The Conviction Ladder: How to Size Bets When You're Not Sure How Sure You Are

The Problem With Knowing Your Edge

Every bet sizing framework — Kelly, fractional Kelly, fixed percentage — starts with the same assumption: you know your edge. You've calculated the true probability is 62%, the market says 55%, and now you plug numbers into a formula.

The problem is you don't actually know it's 62%. You *think* it's somewhere between 58% and 68%. Maybe. On a good day. And that fuzziness — the confidence around your confidence — is the thing that quietly bankrupts smart traders.

This is where the Conviction Ladder comes in. It's not a formula. It's a discipline.

The Five Rungs

Before sizing any bet, force yourself onto one of five rungs. Be honest. The lie you tell yourself here is the money you lose later.

Rung 1 — Vibe (0.25% of bankroll). You have a feeling. You read something. The number looks off. You cannot articulate the model. This is a hunch trade, and hunch trades get hunch-sized.

Rung 2 — Directional (0.5%). You have a real reason to think the market is wrong, but you can't quantify by how much. You know it's mispriced. You don't know if it's a 3-point or 15-point mispricing.

Rung 3 — Quantified (1–2%). You have a specific number in mind. "I think this is 65%, market says 55%." You can defend the number to a skeptical friend without hand-waving.

Rung 4 — Stress-tested (2–4%). You've quantified it *and* you've thought about how you could be wrong. You've considered the base rate, the counterargument, the resolution risk. The bet survived interrogation.

Rung 5 — Structural (4–6%). Rare. You've identified a mechanical mispricing — a platform quirk, a resolution ambiguity in your favor, a correlated market that's out of sync. Not "I think this event happens" but "the market literally cannot be right."

Why This Works Better Than Kelly Alone

Kelly's brutal insight is that if you overestimate your edge by 2x, Kelly betting will blow you up. Half-Kelly is the standard fix, but it treats every trade the same — as if your certainty on a coin-flip election is the same as your certainty on a weather contract you modeled for six hours.

The Conviction Ladder does what half-Kelly can't: it forces a pre-commitment to honesty. You size the *quality of your thinking*, not just the perceived edge.

If you find yourself sizing every bet at Rung 3, congratulations — you're lying to yourself. Most of your trades are Rung 1 or 2. That's fine. Real Rung 4 trades might show up twice a month. Real Rung 5 trades might show up twice a year.

The Downgrade Rule

Here's the part nobody follows: when in doubt, drop a rung.

If you're on the fence between Rung 2 and Rung 3, you're on Rung 2. If you can't decide between 3 and 4, you're on 3. The whole point is that fuzzy confidence gets sized like fuzzy confidence.

This feels bad. You'll look at a trade you *knew* was right and think about the money you left on the table. That regret is the tax you pay for not blowing up on the ten other trades you were equally sure about.

The Journal Test

At the end of every month, look back at your trades and label each one with the rung you assigned. Then check: did your Rung 4 trades actually win more than your Rung 2 trades?

If they didn't, your ladder is broken. You're not calibrated. Your "stress-tested" trades are actually just vibes with extra steps.

Most traders discover their Rung 3s and Rung 4s perform identically. That's a signal to collapse them — you don't actually have that much resolution in your own judgment. Better to know that now than after you've sized a Rung 4 like you meant it.