Parl: Parimutuel Prediction Markets Protocol

Prediction markets are powerful tools for aggregating information and forecasting future events. However, existing implementations face a fundamental trade-off between liquidity and market breadth. AMM-based markets (like Polymarket) require active liquidity provisioning, which concentrates activity on a handful of high-volume markets. Long-tail markets — the thousands of niche or short-lived events — remain underserved.

Parl introduces a parimutuel pool mechanism that eliminates the need for liquidity providers entirely. Each market operates as an independent pool: all bets are collected, and winners split the pool proportionally. This design means any market — no matter how niche — can exist from day one with zero liquidity bootstrapping.

This paper presents the Parl protocol v0.3: its mathematical model, smart contract architecture (PoolEngine v2, MarketFactory, ParlOracle), resolution framework (Optimistic Oracle + Chainlink Functions), auto-categorization system, data API and indexer architecture, comparison with Polymarket, token economics, and value loop game design. The protocol is deployed on Avalanche Fuji testnet with full end-to-end functionality including permissionless market creation, auto-categorization, and a Polymarket-inspired user dashboard with probability displays.

1.1 The Liquidity Bottleneck

Prediction markets require liquidity to function. In traditional order-book markets, market makers provide continuous quotes. In AMM-based markets, liquidity providers deposit assets into pools to enable trading. Both models suffer from a cold-start problem: a market with no liquidity is functionally dead.

This creates a power-law distribution of market activity: 80%+ of volume concentrates in fewer than 5% of markets. Niche, timely, or experimental events — the very class of events where prediction markets create the most informational value — never gain traction.

1.2 The Permission Barrier

In most prediction market platforms, creating a market requires either platform approval (Polymarket, Kalshi) or significant technical expertise (Augur, Gnosis). This gatekeeping limits the diversity of markets and slows the platform's ability to react to current events.

Parl solves this with permissionless market creation via a MarketFactory contract. Any user with an EVM wallet can create a market in a single transaction, paying only gas plus a minimal anti-spam creation fee. No approval, no KYC, no technical barriers.

1.3 Market Maker Incentives

AMM-based prediction markets require LPs to deposit capital into outcome-specific pools. This exposes LPs to adverse selection — informed traders bet against mispriced pools, and LPs absorb the loss. Over time, LPs demand higher fees or exit, reducing market efficiency.

Parl solves both problems by eliminating the concept of a "liquidity provider" entirely. In a parimutuel system, the bettors are the pool. Every new bet increases the pool, and the odds automatically adjust to reflect the new distribution.

2.1 Historical Context

Parimutuel betting (from French pari mutuel, "mutual stake") originated in 19th-century horse racing. All bets on a race are pooled together; after deducting a commission, the pool is divided among winning tickets. The odds are not set in advance — they emerge from the distribution of bets.

2.2 Mathematical Model

2.3 Key Properties

2.4 Implied Probability & Price Display

Parl calculates and displays implied probabilities for every outcome in every market, matching the pattern used by Polymarket. The probability of outcome k winning is:

These probabilities are computed server-side by the Parl Indexer and served through the API. The frontend renders them as visual bars with percentage labels, giving users an intuitive sense of market sentiment. A market with 70% probability on outcome A and 30% on outcome B immediately signals crowd wisdom without requiring users to compute pool ratios manually.

3.1 Design Philosophy

Parl is designed as a minimal protocol layer. The core contract handles three operations: create markets, place bets, and resolve + claim. Everything else — frontend UX, data indexing, notifications — is built on top as separate services.

3.2 Smart Contracts (V2)

Parl v2 introduces a three-contract architecture that separates concerns and enables permissionless market creation:

3.3 Off-Chain Services

4.1 Resolution Model

Each market has a designated resolver address. Only this address can call resolveMarket(marketId, outcomeIndex, proof). After resolution, winners can claim their payout at any time via the claim(marketId) function.

4.2 Optimistic Oracle (ParlOracle)

The ParlOracle implements a propose-dispute-resolve pattern inspired by UMA's optimistic oracle design but simplified for Parl's use case:

4.3 Chainlink Functions (Planned)

For automated, data-driven resolution, Parl includes ParlAutoResolver, a Chainlink Functions consumer contract. This enables markets to resolve automatically based on real-world API data:

4.4 Oracle Options Summary

5.1 Motivation

In Parl v1, only the contract owner could create markets. While this was safe for initial testing, it fundamentally limited the protocol's utility. V2 introduces the MarketFactory contract — a permissionless gateway that any user can call.

5.2 Factory Design

5.3 Anti-Spam & Economics

Market creation requires a 0.01 AVAX fee (~$0.0003 at current testnet prices, negligible in value but sufficient to prevent spam attacks). The fee is configurable by the protocol owner. Collected fees accumulate in the factory contract and can be withdrawn by the owner via withdrawFees().

