Markets
Every market on Dekant is a continuous market over a numerical range. There are no Yes/No markets and no fixed-outcome categorical markets. The only market shape is a number-line question with a bin grid behind it.
What a market looks like
A market is defined by four things at creation time:
| Field | Example | What it does |
|---|---|---|
| Question | "What will SOL be on July 1, 2026?" | The natural-language prompt traders see. |
| Range | $40 to $250 | The lower and upper bounds for plausible outcomes. |
Bin count (N) | 64 | How finely the range is divided. Each bin is (b − a) / N wide. |
| Deadline | 2026-07-01 00:00 UTC | When trading freezes and the oracle resolves. |
The range and bin count together determine resolution granularity. Range $40-$250 with N = 64 gives bins of ~$3.30 each. With N = 256 (the protocol cap) the same range gives ~$0.82 per bin.
Two ways to trade
Inside a continuous market you can place trades in two shapes:
- Bin buy - buy a single bin like betting on one slot. Useful when you have a sharp opinion about exactly one number landing inside one specific bin.
- Distribution buy - set a center (
μ) and conviction (σ), and your stake is allocated across all bins weighted by the Normal PDF. This is the "draw a curve" flow that Dekant is built around, and what almost all real trading uses.
The math is the same on both sides. A distribution buy is just a vector of bin buys done atomically with proportions set by (μ, σ). See the math docs for the closed-form derivation.
Why bins?
A theoretically clean continuous prediction market would let traders post arbitrary payout functions over ℝ. That's the design Paradigm proposes (see the math docs for the comparison). It's elegant, but it makes solvency hard to track on-chain.
Dekant discretizes the range into N bins and treats the position vector as a finite-dimensional ℝᴺ. With finite dimensions, the L₂ invariant ‖x‖₂ = k automatically bounds each xᵢ by k, which means the vault always has enough collateral to honor every redemption. Solvency is provable with no extra constraint. See the math docs for the full solvency theorem.
The trade-off: you can only express belief shapes that decompose into bins, and resolution lands in exactly one bin. In practice this is fine - typical curves spread across 5-12 bins near the peak, and bin sizes are small relative to real-world price movements.
Picking N at market creation
Most markets land at N = 32 or N = 64. The trade-off:
Few bins (low N) | Many bins (high N) | |
|---|---|---|
| Per-bin resolution | Coarse | Fine |
| Distribution-trade compute cost | Cheap | Higher |
| Liquidity per bin | Concentrated | Spread thinner |
| Bin-boundary problem | Worse | Better |
N is fixed at market creation; it cannot be changed mid-market.