Imagine two choices. One wins a little more often. The other puts more money on the table. Which has the better expected result?
MAKE YOUR PREDICTION
Which option has the higher EV?
Imagine these two invented options. A win pays even money, a loss costs the stake, and there are no pushes.
Choose an option to reveal the reasoning, or continue to the worked example below.
Win probability tells you how often a defined event occurs. Expected value (EV) includes the amount gained or lost in each possible outcome.
EV is the average implied by a probability model. Your next hand, or even a whole session, can finish very differently. A push is a tie and contributes zero net profit. Read the expected-value definition.
02 / A FAIR COMPARISON
Keep the original wager as your reference.
In a blackjack action comparison, express each action’s net result relative to the original wager. An ordinary even-money win gains one unit. A loss costs one. Doubling can produce a two-unit gain or loss.
That original wager stays the reference even when an action adds another stake. Otherwise, you would be comparing values measured in different units. Wizard of Odds explains this action-EV convention.
03 / THE REVEAL
Less likely to win. Higher expected value.
Here are the two options from the question. These probabilities are invented to explain the arithmetic, not calculated for a particular blackjack hand. There are no pushes and wins pay even money.
| Option | Win / lose | Net win / loss | EV |
|---|---|---|---|
| A · stake 1 unit | 55% / 45% | +1 / −1 | +0.10 |
| B · stake 2 units | 53% / 47% | +2 / −2 | +0.12 |
Multiply each possible net result by its probability, then add the contributions:
- A: 0.55 × 1 − 0.45 × 1+0.10
- B: 0.53 × 2 − 0.47 × 2+0.12
Option B has the higher EV even though it wins less often. It also puts two units at risk. The higher EV does not remove that larger possible loss.
04 / WHEN ONE HAND BECOMES TWO
Ask what a split percentage measures.
After splitting a pair into two hands, “win probability” can describe different events: winning one particular hand, winning at least one hand, or finishing with a positive combined result.
Those percentages answer different questions. The split EV calculation must account for the resulting hands, extra wagers and the relevant double or resplit rules. A percentage on its own cannot replace that calculation. John A. Nairn’s pair-splitting analysis explores the maths.
05 / THE FULL PICTURE
Read the assumptions beside the answer.
Start with actions that are legal for the current hand. Then check the table rules and whether dealer blackjack has already been ruled out. Published EV tables separate these assumptions.
The highest EV can still be negative. If every legal choice loses on average, the best EV represents the smallest expected loss in that comparison. You can see negative values in published action-return tables.
Later examples apply this comparison to Player Edge and a losing hand, and to Cash Out offers in net units.
In the Advantage Count Analyzer, read the rules and result source alongside the values. Use the Decision Trainer to practise making a choice before seeing the result.
Sources and further reading
The numerical comparison is an original hypothetical example, not an Advantage Count calculation or a product-performance claim.
- OpenStax — expected value
- Wizard of Odds — action EV and wager units
- John A. Nairn — blackjack pair splitting
- Wizard of Odds — expected values by dealer procedure
- Wizard of Odds — infinite-deck action returns
Our methodology explains exact, approximate, simulation and strategy results.