A Super Mega Bowl guide
Flip a coin ten times and you might get seven heads. Flip it ten thousand times and you will land very close to half. That drift toward the true odds is the law of large numbers, one of the most important ideas in probability, and it quietly explains everything from why casinos profit to why opinion polls work.
The law of large numbers says that as you repeat an independent random experiment more and more times, the average of your results moves closer and closer to the expected value, the long-run average the process is built around. For a fair coin the expected share of heads is 0.5. For a fair six-sided die the expected average roll is 3.5. Any single trial can land anywhere, but the running average of many trials settles down near that expected value.
The key word is average. The law is a statement about proportions and means, not about individual outcomes. It does not say results will balance out on any particular flip. It says the average of a long run of flips will be close to the true value, and closer still as the run gets longer.
Suppose you flip a fair coin and track the share of heads after each flip. Early on the share swings wildly. After four flips you might sit at 75 percent heads. After twenty you might be at 40 percent. As the count climbs into the hundreds and thousands, those swings shrink and the share hugs 50 percent.
| Number of flips | Typical share of heads |
|---|---|
| 2 | anywhere from 0 to 100 percent |
| 10 | often 30 to 70 percent |
| 100 | usually 42 to 58 percent |
| 1,000 | usually 47 to 53 percent |
| 10,000 | usually 49 to 51 percent |
The pattern is clear. More trials do not make the coin any fairer. They make your measurement of its fairness more precise.
Here is the part that surprises people. As you flip more, the proportion of heads tightens toward 0.5, yet the raw difference between the number of heads and the number of tails usually gets larger, not smaller.
In 100 flips a gap of 10 (55 heads against 45 tails) is a 10 percent imbalance. In 10,000 flips a gap of 100 (5,050 against 4,950) is only a 1 percent imbalance, even though the raw gap is ten times bigger. The proportion converges because the total grows faster than the gap does. The gap tends to grow with the square root of the number of flips, while the total grows with the number of flips itself.
This is why "it will even out" is only half true. The ratio evens out. The raw count does not.
The law of large numbers is often confused with the gambler's fallacy, the belief that a run of heads makes tails more likely. The law promises nothing about the next flip. Each flip is independent and stays 50/50 no matter what came before. The average converges not because past streaks get corrected, but because they get diluted by an ever-larger pool of future flips.
The law of large numbers quietly runs a great deal of the world.
The fastest way to believe the law is to watch it happen. Flip a large batch of coins and track the running share of heads, roll a handful of dice and watch the average settle near 3.5, or spin a wheel many times and see each option approach its fair share. In the short run the results look lumpy. Give them volume and the averages fall into line.
Try it → Flip 100 or more coins and watch the running tally close in on 50 percent, even as the raw gap between heads and tails keeps drifting.The law of large numbers is not a promise that luck balances out. It is a promise that averages become reliable as samples grow. Individual results stay unpredictable forever. It is the averages, not the outcomes, that become certain.