Parrondo’s Paradox Explained: How Two Losing Games Win
Quick answer: Parrondo’s paradox is the surprising result that two games that each lose money can produce a winning game when you switch between them. In the classic version, game A loses about 1 cent per play and game B about 0.87 cents, but choosing between them at random gains about 1.57 cents per play in the long run. The secret is that game B’s odds depend on your current capital, and switching changes how often you hit its bad state.
Facts checked November 22, 2026.
Imagine two casino games. You lose money playing the first, and you lose money playing the second. Common sense says mixing them must also lose. It does not. Alternate the two games, even at random, and your money tends to grow. This is Parrondo’s paradox, and it is real, simple enough to simulate in a few lines of code, and a great lesson in why probability can fool our intuition.
The Two Games
Both games use biased coins. Win a toss and you gain 1 dollar; lose and you pay 1 dollar. Let \(\varepsilon = 0.005\), a tiny handicap that makes each game a loser.
| Game | Rule | Chance of winning a toss |
|---|---|---|
| A | Toss coin 1 | \(\tfrac{1}{2} – \varepsilon = 0.495\) |
| B (capital a multiple of 3) | Toss coin 2, the “bad” coin | \(\tfrac{1}{10} – \varepsilon = 0.095\) |
| B (otherwise) | Toss coin 3, the “good” coin | \(\tfrac{3}{4} – \varepsilon = 0.745\) |
Game A obviously loses: you win slightly less than half the time. Game B is trickier. Its good coin looks generous, but it loses too, because of how often you get stuck on the bad coin.
Why Game B Loses on Its Own
Track your capital modulo 3 (the remainder when divided by 3). In game B, when your capital is a multiple of 3 you face the bad coin, which pushes you down most of the time, often back to a remainder of 2. From there the good coin tends to push you up, right back to a multiple of 3. The game keeps funneling you back into its bad state. Working out the long-run share of time in each state (a Markov chain calculation) shows you sit on the bad coin about 38% of the time, not the one-third you might expect, and the expected change per play is about −0.0087 dollars.
Why Mixing the Games Wins
Now play A or B at random, each with probability \(\tfrac{1}{2}\). Game A acts like a shuffle: its near-fair coin moves your capital up or down without regard to the remainder, which breaks game B’s trap. You land on B’s bad state less often, so B’s good coin gets used more than enough to cover A’s small losses.
| Strategy | Expected change per play (long run) |
|---|---|
| Always A | −0.0100 |
| Always B | −0.0087 |
| Random mix of A and B | +0.0157 |
The mixed game’s chance of winning a toss is \(\tfrac{0.495 + 0.095}{2} = 0.295\) in the bad state and \(\tfrac{0.495 + 0.745}{2} = 0.62\) in the good states. Solving for the long-run state frequencies shows how mixing breaks the trap:
| Capital remainder (mod 3) | Game B alone | Random A/B mix |
|---|---|---|
| 0 (bad coin for B) | 38.4% | 34.5% |
| 1 (good coin) | 15.4% | 25.4% |
| 2 (good coin) | 46.2% | 40.1% |
Expected gain for the mix: \(0.345 \times (2 \times 0.295 – 1) + 0.655 \times (2 \times 0.62 – 1) \approx +0.0157\) dollars per play.
Does the Switching Pattern Matter?
Yes. Starting from 0 dollars, the exact expected results after 100 plays and the long-run rates are:
| Strategy | Expected result after 100 plays | Long-run gain per play |
|---|---|---|
| Always A | −1.00 | −0.0100 |
| Always B | −1.39 | −0.0087 |
| Random A/B | +1.29 | +0.0157 |
| A, A, B, B, repeating | +1.39 | +0.0147 |
Over 100 plays the A, A, B, B pattern does slightly better because of the starting position, but over the long run random mixing wins by a little. In our own simulation of 20,000 players making 100 plays each, the averages were about −1.06, −1.38, +1.24 and +1.45 dollars: sample results that match the exact values within random variation. These are averages, not guarantees: many individual players still lose, and the model allows negative balances.
