Run The Reference Market Maker#

This page is the canonical on-ramp for the reference market maker. The kit in packages/bot-rust is an educational starter, not a production trading system. It is dry-run by default and should stay dry-run until you have reviewed the config, risk limits, wallet funding, and market behavior.

What It Does#

The reference configuration:

Reads Solana market and orderbook state.
Reads Pyth prices through a Hermes endpoint.
Builds simple two-sided quotes.
Skews quotes based on inventory.
Applies spread, inventory, max-loss, and open-order limits.
Prints intended actions in dry-run mode.

It is not proprietary alpha and it is not a production market-making service.

Start In Dry Run#

The sample configuration does not start infrastructure. Before running it, provide both local dependencies configured in packages/bot-rust/configs/mm.toml:

a Solana RPC with the Seesaw program and discoverable test markets at http://127.0.0.1:8899; and
a Hermes-compatible service with the configured feed data at http://127.0.0.1:8787/hermes.

Then, from the repo root:

bash

cargo test --manifest-path packages/bot-rust/Cargo.toml
cargo run --manifest-path packages/bot-rust/Cargo.toml -- packages/bot-rust/configs/mm.toml

The sample mm.toml uses safe local defaults:

dry_run = true
local Solana RPC
local Hermes URL
virtual dry-run balance
no live transaction sending
conservative quote size and inventory caps

The package README documents the framework internals and other starter strategies. Keep setup and safety guidance on this page so integrators have one market-maker entry point.

Key Config Fields#

Field	Purpose
`bot.dry_run`	Keeps the bot from sending transactions while testing.
`bot.dry_run_virtual_balance`	Simulated settlement balance when no live account is funded.
`bot.treasury_token_accounts`	Live-only ordered list of exactly eight distinct treasury SPL token accounts.
`feeds.hermes_url`	Pyth Hermes endpoint.
`feeds.hermes_api_key` or `PYTH_API_KEY`	Required for upgraded Hermes endpoints.
`strategy.market_making.base_spread_bps`	Neutral spread around the mid.
`strategy.market_making.quote_size`	Size per quote in settlement base units.
`strategy.market_making.max_inventory`	Inventory cap for the strategy.
`strategy.market_making.exit_cutoff_secs`	Time before market end when quotes should stop.
`risk.max_loss`	Max cumulative buy exposure.
`risk.max_open_orders`	Maximum simultaneous open orders.
`risk.min_settlement_balance`	Settlement reserve to keep untouched.

Before Live Trading#

Do all of this before setting dry_run = false:

Use a dedicated wallet.
Fund only the amount you are prepared to risk.
Confirm the target RPC, program ID, settlement mint, market creators, and Pyth feed IDs.
Provide PYTH_API_KEY through the environment or a protected config file.
Configure exactly eight distinct bot.treasury_token_accounts, in the same index order as the on-chain config.treasury_recipients[i] array. A single treasury account, duplicate entries, or a reordered list is invalid.
Confirm startup uses LiveExecutor::new_verified. It fetches the Config PDA and fails closed before trading if the program owner, settlement mint, or any indexed treasury recipient differs from the configured value.
Set strategy.market_making.kill_switch = false only when ready.
Run against a local or staging market first.
Watch order placement, cancel behavior, and settlement claims end to end.
Keep final-minute exposure limits conservative.

Risk Notes#

Makers pay no fee on resting fills, but exiting inventory as a taker can cost fees and slippage.
Thin books can move sharply near resolution.
A dry-run strategy that looks stable can behave differently when live transactions compete for block space.
Always keep enough SOL for transaction fees and enough settlement balance for the configured quote sizes.