The Anatomy of a Market Maker: How We Think About Spreads

You see a token trading at $1.00. The spread is tight, the book is deep. Behind that calm surface, a market maker is making hundreds of micro-decisions per second. Where to place bids. Where to place asks. How wide. How deep. When to pull quotes entirely. Every one of those decisions is a calculated bet — and getting even one consistently wrong means bleeding capital.

In Article 1 of this series, we covered what market makers do and why they matter. Now we go deeper: into the mechanics of how market makers actually think about pricing. This is where the real complexity lives.

---

The Spread Is Everything

The bid-ask spread is the market maker's primary source of revenue and simultaneously their primary expression of risk. Every spread quote is an answer to a single question: what premium do I need to charge for the risk of being on the wrong side of this trade?

Here's a concrete example. Suppose you're making a market on TOKEN/USDT. The fair value is $1.00. You place a bid at $0.998 and an ask at $1.002 — a spread of $0.004, or 0.4%.

If a buyer lifts your ask and a seller hits your bid in roughly equal quantities, you pocket $0.004 per token. On 10,000 tokens traded per day, that's $40 in gross spread revenue. Across 5 pairs and 3 exchanges, that's $600/day or roughly $18,000/month before costs.

Simple enough. But here's the catch: tighter is not always better.

A 0.1% spread earns you $0.001 per token — one-quarter of the revenue — while exposing you to exactly the same risks. If the price moves against you by even 0.2% before you can adjust, you've lost more than you earned on the last 200 round-trip trades.

The optimal spread is a function of several variables:

• Volatility of the underlying asset
• Volume and trade frequency
• Inventory position and exposure limits
• Adverse selection risk from informed traders
• Competition from other market makers

Professional market makers don't set a static spread and walk away. They continuously recalculate, widening when risk increases and tightening when conditions are favorable. The spread is a living number that reflects everything the market maker knows about the current market state.

---

Adverse Selection: The Hidden Tax

Adverse selection is the single most dangerous risk in market making, and the one that amateurs consistently underestimate. It occurs when the counterparty to your trade has better information than you do.

Here's how it works in practice. You're quoting TOKEN/USDT at $0.998 bid / $1.002 ask. A whale who knows that a major exchange listing is about to be announced starts accumulating — buying every ask in sight. Your $1.002 ask gets lifted. Then your $1.004 ask. Then $1.006. By the time you detect the pattern, you've sold 50,000 tokens at an average of $1.004, and the token is now trading at $1.15 after the listing announcement.

You just lost $0.146 per token on 50,000 tokens — $7,300 in a single event. That wipes out weeks of spread earnings.

Adverse selection in crypto is particularly brutal because:

Information asymmetry is extreme. Insider knowledge about listings, partnerships, tokenomics changes, and protocol upgrades circulates through private Telegram groups long before public announcements. The people trading on this information are picking off market maker quotes that haven't adjusted yet.

Whale accumulation is detectable but difficult to act on. A sophisticated buyer will split orders across multiple exchanges, use multiple wallets, and time purchases to avoid detection. By the time the pattern is visible in your order flow data, you've already absorbed significant adverse flow.

Dumps are even worse. When an early investor decides to exit a large position, they sell into your bids relentlessly. You end up holding a growing inventory of a token that's declining — the exact opposite of what you want.

How do professional market makers defend against adverse selection?

Flow analysis. Track the ratio of buy-to-sell volume hitting your quotes. If one side is consistently dominant, something is happening. A sustained buy/sell imbalance above 60/40 warrants immediate spread widening.

Quote speed. The faster you can update quotes after a trade, the less time stale quotes sit in the book waiting to be picked off. Latency is the enemy.

Spread tiering. Widen spreads on larger order sizes. A 50-token trade might get a 0.3% spread; a 5,000-token trade gets 0.8%. This forces informed traders to pay more for their information advantage.

Cross-exchange signals. If the price moves on Binance, your quotes on MEXC and Gate.io need to update within milliseconds. Stale quotes on secondary exchanges are the easiest target for arbitrageurs.

---

Inventory Risk: The Balancing Act

Every trade a market maker executes creates an inventory position. Buy 1,000 tokens, and you're now long 1,000 tokens. If the price drops 5%, you've lost $50 per 1,000 tokens of exposure — regardless of what you earned on the spread.

Inventory accumulation is the slow killer of market makers. It happens gradually: a slight imbalance in buy vs. sell flow over hours or days leads to a steadily growing position in one direction. By the time you notice, you might be holding $200,000 in a token that's trending down.

Professional market makers manage inventory risk through several mechanisms.

Quote skewing. If you're long 10,000 tokens and want to reduce, you shift your quotes: tighten the ask (to encourage sales) and widen the bid (to discourage further buys). Instead of $0.998 / $1.002, you might quote $0.996 / $1.001. This creates a natural flow that pushes your inventory back toward neutral.

The magnitude of the skew depends on how far from neutral you are. A small position gets a small skew. A position approaching your risk limit gets an aggressive skew — you're effectively telling the market "I really want to sell and I'd rather not buy."