Try it: Parrondo Simulator (2,000 players, 100 plays each)
| Strategy | Average result (dollars) |
|---|---|
| Always A | – |
| Always B | – |
| Random A/B | – |
| A, A, B, B, … | – |
Results vary a little each run, but the pattern holds: A and B lose alone and win together.
Where the Idea Came From
Spanish physicist Juan Parrondo devised the games in 1996 to illustrate a “flashing Brownian ratchet,” in which switching a sawtooth-shaped force on and off makes particles drift in one direction even though neither setting alone would. The switching needs an outside energy source, so this is not perpetual motion. Gregory Harmer and Derek Abbott of the University of Adelaide published the paradox in Nature on December 23, 1999, under the title “Losing strategies can win by Parrondo’s paradox.”
Does It Break the Laws of Probability?
No. The key is that game B is state-dependent: its odds depend on your capital. Combining games changes the long-run frequency of those states, so the combined game is a genuinely different game, not an average of the two. If both games were simple fixed-odds coin flips, mixing them could never help. That is also why the paradox does not help at a real casino: standard independent bets with negative expected value, such as roulette spins, cannot be made profitable merely by alternating them.
Where Parrondo Effects Show Up
- Physics: flashing ratchets and models of molecular motors in cells, which move cargo by switching between states.
- Biology and ecology: models where switching between two unfavorable strategies or environments helps a population survive.
- Finance: “volatility pumping,” where rebalancing between assets can grow a portfolio in theory, a related idea with real-world caveats such as costs and changing markets.
- Algorithms and AI: randomly switching between two search or optimization strategies can sometimes beat either strategy used alone, a theme in randomized algorithms.
Common Misconceptions
“Parrondo’s paradox lets you beat the casino.” No. Real casino games have fixed odds, and mixing fixed-odds losing games still loses.
“The math must be wrong.” The result is exact and can be checked with a Markov chain or a simulation like the one above.
“Any two losing games will do.” No. In the classic construction, state dependence is essential, and the games must interact in the right way.
“Mixing guarantees a profit.” No. It gives a positive expected gain; individual runs can still lose.
Watch: Parrondo’s Paradox Explained
If you learn better by watching, these videos are worth your time.
Check Your Understanding
Five quick questions to see how much of this has stuck. Pick an answer and press Submit; you will see a short explanation either way.
Answer each question and press Submit to see your result right away.
Frequently Asked Questions
What is Parrondo’s paradox?
Parrondo’s paradox is the result that two games that each lose on their own can produce a winning outcome when played in alternation or at random.
How does Parrondo’s paradox work?
One game has odds that depend on your current capital. Mixing in the other game changes how often you are in that game’s bad state, so its good state is used more and the combination gains.
Can Parrondo’s paradox be used to win at gambling?
No. Real casino games have fixed odds that do not depend on your bankroll, and combining fixed-odds losing games still loses.
Who discovered Parrondo’s paradox?
Spanish physicist Juan Parrondo devised it in 1996. Gregory Harmer and Derek Abbott published it in Nature in 1999.
Where is Parrondo’s paradox used?
It appears in physics models of Brownian ratchets and molecular motors, in biology and ecology models, and in ideas about portfolio rebalancing and randomized algorithms.
Keep Exploring
- Infinite Monkey Theorem: Can Monkeys Type Shakespeare?
- Friendship Paradox Explained: Why Friends Have More Friends
- Survivorship Bias Explained: Examples and WWII Planes
Sources and Further Reading
Related to This Article
More math articles
- The Best Grade 4 ELA Practice Tests for Wisconsin Students
- Best Digital Writing Pads for Online Teaching in 2027
- How to Factor Polynomials: Every Method with Examples
- 6th Grade Math: The Complete Parent’s Guide for 2027
- Mental Age and Ratio IQ Calculator
- Misinformation, Media Literacy, and Technological Change
- Wisconsin FORWARD Grade 6 Math Free Worksheets: Free Printable PDFs Covering Every Grade 6 Skill
- Tennessee TCAP Grade 6 Math Free Worksheets: Free Printable TCAP-Ready Practice with Answers
- Inverse Function Calculator (Free, with Graph)
- The Best Algebra 1 Book for California Students


What people say about "Parrondo’s Paradox Explained: How Two Losing Games Win - Effortless Math"?
No one replied yet.