Web3 Backtest

Web3 Backtest is a generic Backtesting library for DeFi trading strategies. The library simplifies pullling historic data, simulating protocols, backtest metrics, charting and analytics.

Historic Data

Web3 Backtester pull desired historic data using arkiver. All data is formatted into snaphots with resolutions of 1m, 1h, or 1day. transaction- or block-resolution may be added in the future.

Datasources are specified with a simple configuration

const sources: DataSourceInfo[] = [
    {
        chain: 'arbitrum',
        protocol: 'camelot-dex',
        resoution: '1m',
        config: {
            pairs: [USDCWETH]
        }
    },
]

Supported sources are detailed [below](#Data Sources Supported)

Position Simulation

The library contains helper classes that simulate openning, changing and closing defi positions. For example, the Univ2PositionManager will simulation a position with:

const univ2 = new UniV2PositionManager()
const position = univ2.addLiquidity('WETH/USDC', 1, 1900);

and as cycles pass in the backtest, the position can be monitored and altered

if (position.valueUsd < 2500) {
    { token0, token1 } = position.close()
}

Simulators are in the early stages, the API will change. See Supported Positions Simulators

Data storage and Analytics

There is a docker-compose.yml file that will spin up an influx db and a local instance of grafana so you can store any data you like during the backtest and have plenty of charting flexibility with grafana.

Simplifying Code Structure

Web3 Backtester has a simple API that grants flexibility. The steps are:

Specify Data Sources
Specify Testing Period
Register Handlers
Run the backtest

Here's an example that creates a dummy backtest for a 10 day period, using Camelot 1m resolution data.

const main = async () => {
    const USDCWETH = '0x794a61358D6845594F94dc1DB02A252b5b4814aD'

    const bt = await Backtest.create(
        new Date('2023-01-01'), 
        new Date('2023-01-02'), 
        [{
            chain: 'arbitrum',
            protocol: 'camelot-dex',
            resoution: '1m',
            config: {
                pairs: [USDCWETH]
            }
        }]
    )

    bt.onData(async (update: DataSnapshot<Univ2PoolSnapshot>) => {
        console.log(`we have data for ${update.timestamp}!`)
    })

    bt.onAfter(async () => {
        console.log('backtest complete!')
    })

    bt.run()
}

main()

Features

Feature	Status
Generic Data Sources	✓
Generic Backtest API	✓
Examples: Hedged Farming Strategy	✓
Multi-resolution Support (1m, 1d, 1h)	✓
Influx boilerplate for charting	✓
NPM Module	✗
Example: AAVE Folding	✗
Wallet support for simulators	✗
Local Data Caching	✗

Position Simulators

Protocol	Status
Univ3	✓
Univ2	✓
AAVE	✓
Masterchef V2	✓
Perp V2	✗
Joes V2	✗
GMX	✗

Data Sources Supported

AMMs

Chain	Protocol	Minutely	Hourly	Hourly
Arbitrum	Camelot	✓	✗	✗
Arbitrum	Sushiswap	✗	✗	✗
Arbitrum	Univ3	✗	✗	✗

Lending Markets

Chain	Protocol	Minutely	Hourly	Hourly
Arbitrum	AAVE	✗	partial	✗
Avalanche	AAVE	✗	partial	✗

Farms

Chain	Protocol	Minutely	Hourly	Hourly
Arbitrum	Camelot	✓	✗	✗
Arbitrum	Sushiswap	✗	✗	✗
Arbitrum	Univ3	✗	✗	✗

Setup

Influx and Grafana

Grafana and influx are not required to test strategies but they are useful tools for visualising the results.

Run Grafana and influx locally with docker

docker-compose up -d

You can stop the grafana and influx containers with

docker-compose down -v

Run a Backtest

Install deps

npm install

Run the sample backtest

ts-node ./src/index.ts

or run the larger example (required influx). First copy the .env template, then run

cp .env.template .env
ts-node ./src/examples/cpmm_camelot_aave/index.ts

Graphana Vis

Your grafana instance is a fresh instance so there will be no dashboards, you'll need to create them. First step is setting up the influx data source with the following details

Property	Value
name	backtester
Query Language	InfluxQL
URL	http://localhost:8086
database	backtest
user	admin
password	admin

See the /grafana folder, it contains dashboard json files for all examples.

Robo Labs

If your looking for someone to use this library and write custom back tests for you, hit up Robo Labs on Twitter @robolabs_biz.

SmoothBot/web3-backtest