Thread by @PotionLabs

'The Kelly Machine'

A new approach to risk management in DeFi.

👇👇👇

potion-protocol.medium.com/introducing-th…

❓What is it?

An AMM protocol for decentralized creation and pricing of risk management contracts (eg insurance, puts, calls).

❓How does it work?

You can play as seller or buyer. The system uses the Kelly Criterion to find optimal pricing and apply automatic risk management.

How can it help me❔

📌Buyer: manage the risk of your holdings; get exposure to new assets📈

📌Seller: generate alpha from your holdings💸

Can I trust it❔

📌 It's trustless🔑
📌 Based on Kelly Criterion🤓
📌 Comprehensive back-testing🏁
📌 Automated risk mgmnt

🌊 Sustainable Liquidity:

The Kelly Machine facilitates a virtuous circle of value for market participants

Long term alpha expectation for LPs → Deeper liquidity → Better services for Buyers

Better services for buyers → Higher demand for liquidity → Better LP experience

🧐 Why not Black-Scholes Model (BSM) to price risk?

📌 Empirical results from our testing 👉 BSM misprices risk for major crypto assets and can lead to LP bankruptcy.

📌 BSM Not designed to handle fat tailed assets 👉 ...and crypto assets are very fat tailed!

💡 Utilization based pricing 👉 Intuition:

📌 To offset the higher risk of bankruptcy when more capital exposed

👇👇👇

📌 Charge higher premiums and build a larger safety net.

But which particular curve to use?

⏩ Enter the Kelly Criterion:

📌A mathematical framework for optimal capital allocation under uncertainty

📌It optimizes for “systematic risk taking”, aiming for long term survival and max capital growth

📌It provides a robust link between premiums and allocation/utilization

⏪ Reverse Kelly Criterion:

📌We use the KC "backwards", to produce risk premiums on the basis of incoming order_size

📌From this process we are able to generate "Kelly optimal" bonding curves

📌We found a very simple formula to describe these curves, implementable on chain

🔥The Kelly Machine:

📌A capital manager for sellers with built-in automated risk management.

📌A market maker for buyers with deep liquidity and open reliable pricing.

📌Facilitates natural market convergence point around Kelly Optimal.

📌Easy to use and accessible to all

🏁 Preliminary Backtest results: Alpha

📌 When LPs stay long term, they can mathematically expect asymptotic positive growth rate over time (CAGR or ALPHA)

📌 LPs are able to influence the expected alpha, by setting their custom curves: Programmable Alpha

🏁 Preliminary Backtest results: Risk

📌 Kelly priced options outperform Black Scholes significantly in risk management.

📌 LPs can further tune their curves to control the particular risk exposure they feel comfortable with.

🌊 Liquidity architecture:

📌 We introduce a new distributed bonding curve model.

📌 It has the security advantages of a peer to peer model, with the scalability and price convergence of peer to pool.

📌 We call it peer to virtual pool.

🌍 Routing over Distributed Bonding Curve:

📌 In the network each LP will have its own dedicated curve and utilization 👉 LPs will quote different prices for same order.

📌 We built a distributed order router to ensure buyers could access lowest available prices at all times

🌠 Emerging Distributed Bonding Curve:

📌 From the perspective of users, the entire network behaves as a single LP, with a single bonding curve.

📌 We call this the emerging total market bonding curve.

See below how the router dynamically recalculates lowest available price:

🏁 Benchmarking Kelly Machine Vs Black-Scholes:

📌 Higher LP win-rates

📌 Higher LP survival rates for crypto

📌 Lower max_draw_down

📌 Better Schelling point discovery

More information will be released over the coming weeks and months, gearing up for V0 release 🔥

More info in medium post & community call!

Want to participate in solving some of the coolest and toughest problems in risk management in DeFi?

Join the Potion community today ❤️

discord.potion.finance