Strategy pattern · Market making
Build a market making bot, simulator first and honest about the odds.
In Super Algo Kit's card, a market making bot posts buy and sell quotes around a fair price. It earns the spread from traders who want immediacy. The kit grades the models B. Evidence that a retail maker covers its costs is rare, so that part is C. It belongs to the low-latency build set: simulator first, then shadow mode.
strategy facts
Market making: quoting both sides with inventory control
- Family
- Market making: quoting both sides of the order book
- Markets
- Crypto spot and derivatives, equities, futures and prediction markets
- Timeframe
- Tick by tick
- Evidence grade
- B for the mechanism and models; C for retail results, which are rarely supported
- Status
- Literature: described in published work, untested by the kit
- Build set
- Low-latency engines only: simulator, latency budget and replay, then shadow mode
- Data needed
- Order-book depth and trades with exchange timestamps, measured latency, and fee and rebate tiers
From the kit's own card. Not investment advice.
Paper first
Test it before you trade it.
Super Algo Kit runs inside the coding agent you already use. It interviews you, writes the spec with pass marks set first, then builds, backtests and paper-tests. Real money starts only when you type the authorize command.
The plan, step by step
Tick them off as you go.
How does a market making strategy work?
A market maker is paid the bid-ask spread for supplying liquidity, and two costs push against it. Inventory risk builds when holdings pile up on one side while prices move. Adverse selection happens when informed traders trade with you just before the price moves against you.
Classic models quote around a reservation price that shifts against your inventory. Spreads widen with volatility and with how toxic the incoming order flow is.
Market making rules in plain words
This is a summary of the card's outline, not its exact rules. Every value is declared before a test.
- Take the fair price from the mid, or a microprice adjusted for book imbalance.
- Shift both quotes against your inventory: a long position lowers both.
- Widen spreads with volatility, toxic flow and weak queue priority, never inside one tick.
- Cap inventory, quote size and message rate; halt on stale data; cancel on disconnect.
- Count maker and taker fees and rebates at the tier you really reach.
What the evidence says, and where it breaks
The theory is well studied, from dealer pricing models in 1981 to limit-order-book models in 2008. Those papers explain why spreads exist and how inventory should shift quotes. They do not show that a retail maker can cover its costs, so the kit grades that part C.
Professional firms are faster, so a retail maker tends to sit at the back of the queue. The fills it does get are mostly the toxic ones. Backtests without queue position and latency show fills that would never happen.
How Super Algo Kit tests a market making bot
Market making runs only in the low-latency build set. The kit starts with a simulator, a conservative queue model, a latency budget and deterministic replay. Then the bot runs in shadow mode on live data, computing quotes without sending them.
Pass marks are declared first, fees and rebates are modelled, and every trial counts toward the deflated Sharpe ratio. An independent risk gate enforces inventory, quote-size and message-rate caps, with cancel on disconnect where the venue supports it. For HFT-style systems, the kit's paper minimum is 20 sessions and 1,000 orders.
Before you rely on it
Where it fails.
- Adverse selection: fills cluster just before prices move against your quotes.
- Inventory builds in trends, leaving you holding a losing position.
- Faster firms take the front of the queue, so a retail maker receives mostly toxic fills.
- Rebates assumed at a fee tier the account never reaches.
- An outage or disconnect can leave quotes live in a moving market.
- Backtests without queue position and latency show fills that would never happen.
FAQ
Market making: questions.
What is a market making bot?
A market making bot posts both a buy quote and a sell quote and earns the spread when other traders take them. Super Algo Kit's card adds the two risks that matter: inventory that builds on one side and losses to better-informed traders. The kit treats it as a low-latency build.
Can a retail trader compete at market making?
It is hard. Professional firms are faster, so a retail market maker sits at the back of the queue and receives mostly toxic fills. Super Algo Kit grades the models B but retail results C, because supporting evidence is rare.
What is adverse selection in market making?
Adverse selection in market making is when informed traders trade with your quotes just before the price moves against you. Glosten and Milgrom modelled it in 1985, and later work studied flow toxicity in high-frequency markets. Super Algo Kit widens spreads with toxicity and tests fills with a conservative queue model.
Can I run a market making bot on Polymarket or Kalshi?
Prediction markets are among the card's markets, but each venue's terms on bots and market making were not confirmed in the kit. Eligibility depends on your country, and Super Algo Kit never helps get around a venue's restrictions. Resolution rules and settlement timing become part of the data.
Does Super Algo Kit run the market maker for me?
No. Super Algo Kit is development software inside your own coding agent, and it writes and tests code in your project. Agents never hold live keys or place live orders. Paper first, live only when you type the authorize command.
Keep reading
More guides.
Crypto · Exchange APIBinanceSuper Algo Kit helps you build a Binance trading bot inside your own AI coding agent.
Crypto · Market data feedTardis.devSuper Algo Kit can use Tardis.dev data for backtesting crypto bots that your own AI coding agent builds.Claude Code · Coding agentClaude CodeSuper Algo Kit turns Claude Code into an algo trading team, so you can build your own Claude Code trading bot.Sources the kit's card cites
- Ho and Stoll (1981). Optimal dealer pricing under transactions and return uncertainty. Journal of Financial Economics.
- Glosten and Milgrom (1985). Bid, ask and transaction prices in a specialist market with heterogeneously informed traders. Journal of Financial Economics.
- Avellaneda and Stoikov (2008). High-frequency trading in a limit order book. Quantitative Finance.
- Easley, Lopez de Prado and O'Hara (2012). Flow Toxicity and Liquidity in a High-frequency World. Review of Financial Studies.
Build it with Super Algo Kit.
15 specialist roles and 35 skills inside your coding agent. Paper first, live only when you say.
Super Algo Kit is development software. It is not investment advice, a signal service or a promise of performance. Trading involves substantial risk, including the loss of all capital you trade with.