Sportix.AI
WHITEPAPER
Solana · Fan Consensus Prediction Protocol

Turning fan judgment
into verifiable consensus.

Sportix is a fan-consensus sports prediction protocol on Solana, designed from the ground up with no real-money wagering of any kind. Every prediction is scored and anchored on-chain as a tamper-proof hash — independently verifiable by anyone, with no need to trust Sportix.

On-chain anchoring SPL Memo · no custom contract No proxy-signing · PII-free Non-custodial · non-wagering
01

Abstract

Fans submit structured predictions on real matches; the protocol aggregates them into transparent crowd consensus, scores each prediction against verified outcomes, and anchors a tamper-proof record of every prediction onto Solana.

Sportix combines AI-assisted match and player intelligence with crowdsourced fan consensus, turning the collective judgment of fans into a trustworthy, verifiable crowd-intelligence data asset. The protocol's core trust claim is independent verifiability: every on-chain record carries a SHA-256 hash of the prediction's content, and anyone can recompute that hash from the original prediction — using the protocol's public, deterministic rules — on a public block explorer to confirm the record has not been tampered with.

02

The Opportunity

Every week, hundreds of millions of fans form judgments about matches — yet these collective insights are scattered, unrecorded, and unverifiable. Most existing "prediction" products are built on real-money wagering, carrying heavy regulatory burdens, excluding many jurisdictions, and rewarding the size of a stake rather than the quality of insight.

Sportix takes a different path: it treats prediction as fan participation and a display of skill, not financial speculation.

03

What Sportix Is / Isn't

Sportix is

A fan-consensus prediction and sports-intelligence platform. Users predict outcomes (e.g. win / draw / loss) using non-monetary Contribution Points (CP), scored by long-term accuracy and ranked by skill (Fan Rank).

Sportix isn't

Not a real-money wagering product, not a prediction market, not a market-making or speculation venue. Users never risk real money against each other; rewards come from the protocol's Reward Pool, not from other users' losses.

There is no fiat on-ramp at launch, and predictions carry no real-money outcome. A disclaimer is permanently displayed across all intelligence, prediction, and ranking pages.

04

Architecture · Five Layers

Data flows upward, from raw ingestion to final market output and the user interface.

L1
Data IngestionMulti-source ingestion of match and player data; each record carries source / ingested_at / content_hash, deduplicated by a unique constraint.
L2
Intelligence Engines · MIE / PISTransforms raw data into structured, evidence-backed intelligence with confidence intervals.
L3
Prediction · Scoring · RewardsStructured predictions, non-monetary CP, deterministic scoring (DPV), hit / miss outcomes.
L4
B2B DataAggregated, anonymized consensus and intelligence delivered via API to downstream consumers.
L5
Web InterfaceThe user-facing web application plus an internal admin interface.
Stack: pnpm + Turborepo monorepo · Next.js (frontend) · NestJS (backend) · Python/FastAPI (intelligence) · Privy (wallet & auth) · Solana official SPL Memo (on-chain records) · PostgreSQL.
05

Intelligence Engines

MIE · Match Intelligence Engine

Produces structured, evidence-backed match intelligence. Every output carries a confidence band (low / medium / high) and a non-empty evidence set. Probabilities are presented as a win/draw/loss triple with confidence intervals — a point estimate is never disguised as a definitive conclusion.

PIS · Player Intelligence Stream

Produces player-by-player intelligence and profiles.

Integrity principle (a hard rule): AI-generated text is never the source of truth for statistical data. Statistical facts come only from licensed structured data sources; the AI layer only summarizes and contextualizes, and any AI-generated content is labeled as such.
06

Prediction · Scoring · Rewards

07

On-Chain Anchoring · Solana

Sportix uses the standard, widely-used SPL Memo program to anchor a tamper-proof record of every prediction on Solana mainnet — without deploying any custom on-chain program for anchoring, keeping the trust surface of the anchoring path to a minimum.

Anchored memo payload · PII-free
SX-v1|prediction_id|match_id|outcome_key|confidence|iso_timestamp|sha256=<prediction_hash>

The payload contains no name, email, or wallet identity; user identity, where present, appears only as a wallet public key (the signer). The memo prefix format is frozen and stable across versions.

Two signing modes

PROJECT-ANCHORED

The project's fee-payer wallet signs and pays for the anchoring transaction. The signer is honestly the project fee-payer — the protocol does not claim the transaction was signed by the user.

