동전 던지기 확률: 우연의 수학
동전 던지기의 실제 확률과 도박사의 오류를 이해하세요.
Basic probability: 50/50 or not exactly
Theoretical probability of heads is exactly 1/2 (50%). But in practice:
Stanford study (2007): 350,000+ filmed flips showed coins land slightly more on the side that started facing up (~51% vs 49%). This is the precession bias.
Real coins aren't perfect: A euro coin has different relief on each side. Studies show the heavier side lands slightly more often.
Flip your virtual coin with the NexTools coin flipper — perfectly 50/50 thanks to digital random generation.
The gambler's fallacy: why your intuition lies
You flip 5 heads in a row. What's the probability the next is tails? Exactly 50%. The coin has no memory. Each flip is independent.
The gambler's fallacy is believing past events affect future independent probabilities. It's behind many casino losses.
Monte Carlo (1913): Roulette landed black 26 times in a row. Gamblers bet millions on red. They lost. Red probability stayed ~48.6% every spin.
Use the percentage calculator for compound probability.
Law of large numbers: when 50% actually holds
10 flips might give 7 heads (70%). Loaded coin? Not necessarily.
| Flips | Likely range of heads |
|---|---|
| 10 | 30-70% |
| 100 | 40-60% |
| 1,000 | 47-53% |
| 10,000 | 49-51% |
| 1,000,000 | 49.9-50.1% |
With the NexTools flipper you can do 1000 flips in seconds to see the law in action.
Real applications: when flipping a coin is the right decision
1. Tie-breaking: FIFA used coins before penalty shootouts (1968 World Cup).
2. Statistics: Randomized response uses coins for honest answers in sensitive surveys.
3. Cryptography: Random number generators are essentially digital coins generating random 0s and 1s.
4. Game theory: Optimal mixed strategies require true randomness.
Fair vs loaded coin: how to detect cheating
Chi-squared test: Flip N times, count heads/tails, apply chi-squared to check if deviation from 50% is statistically significant.
Example: 100 flips, 60 heads. Chi = 4.0, p = 0.046. Significant at 5% — probably biased.
But 55 heads in 100: Chi = 1.0, p = 0.317. Not significant — normal variation.
The percentage calculator helps with the math.
Probability of streaks: consecutive heads
| Consecutive | Probability | 1 in... |
|---|---|---|
| 2 heads | 25% | 1 in 4 |
| 5 heads | 3.125% | 1 in 32 |
| 10 heads | 0.098% | 1 in 1,024 |
| 20 heads | 0.000095% | 1 in 1,048,576 |
| 26 heads (Monte Carlo) | 0.0000015% | 1 in 67 million |
Improbable but not impossible. In millions of daily casino flips, 20+ streaks happen regularly.
Coin flips in programming: Math.random() and CSPRNG
JavaScript: Math.random() < 0.5 ? "heads" : "tails"
Python: import random; random.choice(["heads", "tails"])
For cryptography: crypto.getRandomValues() (JS) or secrets.choice() (Python).
For secure passwords (which use "cryptographic coins" internally), use the NexTools password generator.
History of the coin toss: from Rome to algorithms
Ancient Rome: "Navia aut caput" (ship or head) — first known coin toss using Roman coins.
Medieval England: "Cross and pile" — Christian cross on one side.
Super Bowl: The coin toss winner wins ~52% of the time (1999-2024 stats).
Digital: Virtual coin generators use pseudorandom algorithms producing perfectly uniform distributions — "fairer" than any physical coin.
이 도구를 사용해 보세요:
도구 열기→자주 묻는 질문
Is a real coin exactly 50/50
Not exactly. Stanford studies show ~51/49 bias toward the starting-up side. Coins with unequal relief have additional biases. For perfect 50/50, use a digital generator.
If I get 10 heads in a row is the next more likely to be tails
No. Each flip is independent. Probability stays 50%. Believing otherwise is the gambler's fallacy. The coin has no memory.
How many flips to detect a loaded coin
Minimum ~100 for large biases (60/40). For small biases (52/48), thousands needed. Chi-squared test determines statistical significance.
Why is flipping in the air fairer than spinning on a table
Spinning on a table rotates on one axis where unequal weight causes more bias. Air flipping has more uniform rotation. Fairest: high toss with fast spin.
Is Math.random() truly random
Pseudorandom — generates from a seed using a deterministic algorithm. Fine for games. For cryptography use crypto.getRandomValues() which uses OS entropy.
How many possible combinations in N flips
2^N. 10 flips: 1,024 sequences. 20: 1,048,576. 32: 4,294,967,296 (same as IPv4 addresses).