1. Introduction: The Vision

PrediFin is a specialized subnet on Bittensor designed to create the world’s most accurate, transparent, and objective decentralized financial prediction network, focused exclusively on major liquid assets.

Our vision is to build a “Decentralized Macro Prediction Oracle” that leverages collective intelligence from global AI models to deliver reliable probability forecasts for large-cap assets such as BTC, ETH, SPX, Gold, and DXY.

Unlike traditional financial forecasting platforms that suffer from opacity and potential conflicts of interest, PrediFin uses real market outcomes as the sole judge of performance. By combining Bittensor’s incentive mechanism with rigorous objective scoring, we aim to create a self-improving ecosystem where the best financial AI models continuously rise to the top, delivering high-value predictions to traders, DeFi projects, and institutions.

This proposal details PrediFin’s mechanism design, roles of miners and validators, anti-cheating architecture, and strong commercial potential in the multi-trillion-dollar financial prediction market.

2. Incentive & Mechanism Design

PrediFin’s incentive mechanism is built around objective, verifiable market truth — making it one of the most tamper-resistant and transparent subnets on Bittensor.

Core Design Principle Every prediction is scored solely based on actual future market prices. This creates a pure “Proof of Intelligence” system where superior models naturally receive higher rewards.

Emission and Reward Logic

• The subnet runs validation cycles every 4 hours.
• Miners who achieve higher long-term accuracy receive proportionally higher TAO emissions and network weights.
• Top-performing miners capture the majority of emissions, creating strong competitive pressure.
• A portion of future user subscription revenue will also flow back to high-performing miners.

Anti-Cheating & Transparency Mechanisms

• Distributed Task Generation: Tasks are generated using blockchain height + timestamp + random seed, making pre-arranged collusion extremely difficult.

• Multi-Validator Cross Verification: Each task is scored by multiple independent validators. Final scores use median or weighted average to prevent manipulation.

• Delayed Objective Scoring: Scores are only assigned after real market data is available.

• Public Auditability: All tasks, miner submissions, and scoring results are publicly visible on the

   subnet dashboard.