project screenshot 1
project screenshot 2
project screenshot 3

Flume

Smarter, fully automated liquidity manager (ALM) for Uniswap v3/v4. Minimizes impermanent losses while maximizing LPs profitability.

Flume

Created At

ETHGlobal Bangkok

Project Description

Flume is an advanced Automated Liquidity Manager (ALM) tailored for Uniswap v3 and v4, offering a simplistic, Uniswap V2-like deposit experience while handling position building and management on behalf of the end-user.

The Challenge for Liquidity Providers

Many liquidity providers (LPs) prefer not to manage their ranges themselves due to the required expertise and the risk of impermanent losses. Until now, no ALM has successfully provided a profitable, V2-like farming experience on Uniswap V3.

Identified Pain Points:

  • Low data quality
  • Flawed range determination
  • Ineffective rebalancing triggers
  • Excessive position rebalancing
  • Rebalancing slippage

Projected Performance Improvement

Through hypothesis testing, we anticipate Flume can boost LP performance by 1.3x to 5x, depending on the pool's volatility or stability.

Flume's Key Advantages:

  1. Enhanced Data Processing:
  • Aggregating DEX (Uniswap) and CEX (Binance, Bybit, Coinbase) data
  • Computing Volume Weighted Average Price (VWAP) for informed decisions
  1. Dynamic Range Finding:
  • Utilizing a time-series forecasting model trained on live price action to predict volatility and momentum
  • Automatically adjusting the optimal range regardless of the pool's active fee and tick size
  1. Optimized Rebalancing:
  • Leveraging aggregator depth for reduced slippage, ensuring rebalancing swaps cost less than pool executions

Key Benefits:

  • Increased duration, decreased governance (better ranges for a long time)
  • Adaptive Market Response: Instant adjustment to changing market dynamics
  • MEV Resistant

How it's Made

Tech stack:

  • Contracts: forked from Arrakis (https://www.arrakis.finance), selected on security, optimization, quality and features criteria (vs Gamma Strategies, Ichi, Beefy CLM and DefiEdge)
  • Back-end: Python based, using FastAPI and stats libs used for the RangeFinder forecasting model. Chomp indexer is used for collecting CEX/DEX data from a single configuration file (https://github.com/AstrolabDAO/chomp) and expose it as an API for the front-end to use
  • A minimalistic Vue3 front-end will be used as proof-of-concept, allowing deposit and withdrawal of funds on the ALM vaults
  • MEV resistance: the rebalancing triggers are obfuscated and not time-sensitive, and transactions will not sit in mempools (use of MEV resistant RPCs if applicable or low latency passing to L2 sequencers).
background image mobile

Join the mailing list

Get the latest news and updates