Algorithmic trading and market making. From the terminal.

Research, backtest, and execute across centralized and onchain markets from one local runtime.

$ curl -fsSL https://marketlab.sh/install.sh | bash

~/.market-lab/bots/bot_01J4MM82marketlab

mlab bot run grid BTC \--venue bulkf --margin 100 --leverage 10 \--duration 3600 --levels 5 --step-bps 6 \--stop-loss-pct 5

Classic paired grid runningBUY L1 filled

Jobbot_01J4MM82

ModeGRID

VenueBULK

Levels5 / SIDE

Bidspricesize

L165,891.40.00151

L265,851.90.00151

L365,812.30.00151

L465,772.70.00151

L565,733.10.00151

Askspricesize

L165,970.60.00151

L266,010.10.00151

L366,049.70.00151

L466,089.30.00151

L566,128.90.00151

Inventory+0.00151 BTC

Anchor$65,931

Fixed step6 BPS

Net PnL+$0.12

paired orderSELL L1 @ 65,931.0

Setup with your coding agent

Tell your agent to set up Market Lab.

Copy the prompt. Your coding agent reads the docs, installs Market Lab, and verifies the runtime.

01 Copy02 Paste03 Review

marketlab-setup.md

# Set Up Market Lab

Install or upgrade Market Lab on this machine and verify the setup.

Official documentation:

https://docs.marketlab.sh/

Read the Installation and Daemon pages before acting. Use the docs and the installed CLI's `--help` output as the source of truth.

1. Inspect the operating system, architecture, `PATH`, and any existing `mlab` installation.
2. Ask whether `mlabd` should use the **Native** or **Docker** backend. Do not choose or change the backend silently.
3. Follow the documented install or upgrade flow.
4. Verify:

```bash
mlab --version
mlab daemon backend
mlab daemon status
```

Ask before configuring market data, authentication, execution, SSH transport, or Python.

Never ask the user to paste a private key or API key into chat. Keep authentication inside Market Lab's hidden terminal prompts. Never place a live order during setup; use `--dry-run` when a check requires an execution plan.

Finish with the installed version, selected backend, daemon status, any optional setup completed, and one safe next command.

01 / OPERATE

Built to trade.

Run a native algorithm, deploy a market maker, or bring your own code.

Work an order

Execution algorithms

Volume-paced execution.

Volume-pacedMaker-firstParticipation cap

$ mlab strategy run vwap BTC \
  --venue bulkf --side buy --margin 100 --duration 3600 \
  --volume-sources binancef@mmt,okxf@mmt,bulkf --dry-run

Quote liquidity

Market makers

Fixed paired grid.

Fixed pricesPaired cellsStop loss

$ mlab bot run grid BTC \
  --venue bulkf --margin 100 --leverage 10 \
  --duration 3600 --levels 5 --step-bps 6 \
  --stop-loss-pct 5 --dry-run

03Research

Explore data in JupyterLab.

$ mlab notebook

04Scripting

Backtest your own strategy.

$ mlab script backtest strategy.py \
  --from 2026-08-01 --to 2026-08-20

05Job control

Run and supervise detached jobs.

$ mlab bot status <JOB_ID>
mlab bot logs <JOB_ID> --follow
mlab bot stop <JOB_ID>

02 / ARCHITECTURE

Data in. Orders out.

One local runtime connects market data, strategy logic, and execution.

Market data connected01 / 04

PROVIDERS / VENUESSelect a provider

MMT venues20+ venues

binanceCEX / Spot

Binance spot order books and trades.

SOURCE LAYERcandlestradesorderbookvdvolumes

MARKET LABTrading terminal

LOCAL

DataBacktestsStrategiesBotsScriptsExecution

OPERATOR SURFACETerminal / JSON / JSONL

EXECUTION LAYERmlabd / Native or Dockerlocal

EXECUTIONThree providers

BULK

Execution provider--venue bulkf

Testnet

Hyperliquid

Execution providerhyperliquid / hyperliquidf

Direct

HyperLink

Execution providerhyperlink / hyperlinkf

Mainnet

03 / MARKET CATALOG

Find the exact symbol.

Search the bundled catalog by provider, venue, or symbol.

Loading catalog

ProviderProvider symbolVenueMarketMarket Lab input

Loading local market catalog...

04 / BOT WORKFLOW

One job. Four clear states.

Dry-run first. Then deploy, observe, and stop.

marketlab01 / 04

$ mlab bot run grid BTC \
  --venue bulkf --margin 100 --leverage 10 \
  --duration 3600 --levels 5 --step-bps 6 \
  --stop-loss-pct 5 --dry-run

RESULT-levels per side 5-fixed step 6 bps-initial cells 10 equalokorders submitted 0

05 / EXTEND

Write the strategy in Python or JavaScript.

Use your Python environment, or run lightweight JavaScript in QuickJS.

Research in JupyterLab

Backtest and run plain Python

Embed lightweight JavaScript

Explore scripting

Python V2

01 from statistics import fmean

02 SOURCE = "btc@candles@hyperliquidf:timeframe=60"

03 script = {
04   "name": "mean-cross",
05   "version": "2",
06   "lookback": 30,
07 }

08 def on_data(ctx, history):
09   rows = history.source(SOURCE)
10   if len(rows) < 20: return
11   fast = fmean(row["c"] for row in rows[-5:])
12   slow = fmean(row["c"] for row in rows[-20:])

13   if fast > slow:
14     ctx.trade({"exchange": "hyperliquidf", "symbol": "BTC", ...})

Market Lab on GitHub

View repository