# FRY: Turning Trading Loses into Productive Assets

By [GH Systems](https://paragraph.com/@gh-systems) · 2025-10-06

---

**TL;DR**: We built infrastructure that converts DEX trading losses into a stablecoin with 7.4x better capital efficiency than traditional approaches.

* * *

The Problem: $50M+ in Daily Wreckage
------------------------------------

Every day, decentralized exchanges generate massive losses from:

*   Liquidations (longs/shorts getting rekt)
    
*   Slippage (price moves against you mid-trade)
    
*   Funding rate payments (perps bleeding money)
    
*   Getting picked off by informed traders
    

Traditional solution? Socialize the losses across all LPs. Everyone loses.

The FRY Solution: Liquidity Rails
---------------------------------

We built a three-tier system that routes losses through optimal paths:

### Tier 1: P2P Matching (1.4 FRY per $1)

If you're paying funding and someone else is receiving it, we match you directly. Cash-settled swap, no token transfers. Both sides mint enhanced FRY.

### Tier 2: Liquidity Rails (1.2-2.2 FRY per $1)

Smart routing across 5+ DEXes (Hyperliquid, Aster, dYdX, GMX, Vertex). Multi-hop paths, liquidity aggregation, efficiency bonuses.

### Tier 3: fryboy AI (0.8-1.0 FRY per $1)

ML-enhanced market maker as fallback. Slippage harvesting, adaptive hedging, reinforcement learning. +11% better than traditional hedging.

**Result**: 2.26 FRY per $1 average (vs 0.5 base rate)

* * *

Why This Works: Native Token Magic
----------------------------------

Here's the key insight: **denominate losses in the DEX's native token, not USD**.

When you measure losses in $HYPE or $USDF instead of USDC:

*   Higher token price → More valuable loss pool
    
*   More FRY minted per dollar of losses
    
*   Creates positive feedback loop
    

**Proof**: 61.5% reduction in funding rate volatility, 7.4x capital efficiency advantage.

* * *

Privacy Layer: zkML + Pedersen Commitments
------------------------------------------

**Problem**: Market makers don't want to reveal their positions/strategies.

**Solution**:

*   **zkML proofs** (EZKL): Prove your model works without showing validation data
    
*   **Pedersen commitments**: Hide collateral amounts while proving you're not overleveraged
    
*   **Federated learning**: Train AI across venues without sharing raw data
    

Bonus: 30% higher FRY minting rate if you provide zkML proofs.

* * *

The Numbers
-----------

**Test Results** (20 wreckage events):

*   $2.33M wreckage processed
    
*   3.74M FRY minted
    
*   221% improvement vs base rate
    
*   57% average liquidity utilization
    

**ML Performance**:

*   +11% hedge ratio optimization
    
*   +15.7% in crisis scenarios
    
*   85%+ regime detection accuracy
    

**Capital Efficiency**:

*   7.4x vs traditional stablecoins
    
*   61.5% funding rate volatility reduction
    
*   70% liquidity rails / 30% AI reserve allocation
    

* * *

Who This Is For
---------------

**DEXes**: Reduce LP losses, stabilize funding rates, attract liquidity

**Market Makers**: Convert losses to FRY, access optimal routes, ML-enhanced hedging

**Liquidity Providers**: Earn FRY from provision, reduced IL, confidential positions

* * *

The Tech Stack
--------------

*   **Routing**: Dynamic programming for optimal paths (up to 3 hops)
    
*   **Matching**: Cash-settled funding swaps (no token transfers)
    
*   **AI**: Reinforcement learning + regime detection
    
*   **Privacy**: EZKL zkML + Pedersen commitments
    
*   **Contracts**: Solidity on Arbitrum (ready for audit)
    

All production-ready. All open source.

* * *

What Makes This Different
-------------------------

**Traditional stablecoins**: Backed by fiat or crypto reserves **Native stablecoins** (USDF/USDH): Backed by DEX native tokens **USD\_FRY**: Backed by _wreckage_ (trading losses)

We're not competing with USDC. We're infrastructure for native stablecoin DEXes to recycle losses productively.

* * *

Roadmap
-------

**Q1 2026**:

*   10+ DEX integrations
    
*   $50M+ TVL
    
*   500+ Agent B instances
    

**Q2 2026**:

*   Cross-chain (Solana, Base)
    
*   Advanced ML (transformers)
    
*   Options market
    

* * *

Try It
------

**Website**: [https://aidanduffy68-prog.github.io/USD\_FRY/](https://aidanduffy68-prog.github.io/USD_FRY/) **GitHub**: [https://github.com/aidanduffy68-prog/USD\_FRY](https://github.com/aidanduffy68-prog/USD_FRY) **Docs**: Full technical whitepaper available

Built by liquidity engineers. Powered by Greenhouse & Company.

* * *

_The first wreckage-backed stablecoin. Because losses shouldn't be wasted._ 🍟

---

*Originally published on [GH Systems](https://paragraph.com/@gh-systems/fry-turning-trading-loses-into-productive-assets)*