bet-sizing

Bet Sizing

The founder has already computed the upside. That arithmetic is why they are in the room, and it is not in dispute. This skill computes the other number.

The cap is the product here. A maximum spend chosen now, before anything is sunk, is a real constraint. The same number chosen in month three, with 60 hours already in the ground, is a rationalization with a spreadsheet attached.

When to use

Before any bet is written into goals.md — every bet in quarterly-planning passes through here.

Also whenever an opportunity arrives mid-quarter and the founder wants to know whether it is worth doing. It usually is, at some size. The question is which size.

Inputs

Read first — house rule 1:

• metrics.md — the effective hourly rate, cash position, and runway, with their dates
• goals.md — what is already committed, because the founder's hours are already spent on paper
• decisions/ — has a bet of this shape been made before, and what did it actually cost?

Beliefs

• The upside is not in dispute and it is not an input. Every bet that reaches this room has a good upside — that is why it arrived. A number that is high for every option cannot discriminate between them, so computing it again is not rigour, it is company.
• Almost every bet is worth taking at some size, and almost none at the size the founder first proposed. "Should I do this?" is the wrong question wearing the clothes of caution. The question is how much of the company they are willing to lose finding out.
• Deliberation is spend. A week of analysis on something reversible in an afternoon has already cost more than the mistake could, and a founder carefully evaluating a cheap reversible bet is busy not deciding something else. Ask which one.
• Founder hours are not free on a quiet week. "I'd have been working anyway" is the sentence that funds every unpriced bet in this company, and the hours it hides are the hours goals.md had already committed to something else.

Steps

1. Price the hours at the rate in metrics.md. Founder hours are the most expensive input in the company and the only one that gets called free. If metrics.md has no effective rate, stop and hand to the CFO — you cannot size a bet in a currency that does not exist.
2. Answer the question that is being avoided: what does this cost if it is wrong? Not "what if it works". Hours, cash, and the thing you cannot get back — the client you deprioritized, the quarter you spent, the positioning you moved. Write it as a sentence a person could be held to.
3. Set the cap before the start. The maximum hours and cash spent before the judgement date, whichever comes first. Written into goals.md now, while saying a small number is free.
4. Apply the runway rule. If the downside exceeds one month of runway from metrics.md, this is not a bet — it is a bet-the-company move. It goes through the Board Member (premortem) first, and it gets logged in decisions/ by the Chief of Staff whichever way it goes. At this size, ask the Chief of Staff to convene first: the two or three agents whose files the bet touches each state a written position — two or three sentences from their own book, numbers and dates — before the premortem runs. The premortem then attacks a plan whose owners are already on record, and the record follows the decision into decisions/.
5. Apply the reversibility rule. How long to undo if wrong?
   • Under a week — stop analyzing and run it. The deliberation now costs more than the mistake would. A founder who red-teams a reversible bet is avoiding a different decision; ask which one.
   • A week to a month — cap it and go.
   • Over a month — red-team before committing. Irreversibility is the only thing that earns a delay.
6. Write the size into the bet's block in goals.md. A bet whose size lives in the conversation is unsized.

Output

Append to the bet's block in goals.md:

Cost: <hours> h @ <rate from metrics.md, dated> + <cash>
Cap: stop at <hours> h or <cash>, or <date> — whichever comes first
If wrong: <what is lost, concretely> — reversible in <time>

Guardrails

Never size a bet without metrics.md. A cost estimate with no rate and no runway behind it is a wish with units.

Never size the upside. It is not this skill's job, the founder has done it, and adding a second optimistic number to the page will not improve the decision — it will balance it, which is worse.

Refuse to size a bet with no judgement date. Send it back to quarterly-planning. Without a date, "the cap or the date, whichever comes first" has no second half, and the cap will be quietly renegotiated instead.

Do not decide whether to take the bet. You price it. Taking it is the founder's call, and if the price kills it, that is the price talking.