Excess payments are automatically refunded to the user. If a user sends 0.1 AVAX for a 0.01 AVAX fee, 0.09 AVAX is returned immediately in the same transaction.

6.1 Category Detection

Parl implements a server-side auto-categorization system that analyzes market outcome strings and assigns a category. Categories are stored in the database and served via the API, enabling Polymarket-style category filtering in the frontend.

6.2 Category Hierarchy

The categorization engine uses keyword matching with a priority hierarchy to avoid false positives. For example, "Party A wins election" contains "win" (sports keyword) but "election" (politics keyword) takes priority:

6.3 Frontend Integration

The frontend displays category badges with distinct colors:

Users can filter markets by category via tab buttons in the frontend, and the API supports ?category=sports query parameters for programmatic filtering. The /api/categories endpoint returns all categories with market counts.

7.1 Indexer

The Parl Indexer is a Rust binary that connects to Avalanche Fuji via WebSocket/HTTP and listens for PoolEngine events (MarketCreated, BetPlaced, MarketResolved, ClaimProcessed). It maintains a PostgreSQL database with 6 tables:

7.2 API Endpoints

7.3 Simplified Markets Format

The /api/simplified-markets endpoint returns a lightweight JSON format designed for frontend tables, inspired by Polymarket's API format:

[
  {
    "id": "9a228815-...",
    "question": "BTC exceeds $100k | BTC below $100k",
    "outcomes": ["Yes", "No"],
    "status": "active",
    "category": "crypto",
    "total_pool_avax": "0.0100",
    "volume_avax": "0.0100",
    "probabilities": [55.0, 45.0],
    "created_at": "2026-07-02T00:26:50Z"
  }
]

8.1 Fee Model

Parl generates revenue through platform fees on each market. Market creators set the fee at creation time, expressed in basis points (100–500 bps = 1%–5%). Fees are deducted from the total poolbefore payout distribution.

Fee Tiers

8.2 $PARL Token Model

$PARL is the native protocol token, designed to align incentives across market creators, bettors, stakers, and the broader ecosystem. No token is deployed on testnet; all mechanics below are planned for mainnet launch.

Token Utility

Token Supply & Distribution (Proposed)

Token Flywheel

Prediction markets, at their core, are games of coordination and prediction. The parimutuel model unlocks game-design primitives that AMM-based markets cannot replicate. This section explores the engagement loops, social mechanics, and game systems that turn Parl from a financial protocol into a platform people want to use.

9.1 Pool Boosters (Tipping)

Anyone can deposit additional funds into a market pool aftercreation. This increases the total payout for winners without affecting the distribution of existing bets.

9.2 Outcome NFTs (Bets as Collectibles)

Every bet is minted as an NFT — a non-transferable or optionally transferable token representing a position in a market. This turns betting into a collecting mechanic.

9.3 Manager Mode

Inspired by the engagement systems behind EA Sports, Monopoly Go's event loops, and NBA Top Shot's collection mechanics, Manager Mode is a gamified layer over Parl markets:

9.4 Bettor Reputation & Trust Scores

Every address builds an on-chain reputation based on prediction accuracy, volume, and longevity. This reputation feeds into market creation trust scores.

9.5 Social & Viral Mechanics

9.6 The Flywheel

Polymarket is the dominant prediction market platform by volume, using a CLOB (Central Limit Order Book) model with AMM-like liquidity provision via CTF (Conditional Token Framework). Below is a detailed comparison with Parl's parimutuel approach:

10.1 Key Takeaways

Polymarket's CLOB model provides deep liquidity for high-volume markets (like political events) and enables sophisticated trading strategies. However, it suffers from the cold-start problem — new markets require LP deposits before any trading can occur.

Parl's parimutuel model is fundamentally better suited for long-tail markets — the thousands of niche events where prediction markets create the most informational value. A market for "Will it rain in Jakarta tomorrow?" can exist with just 0.01 AVAX in the pool and still function perfectly.

Additionally, Parl's permissionless creation, game mechanics, and open-source frontend make it more accessible for developers and communities to build on top of the protocol.

Parl introduces a novel approach to on-chain prediction markets by leveraging the parimutuel mechanism. By eliminating the need for liquidity providers and AMM curves, Parl enables a fundamentally different market structure — one where any event, no matter how niche, can have a liquid market from the moment of creation.

The protocol is live on Avalanche Fuji testnet (v2 contracts) with full functionality: permissionless market creation via MarketFactory, auto-categorization across six categories, an Optimistic Oracle for honest resolution, Chainlink Functions integration for automated data-driven resolution, a comprehensive data API, and a Polymarket-inspired user dashboard with probability displays.

With the inclusion of token economics designed to align incentives and game mechanics inspired by the best engagement systems in consumer technology, Parl is positioned not just as a financial protocol — but as a platform that people genuinely enjoy using.