete while leaving whole rows untested.

## Specific chart cells are rarer still

Everything so far counted only your two cards. A strategy chart cell also needs a dealer upcard, and the upcard has its own skew: a ten-value card shows as the upcard 4 times in 13, about 30.8 percent, while any other specific rank shows 1 time in 13, about 7.7 percent. Multiply the hand frequency by the upcard frequency, treating the two as independent. That is only an approximation, since all three cards come from the same shoe, but with six decks it is a close one. The result is how often each specific cell really appears:

| Hand and upcard | Arithmetic | Comes up about |
| --- | --- | --- |
| Hard 16 against a 10 | 6.51% x 4/13 | 2.0%, one hand in 50 |
| Hard 12 against a 3 | 8.88% x 1/13 | 0.68%, one hand in 146 |
| Hard 11 against an ace | 4.75% x 1/13 | 0.37%, one hand in 274 |
| Soft 18 against a 9 | 1.19% x 1/13 | 0.092%, one hand in 1,092 |
| 8,8 against an ace | 0.57% x 1/13 | 0.044%, one hand in 2,281 |

Two of those rows are slightly generous: the hard 16 and hard 12 shares include 8,8 and 6,6, which the chart reads from the pair rows. Take the pairs out (6.51 minus 0.57, and 8.88 minus 0.57) and those two cells come up about once in 55 hands and once in 156.

Some of the cells learners find hardest are among the ones the deal tests least. Soft 18 against a 9 is a genuinely tricky decision, and a hundred hands gives roughly a one in eleven chance of seeing it even once. A pair of 8s against an ace shows up about once per 2,281 hands. That scarcity is why our piece on [drilling pair splits](https://21trainer.app/articles/drilling-pair-splits/) argues that pairs need forced repetitions rather than whatever the random deal happens to offer.

## What a hundred hand accuracy score is worth

An accuracy score over 100 hands is a sample proportion, and sample proportions wobble. The standard binomial formula, the square root of p times 1 minus p divided by n, puts two standard errors at roughly 8 points if your true accuracy is 80 percent, 6 points at 90 percent, and 4.4 points at 95 percent. At 300 hands those shrink to 4.6, 3.5 and 2.5 points, and at 1,000 hands to 2.5, 1.9 and 1.4.

In practice that means a 92 today and an 88 tomorrow are the same score at the hundred hand mark. This lines up with what our guide to [what your accuracy percentage really means](https://21trainer.app/articles/blackjack-accuracy-percentage/) says: ignore any accuracy figure built on fewer than about 200 decisions, and expect that scores in the low 90s usually mean one or two hand families are still wrong every time they appear.

The lopsided deal adds a quieter distortion. Because stiffs and pat hands are most of what you are dealt, an overall score mostly measures the rows you see constantly. Hard 16 against a 10 appears about twice per hundred hands, and our deeper dive into [16 against a dealer 10](https://21trainer.app/articles/sixteen-vs-dealer-ten/) explains why it feels wrong even when played right. But the soft doubles and pair splits barely register in the sample at all. Soft hands and the pairs that create split decisions are together under 15 percent of the deal, so a learner who has the hard totals down cold and gets every soft hand and every one of those pairs wrong would still get roughly 85 of a hundred opening decisions right.

## What to do about it

None of this is an argument against the first hundred hands. Do them anyway: the first hundred is where the big rows, the stiffs and the pat totals, get their repetition, and those rows carry most of the game's decisions. Accuracy tracking, streaks and session goals are part of the free tier of 21 Trainer, so it is still a satisfying milestone.

What changes is what you trust it for. After the first hundred, stop relying on the random deal to cover the rare rows. Give pairs their own focused repetitions, work soft hands with our piece on the [soft hand recognition drill](https://21trainer.app/articles/soft-hand-recognition-drill/) until recognizing them is instant, then scramble everything together with [mixed hand drills](https://21trainer.app/articles/mixed-hand-drills/) so the recognition survives without a label on it. In 21 Trainer, the Weakness Drill and accuracy heat maps by hand type and dealer upcard are Pro features built for exactly this.

Judge readiness over hundreds of hands, not one hundred. Our piece on [knowing when the chart is finally learned](https://21trainer.app/articles/when-is-the-chart-learned/) collects the thresholds this site publishes. Two come from the [four week practice schedule](https://21trainer.app/articles/blackjack-practice-schedule/): accuracy in a single category steady above roughly 95 percent across at least a few hundred hands, and, before moving on to anything beyond basic strategy, accuracy in the high nineties on mixed drilling at table pace. The third, 99 percent or better across several hundred mixed hands as the working target, comes from the accuracy piece. None of those numbers can be read off a single hundred hand session.

## Frequently asked questions

### How many blackjack hands does it take before an accuracy score means anything?

Treat 100 hands as a first checkpoint, not a verdict. At that sample size, two standard errors of sampling noise are roughly 6 to 8 points when true accuracy is between 80 and 90 percent, so a 92 today and an 88 tomorrow can be the same performance. Reliable readings need a few hundred mixed hands; ignore any figure built on fewer than about 200 decisions.

### Why does blackjack practice deal mostly hard hands?

Because the deal itself is lopsided. Per our starting hand frequencies piece, in a six deck game hard hands are about 85 percent of two-card totals, and hard 12 through 16 alone is 38.51 percent. Soft hands are about 9.5 percent and specific pairs about 0.57 percent each, so a random hundred hands is dominated by stiff totals and pat hands, whether the cards are dealt by hand or by a trainer dealing at random.

### Will a specific pair show up in a hundred blackjack hands?

More often than not, no. A specific pair such as 8,8 arrives 0.57 percent of the time, so the chance of seeing none in 100 hands is about 56.5 percent. Even in 300 hands there is about an 18 percent chance a given pair never appears. It takes about 122 hands for an even chance of a specific pair, and about 403 hands for a 90 percent chance.

### Is 90 percent accuracy good after 100 hands of blackjack practice?

It is a promising start, but not a measurement. Sampling noise at 100 hands is several points wide, and because stiff and pat hands dominate the deal, the score mostly reflects the high volume rows. Low nineties usually mean one or two hand families are still wrong every time they appear, so check your misses by hand type rather than trusting the headline number.

### What is the best way to practice rare blackjack hands?

Stop waiting for the random deal to supply them. Deal pairs and soft hands on purpose in focused sessions, then mix them back in with confusable hand types so recognition stays sharp. Judge readiness over hundreds of hands, and use accuracy heat maps or a weakness drill to find the cells that are still weak.

## Keep reading

- [How Often Each Blackjack Starting Hand Comes Up](https://21trainer.app/articles/starting-hand-frequencies/)
- [What Your Blackjack Accuracy Percentage Really Means](https://21trainer.app/articles/blackjack-accuracy-percentage/)
- [Drilling Pair Splits When Pairs Rarely Show Up](https://21trainer.app/articles/drilling-pair-splits/)