even with 50/50 odds, most players lose everything. and copying successful risk-takers is a trap.

If you hear "50/50 fair odds" and your brain goes "oh, that's safe" that's a mistake. Let me show you something fascinating: two people, $100 each, same exact chances and wildly different results.

I've been using AI to simulate how markets work in a simplified way, and what I discovered might change how you think about risk forever.

The Setup: A Fair Game

Imagine two people, each with $100. They bet 20% of the lowest amount each round. Whoever wins gets that amount added to their wealth; whoever loses pays it out. Simple. Fair. 50/50 odds every single round.

Round 0: Both have $100
Round 1: Person one wins, goes to $120. Person two loses, drops to $80
Round 2: We bet 20% of $80 (the lowest amount). Person one wins again, now at $136
Round 3: Person one wins yet again...

After 50 rounds? They're almost back where they started. Makes sense, right? Fair game, fair results.

But watch what happens after 400 rounds.

The Shocking Result

Person one has $189. Person two has $11.

The bet is now only $2.70 (20% of $11). If person one loses this round, it's barely noticeable. But if person two loses, it's 20% of everything they have left. Person two can fight, but they have no chance. Eventually, they're down to pennies, betting cents while person one has hundreds.

Here's the key insight: when you lose, you lose 20% of your wealth. When you win back that same amount, you only gain back 16% of your new total. Losses compound faster than wins.

Scaling Up: 50 Players

I ran 50 different simulations of this game. After 400 rounds, almost every single game ended the same way: one person had nearly $200, and the other had nearly $0. Only 3 out of 50 games weren't decisive yet—but give them more rounds, and they would be.

This isn't about luck or skill. It's pure mathematics. At some point, the bet size becomes significant only to one person.

Now let's scale it further: 50 people, all playing together in a free-for-all. Every round, two random players face off with 50/50 odds.

After 10,000 rounds? One person has essentially all the money. Everyone else has zero. There's always one survivor.

The Yard Sale Model

This phenomenon is called the Yard Sale Model. It simulates pure free-market capitalism where everyone has fair chances at every transaction. The result? It always leads to unfair outcomes. One player accumulates everything because once they get ahead, bet sizes become insignificant to them but devastating to others.

The Strategy Experiment: 1,500 Players

I wanted to know: is there a winning strategy? So I created 1,500 agents, each starting with $100, but each with different risk preferences—from 0% to 100% of their wealth per bet.

Round 1: Scattered results. One unlucky 90%-better lost immediately. Some 70%-betters did well.

Round 10: Higher betters' median performance was below $100. Conservative betters still around $100.

Round 100: Any group willing to risk more than 50% had a median player who was gone—out of the game completely. No group's median player had more money than they started with.

Round 50,000: The results were stark.

The median player for every group betting more than 5% of their wealth was eliminated. Gone. Zero dollars.

Yes, there were individual survivors with high bet percentages—sometimes a 90%-better doing exceptionally well. But their entire group was dead. They were lone survivors in a field of corpses.

The top 30 performers averaged 9.6% bet size. The bottom 30 performers averaged 91% bet size.

I ran this simulation four times. Same pattern every time.

The Winner's Illusion

Here's the trap: we see that one person who bet aggressively and won everything. We think, ""That's the strategy!"" So we copy them.

But we don't see the 1,499 other players who used similar strategies and lost everything.

That winner didn't win because of skill. It was pure randomness. There's no way to know in advance who will be that statistical outlier.

This is the Winner's Illusion—we study successful risk-takers and think their strategy is proven, when really they're just the visible survivors of a massacre.

The Kelly Criterion

This model demonstrates what's called the Kelly Criterion Zone:

In a fair game (50/50), the optimal bet is 0%
Only bet when you have a real edge (better than 50/50)
Even then, bet a very small percentage of your wealth
Overbetting will destroy your wealth more times than not

The Lessons

Even if the odds are fair, the results won't be fair. You might win individual rounds 50% of the time, but your chance of winning the overall game is far lower—potentially 1 in 150 or worse.

Losses compound faster than wins. When you lose 20% then win back that dollar amount, you haven't recovered—you're still down because the percentage lost was larger than the percentage gained back.

Don't copy the winner. That aggressive better who won big? They had the same strategy as hundreds who lost everything. Their success was randomness, not skill.

The only guaranteed strategy is to bet as little as possible. In my simulations, only the near-zero betters consistently survived with their wealth intact.

How to Actually Win

If you want guaranteed results:

Bet as little as possible on any single opportunity
Diversify widely across many small bets, not one big one
Use discipline, not glory as your teacher—success stories are bad teachers
Focus on surviving the long game, not fast gains
Understand that fair odds ≠ fair outcomes over time

The market, investing, entrepreneurship—they're not perfectly 50/50 games, but the math of compounding losses applies everywhere. The person who survives long enough to play many rounds has the advantage, not the person who bets big and hopes for luck.

Fair odds can be deadly. The game is survival, not glory.