USER-SIGNED

The user co-signs their own prediction with their own wallet while the project sponsors the fee, establishing an on-chain record verifiably belonging to that user.

Layered verifiability (a core integrity property): embedded-wallet users co-sign their own predictions (the on-chain co-signer is the user's wallet, independently verifiable); external-wallet users are anchored by the project with identity bound off-chain at login. The protocol never impersonates a user's wallet with the project wallet (no proxy-signing) — under either path, who signed on-chain is recorded faithfully.

On-chain anchoring counts are reported as the count of successfully anchored memos, kept strictly separate from off-chain database counts and never presented as on-chain data.

08

Data Integrity

09

Deterministic Canonicalization

Before predictions enter aggregation, scoring, or the chain, their content passes through a deterministic, cross-language canonicalization that produces an outcomeKey.

The key property: the same prediction produces a bit-for-bit identical hash across different languages and layers (TypeScript backend, Python intelligence layer, Rust on-chain program), locked by cross-language test fixtures — eliminating any on-chain / off-chain divergence caused by serialization differences.
10

Consensus Aggregation · Anonymization

Crowd consensus is aggregated from predictions by proposition and outcome key, producing the count and share of each outcome. Consensus aggregation never returns a user identifier — anonymization of consensus data is a hard rule. Aggregate views expose only the distribution, never "which user predicted what."

11

Security & Trust Boundaries

12

Vision

To become the trustworthy, verifiable crowd-intelligence infrastructure for sports.

Every day, fans worldwide generate an enormous volume of match judgments — yet today these judgments are unrecorded, unverifiable, and impossible to measure. Sportix turns each prediction into a structured, scored, on-chain-verifiable record, settling scattered fan intuition into a consensus data asset that thickens over time and is weighted by accuracy.

When a prediction network grows large enough — and every prediction is verifiable and scored against real outcomes — the crowd consensus it aggregates becomes a uniquely credible signal source for sports. Not because some authority says so, but because anyone can independently verify, on-chain, where it came from and whether it has been tampered with.

13

Business Model

Sportix's value comes from trustworthy, anonymized, skill-weighted crowd consensus and intelligence data. Monetization is built around this data asset:

B2B Data API

Aggregated, anonymized consensus distributions and intelligence delivered via API to media and content platforms, sports organizations and clubs, and data & research analysts. The output is crowd-scale anonymized signal — never any personal identity data.

Value-added / Subscription

Premium capabilities for power users and institutions (finer-grained intelligence, historical track-record and ranking analytics) offered on a subscription / value-added basis.

14

Value Flywheel · Contribute → Reward

At the core is a flywheel where user contribution and protocol value reinforce each other:

Users contribute predictions

Fans submit structured predictions based on the intelligence layer, spending non-monetary CP.

Settles into a consensus data asset

Predictions are anonymized, aggregated, scored against real outcomes, and weighted by skill — forming verifiable crowd consensus data.

B2B monetization

The data asset is delivered through the B2B API to downstream consumers, generating protocol revenue.

Reward contributors

The value the protocol generates is returned, through the protocol's reward mechanism (Reward Pool), to users who contribute high-quality predictions. The reward recognizes the act and skill of contributing accurate predictions — not a return on any capital — and users bear no real-money risk at any point.

Incentivize more, stronger predictions

Rewards drive continued contribution, growing the consensus data in scale and quality and making the asset more valuable — turning the flywheel.

The key to the flywheel: it rewards insight and accuracy, not stake size. The more accurate the predictor, the more reward they earn — so the data asset converges toward high quality.
15

Roadmap · Direction

Development follows the order of "first grow consensus data large and trustworthy, then monetize around the data asset":

01
Consensus NetworkGrow the scale and coverage of user predictions, settling verifiable, skill-weighted crowd consensus data; deepen the intelligence layer and on-chain verifiability.
02
Data Asset MaturityAs scale and quality grow, deepen DPV scoring, ranking, and intelligence so anonymized consensus becomes a mature, deliverable data product.
03
B2B Monetization & RewardLaunch the B2B data API and value-added services, monetize the data asset, and return value to contributors through the Reward Pool — keeping the flywheel turning.

Roadmap items are directional and may change. Any custom on-chain program is gated by an independent security audit and multisig approval before mainnet deployment.