Asymmetric Bet Sizing (Truncated vs. Long-Tail Outcomes)
In business, unlike baseball, a single swing can score a thousand runs — size your bets accordingly.
- Difficulty
- Advanced
- Time to result
- ~ongoing to results
- Steps
- 5
- Confidence
- 83%
Drawing on a Jeff Bezos shareholder letter, this framework is a capital allocation decision rule for evaluating high-risk, high-reward bets. The core insight is that business outcomes, unlike baseball, are not truncated — a single successful bet can generate returns many multiples larger than a typical win, while the downside of a failed bet is capped at the capital invested. Given a bet with a 10% chance of a 100x payoff, the expected value strongly favors taking it, even though you'll be wrong nine times out of ten. This reasoning underpinned Amazon's willingness to fund exploratory bets like AWS itself, treating the company's balance sheet like a venture portfolio rather than requiring each individual bet to be safe or highly likely to succeed.
Origin
Drawn from Jeff Bezos's 2015 shareholder letter explaining Amazon's philosophy toward big, bold bets.
Core principles
- 01Business payoffs are not capped the way a baseball swing is
- 02Expected value, not win probability alone, should drive the decision
- 03A high failure rate is an expected and acceptable cost of pursuing asymmetric upside
- 04A small number of big wins can fund and outweigh many losses
- 05Boldness is a rational strategy when the payoff distribution has a long tail
How to run it
- 1
Estimate Probability and Payoff
Assess the rough probability of success and the magnitude of payoff if the bet succeeds.
- 2
Recognize Uncapped Upside
Confirm that, unlike a truncated outcome (a fixed maximum win), the potential payoff of this bet could be many multiples of a typical win.
- 3
Take the Bet If Expected Value Justifies It
Proceed with the bet if the probability-weighted payoff justifies the investment, even when the probability of success is low.
Pro tip A 10% chance at 100x payoff should be taken every time on expected value grounds.
- 4
Expect and Accept a High Failure Rate
Mentally and organizationally prepare to be wrong most of the time — this is a feature of the strategy, not a flaw.
Watch out Punishing individual failures discourages the bold bets this strategy depends on.
- 5
Let a Few Big Wins Fund the Many Losses
Structure the portfolio of bets so that the rare large wins more than compensate for the frequent small losses.
In the wild
AWS began as an uncertain internal initiative with no guaranteed market, requiring years of investment before it proved out; it ultimately became a business generating more operating income than Amazon's entire retail operation.
→ One asymmetric bet came to dominate Amazon's total profitability.
Amazon passed on funding an outside e-reader inventor and instead chose to build its own device internally years later, betting on owning the category outright rather than funding a smaller, safer stake.
→ Amazon created the Kindle as a proprietary, category-defining product.
Common mistakes
Sizing Bets Like a Baseball Swing
Assuming business upside is capped the way a single at-bat is capped leads to under-investing in bets with true long-tail potential.
Avoiding Bets Due to Fear of Being Wrong
Focusing on the high individual failure rate rather than the expected value of the full bet distribution leads to excessive risk aversion.
Not Letting Winners Run
Cutting a successful bet short before it reaches its full long-tail potential forfeits the very payoff that justified taking the risk in the first place.
Is it for you?
Best for
Executives and capital allocators deciding whether to fund ambitious, uncertain new initiatives.
Not ideal for
Organizations without the balance sheet or risk tolerance to absorb a high rate of individual project failures.
From the transcript
“given a 10 chance of a hundred times payoff you should make that bet every time but you're still gonna be wrong nine times out…”
“in business every once in a while when you step up to the plate you can score a thousand runs”
From the episode
Amazon Web Services