Scratch cards guide
Classic Scratch: the arithmetic of buying a stack
Classic is the page's label and does not describe any card format. This guide looks at a common habit, buying several tickets in one go, and asks what the numbers say about it.
The run used here
The tickets in this example come from a hypothetical run of 900,000 cards at 3 units each. A single ticket wins some prize with probability 20.80%.
| Prize | Tickets | Odds | Expected value per ticket |
|---|---|---|---|
| 15,000.00 | 6 | 1 in 150,000.0 | 0.1000 |
| 300.00 | 150 | 1 in 6,000.0 | 0.0500 |
| 60.00 | 2,000 | 1 in 450.0 | 0.1333 |
| 15.00 | 15,000 | 1 in 60.0 | 0.2500 |
| 6.00 | 50,000 | 1 in 18.0 | 0.3333 |
| 3.00 | 120,000 | 1 in 7.5 | 0.4000 |
187,156 winning tickets out of 900,000 (overall 1 in 4.81). Prize fund 1,140,000.00 on sales of 2,700,000.00: return 42.22%, so a ticket at 3.00 is worth 1.2667 on average. Hypothetical print run.
At least one winner
With k tickets, the chance that none of them wins is (1 - p)k, so the chance of at least one winner is 1 - (1 - p)k. This treats each ticket as an independent draw from the run. Strictly, tickets are drawn without replacement, but with 900,000 cards in the run, removing a handful changes the odds in the fifth or sixth decimal place, so the shortcut is safe.
| Tickets (k) | Cost | P(at least one prize) | P(at least one prize above 3) | Expected net |
|---|---|---|---|---|
| 1 | 3 | 20.80% | 7.46% | -1.73 |
| 5 | 15 | 68.83% | 32.14% | -8.67 |
| 10 | 30 | 90.28% | 53.95% | -17.33 |
| 20 | 60 | 99.06% | 78.80% | -34.67 |
Independent-draw approximation. Expected net is the average of total prizes minus total cost.
The pattern is clear. A stack of 10 tickets almost always contains a winner, which is exactly why buying a stack feels rewarding. But the expected net falls in a straight line: every extra ticket adds the same average loss of 1.7333 units. The stack makes a win more likely; it does not make the purchase better value.
How many winners in a stack of 10
The number of prize tickets among 10 follows a binomial distribution. The table gives the exact chance of each count under the independent-draw model.
| Prize tickets | Chance |
|---|---|
| 0 | 9.72% |
| 1 | 25.51% |
| 2 | 30.14% |
| 3 | 21.10% |
| 4 | 9.70% |
| 5 or more | 3.83% |
The most likely count is two, with one and three close behind, and most of those will be refunds or the 6-unit tier. A stack of 10 costs 30 units and its expected prize total is 12.67 units.
What a stack does not change
- The return per unit spent stays at 42.22% whether you buy one card or a hundred.
- The chance of the top prize grows roughly in proportion to the count. Twenty tickets give about 1 in 7,500.
- Consecutive tickets from one pack are not a better sample than scattered ones, since prize positions are not known to the buyer.