Phase 0 prototype · Local anvil · Not on public testnet yet

Liquidity, where it trades.

A binary prediction market with concentrated LP bands. The curve traders interact with is unchanged from flat LMSR; the capital backing it doesn't have to be spread across the parts of the curve nobody touches.

// Max η (tight band)

17.3×

// Parity tests

65/65

// Math tolerance

1e-9

Read Whitepaper → View on GitHub ↗

tessera@phase-0 ~ % pytest tests/

▸ # Reference simulator + fixed-point parity
tests/test_dclmsr.py ............ 29 passed
tests/test_fixedpoint_parity.py . 36 passed
 
▸ # Capital efficiency, exact from mpmath ref
  η[0.05, 0.95] = 0.235×  // wide band
  η[0.45, 0.55] = 3.454×
  η[0.49, 0.51] = 17.326× // tight band
  η[0.495,0.505] = 34.656× // extreme
 
▸ # LP strategy Monte Carlo, p* = 0.62, 30 runs
  aligned [0.55,0.70]   +7.59%
  flat    [0.02,0.98]   −2.93%
  misalign[0.30,0.45]  −28.26%
 
65 passed in 0.16s
▸

Same capital. More market.

Flat-AMM prediction markets spread liquidity across probabilities that never trade. Tessera concentrates it where LPs have conviction. Numbers below are from the reference simulator at simulator/veritas_sim/dclmsr.py.

Metric

Flat LMSR (Polymarket-style)

Pari-mutuel (Kalshi-style)

Tessera DC-LMSR

Capital efficiency η at [0.49, 0.51]

1.0×

n/a (no continuous LP)

17.3×

Capital efficiency η at [0.45, 0.55]

1.0×

n/a

3.45×

LP chooses price range

No — full (0,1) support

No LP role

Yes — any [p_min, p_max]

Trader-facing curve

LMSR

Pool odds

LMSR (unchanged)

Multi-LP layering

Single shared pool

Single pool

Stacked bands at price p

// Comparison covers liquidity mechanics only. Resolution layer is a separate spec — see MMAN below. Polymarket and Kalshi run production resolution today; Tessera does not yet.

Two primitives. One built, one specified.

Most prediction markets fail at the seams between resolution and liquidity. Tessera treats them as one design problem. The liquidity primitive is implemented and tested; the resolution primitive lives in the whitepaper.

// 01 — DC-LMSR · Implemented

Dynamic Concentrated LMSR.

LPs deposit into specific probability bands. Trader sees a vanilla LMSR curve; LP gets concentrated exposure. Capital required for a given band is derived in closed form (ADR-003), with the exact formula verified to 1e-9 across Python and Solidity.

Max η (tested)17.326×
LP band[p_min, p_max]
BootstrapFirst LP anchors
Math parity1e-9 vs mpmath
Solidity portPRBMath UD60x18

Read DC-LMSR spec →

// 02 — MMAN · Whitepaper only

Multi-Modal Attestation Network.

Specified resolution layer: three independent paths attest outcome data in parallel — zkTLS proofs of HTTPS retrieval, TEE-hosted multi-model consensus, and quadratic-staked jurors on dispute. Phase 0 uses a mock resolver; MMAN itself is not yet implemented.

StatusSpec only
Paths (designed)3 parallel
Juror weighting√(stake)
Dispute window10 min (target)
Phase 0 stand-inMockResolver.sol

Read MMAN spec →

Where the protocol actually is.

Tessera is a research-stage protocol. The liquidity primitive is implemented and tested locally. Public testnet, audit, and mainnet are subsequent phases.

// Phase 0 · current

Prototype, local

DC-LMSR market contract, mock resolver, reference simulator, frontend on local anvil. 65 parity tests passing.

// Phase 0.5 · next

Correctness gaps

Per-LP inventory attribution. Exact band-boundary crossings. Tick-walking instead of linear scan. Withdrawal mechanics.

// Phase 1 · pending

Public testnet + MMAN v1

Sepolia deployment. First end-to-end resolution path (likely zkTLS only). Audit scoping. Legal review.

// Phase 2 · TBD

Mainnet candidate

Full three-path MMAN. Audit complete. Jurisdiction-compliant deployment. Real USDC, real markets.