Skip to main content

​
The Binary Problem

Every major prediction market today forces you into binary choices:
  • Polymarket: “Will X happen?” → Yes / No
  • Kalshi: “Will Y be above Z?” → Yes / No
  • Metaculus: “Will this occur?” → Yes / No
But real predictions are rarely binary.

​
A Tale of Two Predictions

Binary vs Continuous Outcome Resolution

​
The Binary Way

Question: “Will Bitcoin hit $100,000 by December 31st?”
ScenarioActual BTC PriceYour BetResult
A$99,800YesLOSE
B$100,001YesWIN
C$150,000YesWIN
Scenario A and B are almost identical prices but produce opposite outcomes. Your prediction of “around $100K” was correct; the binary framing punished you.

​
The Skepsis Way

Question: “Where will Bitcoin be on December 31st?”
Your RangeActual BTC PriceResult
95K−95K - 105K$99,800WIN
95K−95K - 105K$100,001WIN
95K−95K - 105K$150,000LOSE
You bet on what you actually believed, “Bitcoin will be around $100K,” and were rewarded for being right.

​
Information Density

​
Binary Markets: One Data Point

A binary market tells you one thing: the probability of crossing a threshold. 
"BTC > $100K by Dec 31" = 45% Yes
That’s it. You don’t know if traders think it’ll be 101Kor101K or 200K if it passes.

​
Continuous Markets: Full Distribution

A Skepsis market shows you the entire probability landscape: 
$80K - $90K:   5%   ░░
$90K - $95K:   15%  ░░░░
$95K - $100K:  25%  ░░░░░░░░
$100K - $105K: 30%  ░░░░░░░░░░  ← Most likely
$105K - $110K: 15%  ░░░░
$110K - $120K: 8%   ░░░
$120K+:        2%   ░
This tells you:
  • Where the crowd thinks BTC will land
  • How confident they are (narrow vs. wide distribution)
  • Where the contrarian opportunities are (underpriced tails)

​
Capital Efficiency

​
Binary: Need Many Markets

To express “BTC will be between 95K−95K-105K,” you’d need to:
  1. Buy “Yes” on “BTC > $95K”
  2. Buy “No” on “BTC > $105K”
  3. Hope both markets have liquidity
  4. Pay fees on multiple trades
  5. Track multiple positions

​
Continuous: One Market, Full Expression

On Skepsis:
  1. Select the 95K−95K-105K range
  2. Place one bet
  3. Done.
Same belief, one transaction, lower fees.

​
Real-World Examples

​
Release Date Predictions

Binary:
  • “Will GPT-5 launch before April 2025?” → Yes/No
  • “Will GPT-5 launch before July 2025?” → Yes/No
  • “Will GPT-5 launch before October 2025?” → Yes/No
  • (Need to bet on multiple markets)
Continuous:
  • “When will GPT-5 launch?” → Select: May - August 2025
  • (One market captures your full belief)

​
Weather Forecasting

Binary:
  • “Will NYC hit 90°F tomorrow?” → Yes/No
Continuous:
  • “What will NYC’s high be?” → Select: 85°F - 92°F
  • (Actually matches how meteorologists think)

​
Economic Indicators

Binary:
  • “Will unemployment be below 4%?” → Yes/No
Continuous:
  • “What will unemployment be?” → See full distribution from 3.5% to 5.5%
  • (Captures market uncertainty, not just a threshold)

​
The Mathematical Edge

Continuous markets using LMSR (our pricing algorithm) have proven properties:
PropertyWhat It Means
Bounded LossMarket makers know their max downside
Infinite LiquidityAlways able to trade, any size
Path IndependenceFair pricing regardless of trade order
Proper ScoringRewards honest probability estimates
Together these properties produce fair, efficient markets.

​
Summary: Why Continuous Wins

AspectBinary MarketsContinuous Markets
ExpressivenessYes/No onlyAny range
InformationSingle probabilityFull distribution
Capital efficiencyMultiple markets neededOne market
Edge case handlingAll-or-nothingNuanced outcomes
Real-world fitRarely perfectNatural fit

​
Try It

Your first continuous prediction: Launch Skepsis →