Skip to main content

Agent interface for on-chain protocols.

Project description

linting: pylint code style: black testing: pytest license: Apache 2.0 DELV - Terms of Service

testing: coverage docs: build

agent0

A Framework for Hyperdrive Trading Strategies

This repo by DELV contains tools for you to deploy automated trading agents, perform market simulations, and conduct trading research. Read on for more info or jump to the quickstart guide. This docs page can be found via https://agent0.readthedocs.io/en/latest/.


Hyperdrive Background

Hyperdrive is an automated market maker (AMM) protocol that enables fixed-rate markets to be built on top of arbitrary yield sources. It deploys assets into those yield sources and wraps them as Hyperdrive positions represented as hy[Tokens] that trade at a discount (%) and can be redeemed for their full face value at maturity. Since the hy[Token]’s initial cost and value at maturity are known upfront, this discounted purchase represents a fixed rate of return.

Abstracting interest rate dynamics into a single price opens up interesting Market Dynamics while giving users the freedom to employ a number of Trading Strategies.

To ensure that a balanced market exists where the market price can be increased and decreased by supply and demand, the AMM supports three basic operations:

  • Opening Longs, which provides exposure to the fixed rate by purchasing hy[Tokens] at a discount to their face value for the price of forgoing the variable rate. Longs pay trading fees to the pool.

  • Opening Shorts, which provides exposure to the variable rate generated by the capital that backs hy[Tokens] for the price of paying the fixed rate. Shorts pay trading fees to the pool.

  • Providing Liquidity, which facilitates Long and Short trading by automatically taking the other side of user positions, and charges trading fees. Part of the pool's fees may go to a governance address.

Read more at docs.hyperdrive.box

What is Agent0?

Agent0 is DELV's Python-based library for testing, analyzing, and interacting with Hyperdrive's smart contracts. It provides ready-for-use trading policies as well as a framework for building smart agents that act according to policies that can be strictly user-designed, AI-powered, or a combination. These agents are deployable to execute trades on-chain or can be coupled with a simulated environment to test trading strategies, understand Hyperdrive, and explore integrations or deployment configurations.

When running Hyperdrive on a local blockchain, agent0 also provides a managed database delivered to you as Pandas dataframes via an API as well as a visualization dashboard to enable analysis and understanding.

Read more about how agent0 works in our docs.

Quickstart Examples

This repo contains general purpose code for interacting with Ethereum smart contracts. However, it was built for the primary use case of trading on Hyperdrive markets.

First, install Foundry and Docker.

Next, using a Python 3.10 environment, you can install agent0 via uv:

uv pip install --upgrade agent0

You're ready to go! You can use agent0 to do datascience, automate trading, or simulate trading strategies. Below are some quick examples to get you started. When you're ready, check out our examples folder for more information, including details on executing trades on remote chains.

Execute Hyperdrive trades in a simulated blockchain environment

import datetime
from fixedpointmath import FixedPoint
from agent0 import LocalHyperdrive, LocalChain

# Initialize.
chain = LocalChain()
hyperdrive = LocalHyperdrive(chain)
hyperdrive_agent0 = chain.init_agent(base=FixedPoint(100_000), eth=FixedPoint(10), pool=hyperdrive)

# Run trades.
chain.advance_time(datetime.timedelta(weeks=1))
open_long_event = hyperdrive_agent0.open_long(base=FixedPoint(100), eth=FixedPoint(10))
chain.advance_time(datetime.timedelta(weeks=5))
close_event = hyperdrive_agent0.close_long(
    maturity_time=open_long_event.maturity_time, bonds=open_long_event.bond_amount
)

# Analyze.
pool_info = hyperdrive.get_pool_info(coerce_float=True)
pool_info.plot(x="block_number", y="longs_outstanding", kind="line")

# Shut down the chain.
chain.cleanup()

Query all positions

import os
from agnet0 import Chain, Hyperdrive

# We recommend you use env variables for sensitive information.
PUBLIC_ADDRESS = os.getenv("PUBLIC_ADDRESS")
RPC_URI = os.getenv("RPC_URI")

# Address of the Hyperdrive registry (this is for Sepolia).
REGISTRY_ADDRESS = "0x4ba58147e50e57e71177cfedb1fac0303f216104"