Inventory limits. Hard caps on how much net exposure you're willing to hold. When limits are hit, the market maker stops quoting on the side that would increase exposure. Better to miss trades than to take on unmanageable risk.

Portfolio-level hedging. For tokens that have correlated assets or derivatives, market makers can hedge exposure. Long TOKEN on MEXC? Short a correlated asset on another exchange. This reduces net exposure while allowing continued market making.

Time-based decay. Many strategies include a time penalty for inventory age. A position that's 5 minutes old might be fine. The same position held for 2 hours starts incurring an internal cost that forces the algorithm to trade more aggressively to flatten.

The goal is never to hold zero inventory — that's impossible when you're actively trading. The goal is to keep inventory oscillating around zero, spending as little time as possible at the extremes.

---

Volatility: Friend and Foe

Volatility has a dual relationship with market makers. Higher volatility means wider spreads are justified — which means more revenue per trade. But it also means faster price moves, more adverse selection, and larger potential losses on inventory.

How volatility drives spread width. A simple but effective model: set the minimum spread to a multiple of recent realized volatility. If the token's 1-hour realized volatility is 0.5%, you might set your spread at 1.5x that — 0.75%. If volatility spikes to 3%, your spread widens to 4.5%.

This isn't arbitrary. The spread needs to be wide enough to compensate for the expected price move during the time your quote is exposed to the market. If prices are moving 2% in the time it takes you to update quotes, a 0.5% spread guarantees losses.

Flash crashes. Crypto markets experience sudden, violent moves that can drop prices 15-30% in seconds. During these events, market makers face an impossible choice: keep quoting and risk massive losses on stale bids, or pull quotes and let the market freefall.

Most professional market makers have circuit breakers — automatic systems that widen spreads dramatically or pull quotes entirely when price moves exceed predefined thresholds. A 5% move in 30 seconds might trigger a spread widening to 3%. A 10% move might trigger a full quote pullback.

Sudden pumps. Equally dangerous. A coordinated pump pushes the price up quickly, and any market maker with asks in the book sells tokens that are immediately worth more. This is adverse selection at scale — the pump organizers know the price is going up, and your asks are free money for them.

Volatility regime detection. Sophisticated market makers don't just react to volatility — they try to predict regime changes. A token that's been quiet for three days and suddenly sees a spike in order flow is likely entering a new volatility regime. The algorithm should preemptively widen spreads before the move happens, not after.

The best market makers embrace volatility rather than fear it. Wide spreads during volatile periods are enormously profitable — if you survive the inventory swings. The key is having enough capital reserves and risk management discipline to stay in the game during the wild moves.

---

How OpenMM Handles Spread Management

Everything described above — spread optimization, adverse selection defense, inventory management, volatility adjustment — is what makes professional market making a complex engineering challenge. OpenMM is built to handle this complexity through configurable, automated strategies.

Grid strategies with configurable spacing. OpenMM's grid trading engine places layered orders at defined intervals around the current price. You choose between two spacing models:

• Linear spacing places orders at equal price intervals (e.g., every 0.5% from the midpoint). This provides uniform coverage and is suitable for stable, range-bound markets.
• Geometric spacing increases the distance between orders as they move further from the midpoint (e.g., 0.5%, 0.65%, 0.85%, 1.1%). This concentrates liquidity near the current price — where most trading happens — while still providing depth further out. The spacing factor is configurable, letting you tune how aggressively levels spread apart.

Pyramidal sizing to concentrate liquidity. Instead of placing equal-sized orders at every level, pyramidal sizing places larger orders near the midpoint and smaller orders at the extremes. This is how institutional market makers allocate capital: heavy where trades are likely, light where they're speculative.

Combined with geometric spacing, pyramidal sizing creates a liquidity profile that mimics what you'd see on a Tier 1 exchange — deep in the center, tapering at the edges.

Real-time adjustment via the MCP interface. The OpenMM MCP server exposes 13 trading tools through the Model Context Protocol, letting you monitor and adjust strategies conversationally. Check current grid status, modify spacing parameters, adjust order sizes, or stop strategies entirely — all through natural language commands to an AI agent.

This means you can react to changing market conditions without writing code or editing config files. If volatility spikes, tell the agent to widen spacing. If a token is trending, adjust the skew. The feedback loop between market observation and strategy adjustment drops from minutes to seconds.

For a detailed comparison of how OpenMM stacks up against other tools, see our OpenMM vs Hummingbot comparison. For a breakdown of what market making typically costs and how self-service tools change the economics, read the Crypto Market Making Pricing Guide.

---

Market Making Decoded: Full Series

• Article 1: What Is Market Making? The Engine Behind Every Trade
• Article 2: The Anatomy of a Market Maker: How We Think About Spreads (this article)
• Article 3: Multi-Exchange Market Making: Running on CEXs and DEXs Simultaneously
• Article 4: What Token Projects Should Know Before Engaging a Market Maker

---

QBT Labs builds professional trading infrastructure and provides market making services across multiple CEXs and DEXs. Based in Cyprus, serving Web3 teams globally.

Learn more: qbtlabs.io | [email protected]