Djinn
SN103Cryptographic web attestation that proves any website showed specific content at a specific time
An encrypted prediction marketplace where analysts sell signals that nobody, not even the platform, can ever see. Track records are cryptographically verifiable. The protocol runs itself.
// Buy intelligence you can trust.
Djinn is a marketplace for buying and selling predictions, primarily for sports betting right now. Analysts (called "Geniuses") post encrypted predictions with collateral locked up as a guarantee. Buyers (called "Idiots") purchase access to those signals based on the analyst's provable track record, without ever seeing the actual picks until they pay.
The simple version: Imagine a tipster marketplace where the tips are sealed in cryptographic envelopes. You can verify the tipster's past win rate without seeing any individual tip. When you pay, only your envelope opens. The marketplace operator never sees any of the tips.
Centralized equivalent: Think Cappers or Action Network, but the platform can't steal signals, fake track records, or front-run your bets.
How it works:
- Miners ("Geniuses") verify real-time betting line availability using TLSNotary proofs, confirming that odds are actually available at the stated price.
- Validators hold fragments of encryption keys using Shamir secret sharing. They coordinate multi-party computation (MPC) to decrypt signals only for verified buyers, and they attest game outcomes through consensus.
- The problem it solves: Prediction marketplaces are plagued by fake track records, signal theft, and platform manipulation. There's no way to prove your picks are real without revealing them, and no way to buy tips without trusting the seller.
- The opportunity: The global sports betting market generates hundreds of billions annually. Verified analyst signals are a premium product with no trustworthy marketplace.
- The Bittensor advantage: Decentralized key management means no single entity can access signals. Validators collectively hold encryption keys, so compromising one node reveals nothing. This is security through architecture, not policy.
- Traction signals: 853 commits across 3 contributors with 85+ commits per week in the last month. 1,183 automated tests (857 validator, 326 miner). Smart contracts deployed on Base chain. Public web attestation service live at djinn.gg/attest. Jason Calacanis highlighted the team as "very legit" by Mark Jeffrey.
Category: Financial Forecasting and Trading Signals | Centralized Competitor: Action Network, Cappers, OddsJam
Djinn is doing something genuinely novel in the Bittensor ecosystem. Instead of competing in inference, training, or data, it's building a cryptographic marketplace for information itself. The core insight: intelligence has value, but selling it destroys it. Djinn solves this with encryption that persists even after the sale.
Mechanism:
The protocol has two layers. On Base chain, Solidity smart contracts handle escrow, collateral, audit settlement, and credit accounting. On Bittensor, validators and miners handle the cryptographic operations: TLSNotary proofs for line verification, Shamir secret sharing for key management, and MPC for decryption.
When an analyst posts a signal, it's encrypted and committed to Base chain with a timestamp. The encryption key is split into fragments distributed across validators using Shamir's scheme. When a buyer purchases access, validators coordinate an MPC ceremony to decrypt only that buyer's copy. The analyst's actual picks are never visible to Djinn, the validators, or anyone except the buyer.
This is a 41,330KB codebase, one of the larger subnet repos in the ecosystem. The repository spans Solidity contracts (Foundry), a Next.js 14 frontend, Python validator and miner packages, and a Graph subgraph for indexing. Development velocity is exceptional: 85+ commits per week over the last 4 weeks, all primarily from Philip Maymin. Recent commits show active work on signal purchasing flows, odds validation, ESPN score integration, and miner line matching.
The web attestation service at djinn.gg/attest is an interesting secondary product. For a burn of 0.0001 TAO in SN103 alpha, anyone can generate a TLSNotary proof that a website showed specific content at a specific time. Use cases include legal evidence, journalism verification, and governance transparency.
Financially, Djinn is small but well-distributed. At 15,485 TAO market cap, it's early-stage. The Gini coefficient of 0.483 is the lowest of any active subnet in the ecosystem, meaning holder distribution is remarkably even. HHI of 0.017 confirms minimal concentration. Root proportion of 0.242 shows the majority of its value comes from organic demand rather than protocol subsidy.
- Single developer: Philip Maymin accounts for the vast majority of recent commits. While prolific (85+ commits/week), this creates key-person risk. The project is highly dependent on one individual.
- Small market cap: At 15,485 TAO, liquidity is thin. The pool holds roughly 5,700 TAO. Significant positions are hard to enter or exit without slippage.
- Sports-first focus: The current product is narrowly focused on sports betting signals. Expanding to financial markets, crypto, or other prediction domains would significantly increase the addressable market but remains unproven.
- Execution complexity: The stack spans five technologies (Solidity, Python, TypeScript, AssemblyScript, TLSNotary). Maintaining and scaling this across a small team is ambitious.