# View open and closed positions across all pools.
with Chain(RPC_URI) as chain:
    agent = chain.init_agent(public_address=PUBLIC_ADDRESS)
    registered_pools = Hyperdrive.get_hyperdrive_pools_from_registry(
        chain,
        registry_address = REGISTRY_ADDRESS,
    )
    all_positions = agent.get_positions(
        pool_filter=registered_pools,
        show_closed_positions=True,
    )
    print(all_positions)

Automate withdrawing mature positions

import os
from agnet0 import Chain, Hyperdrive

# We recommend you use env variables for sensitive information.
PRIVATE_KEY = os.getenv("PRIVATE_KEY")
RPC_URI = os.getenv("RPC_URI")

# Address of the Hyperdrive registry (this is for Sepolia).
REGISTRY_ADDRESS = "0x4ba58147e50e57e71177cfedb1fac0303f216104"
chain = Chain(RPC_URI)
registered_pools = Hyperdrive.get_hyperdrive_pools_from_registry(
    chain,
    registry_address = REGISTRY_ADDRESS,
)

agent = chain.init_agent(private_key=PRIVATE_KEY)
for pool in registered_pools:
    # Close all mature longs.
    for long in agent.get_longs(pool=pool):
        if long.maturity_time <= chain.block_time():
            agent.close_long(
                maturity_time=long.maturity_time,
                bonds=long.balance,
                pool=pool,
            )
    # Close all mature shorts.
    for short in agent.get_shorts(pool=pool):
        if short.maturity_time <= chain.block_time():
            agent.close_short(
                maturity_time=short.maturity_time,
                bonds=short.balance,
                pool=pool,
            )

# Shut down the chain.
chain.cleanup()

Advanced Install

Please refer to INSTALL.md for more advanced install options.

Deployment

Please refer to BUILD.md.

Testing and Contributions

Please refer to CONTRIBUTING.md.

Number format

We frequently use 18-decimal fixed-point precision numbers for arithmetic.

Disclaimer

This project is a work-in-progress. The language used in this code and documentation is not intended to, and does not, have any particular financial, legal, or regulatory significance.


Copyright © 2024 DELV

Licensed under the Apache License, Version 2.0 (the "OSS License").

By accessing or using this code, you signify that you have read, understand and agree to be bound by and to comply with the OSS License and DELV's Terms of Service. If you do not agree to those terms, you are prohibited from accessing or using this code.

Unless required by applicable law or agreed to in writing, software distributed under the OSS License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the OSS License and the DELV Terms of Service for the specific language governing permissions and limitations.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

agent0-0.25.1.tar.gz (225.6 kB view details)

Uploaded Source

Built Distribution

agent0-0.25.1-py3-none-any.whl (296.4 kB view details)

Uploaded Python 3

File details

Details for the file agent0-0.25.1.tar.gz.

File metadata

  • Download URL: agent0-0.25.1.tar.gz
  • Upload date:
  • Size: 225.6 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/5.1.1 CPython/3.12.7

File hashes

Hashes for agent0-0.25.1.tar.gz
Algorithm Hash digest
SHA256 c5b94bac52f9b6b54ef54ae5112e4895aa9e265e82dcb8bbfa74bc18fc4bdf44
MD5 66ca5e8e1b3afc5a6711b79e58423f5a
BLAKE2b-256 3593e881d2f51cada6ea6a0f2727e90bffb204894240f3b1b8133ccdade7dcfc

See more details on using hashes here.

File details

Details for the file agent0-0.25.1-py3-none-any.whl.

File metadata

  • Download URL: agent0-0.25.1-py3-none-any.whl
  • Upload date:
  • Size: 296.4 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/5.1.1 CPython/3.12.7

File hashes

Hashes for agent0-0.25.1-py3-none-any.whl
Algorithm Hash digest
SHA256 67d8d7c31462d220594fd412767ebbb86f0953606aaf811838a43cc9f91a965b
MD5 f15f1498a1482a7107d3d9c6497970ed
BLAKE2b-256 030076554967607eca1bf99d771e6a0697814e5b02bc009a491412b963452c6b

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page