Market Making Basics: How Liquidity Providers Earn (and Why It's Hard)

Market makers (MMs) provide the liquidity that lets takers trade. They post bids and asks, earn the spread between them, and capture maker rebates from exchanges. Done at scale by sophisticated firms, market making is a real business with consistent returns. Done casually by retail, it's usually a way to give up money to faster participants. Understanding how it works clarifies what the institutional liquidity layer is actually doing, and why retail attempts at MM rarely succeed.

What market makers actually do

The mechanic:

A market maker posts a bid (e.g., $66,000) and an ask (e.g., $66,002) on an order book. If a taker buys, they fill against the MM's ask. If a taker sells, they fill against the MM's bid. The MM captures the $2 spread.

If the MM does this repeatedly with balanced flow (equal buy and sell volume from takers), they accumulate profit at the spread rate. They also earn the maker rebate on each fill (often 1-2 bps).

The challenge: flow isn't always balanced. Sometimes takers buy more than they sell, leaving the MM short the asset they kept selling. Sometimes the opposite. The MM has to manage this inventory risk, the difference between their flow direction and the underlying price direction.

The two profit sources

Spread capture. Every taker that fills against you pays the spread. With high volume, even small spreads add up.

Maker rebate. Many exchanges pay makers a small rebate (sometimes negative fees, e.g., -0.005%) for providing liquidity. Adds to the per-trade profit.

Combined, a busy MM can capture 1-3 bps per round trip across hundreds or thousands of fills per day.

The hard part: inventory risk

If you post a bid and your bid gets hit, you bought the asset. Now you're long. If the price keeps falling, your inventory loses value. The spread you captured is trivial compared to the inventory loss.

Symmetrically, if your ask gets hit, you're now short. Rising prices hurt your inventory.

Professional MMs manage inventory risk through:

1. Symmetric quoting. Adjust bid and ask to encourage flow that rebalances inventory. If you're long, raise your bid (less attractive to sellers, discouraging more buying) and aggressively quote your ask (encouraging takers to buy from you).

2. Hedging. Take an offsetting position in a correlated instrument (perpetual hedge for spot position). The hedge neutralizes price exposure while the MM continues to capture spread.

3. Inventory limits. Hard caps on how long or short the MM is willing to be at any given moment. When limits hit, the MM stops adding inventory regardless of flow.

4. Fast cancellation. If the market moves unexpectedly, immediately cancel quotes that are now out-of-line. Speed matters, slow MMs get adverse- selected (filled by people who saw the move first).

These management techniques require infrastructure (low-latency execution, real-time risk monitoring, fast cancel/replace). Casual retail attempts without this infrastructure get adversely selected and lose to informed flow.

Adverse selection, the killer

Adverse selection is the phenomenon where the takers who hit your quotes know something you don't. They're hitting your bid because they have a reason to sell at that price (maybe they saw a sell signal you didn't); they're hitting your ask because they have a reason to buy.

If you post passive quotes and don't move them as information arrives, the takers who fill you are disproportionately the informed ones. You consistently buy right before declines and sell right before rallies. The spread you captured is tiny compared to the adverse-selection cost.

This is why naive MM doesn't work. The math seems good ("I'm capturing the spread!") but the realized performance is bad because the flow you fill is biased against you.

Pros defend through speed (cancel before the informed take) and through narrowing quotes when they sense informed flow (wider spread = compensation for adverse selection).

Three ways retail attempts MM (and why they usually fail)

1. Manual order placement on CEXs. A retail trader posts limit bids and asks around the mid-price. Hopes to capture the spread. Usually:

• They're too slow to cancel when the market moves
• They get adversely selected by informed flow
• The spreads they can post are wider than pro MMs' (because they can't cancel fast enough), so they don't get filled often
• Net result: small wins offset by occasional big inventory losses

2. Providing liquidity in AMM pools. Deposit equal value of two assets into a Uniswap pool. Earn a share of swap fees. The mechanism is automated no need for fast cancel/replace.

But: impermanent loss. When the price ratio diverges, the LP's portfolio underperforms simply holding the two assets. The fees might or might not cover the IL.

For volatile pairs (most crypto pairs), IL often exceeds fee income. Net negative for many LPs.

For stable-pair pools (USDC/USDT, etc.), IL is minimal but fees are also low. Returns of 1-5% annually, not exciting.

3. Concentrated liquidity (Uniswap V3 style). LP positions a price range to concentrate capital where they think the price will be. Higher fee capture per dollar deployed.

But: requires active management as price moves. The position needs to be repositioned as the market shifts. Without active management, the LP gets stuck out-of- range and earns nothing while still bearing IL.

In practice, for most retail, AMM LPing produces returns below the simple "hold the assets" strategy once IL is accounted for. The advertised APRs are gross; the net returns are often unimpressive or negative.

The market-making business, what it looks like at scale

Professional MM firms (Cumberland, GSR, Wintermute, Jump, Jane Street's crypto desk, etc.) operate differently:

• Run thousands of trades per second across many venues
• Sub-millisecond decision latency
• Co-located infrastructure near exchanges
• Sophisticated inventory and risk models
• Integrated hedging across spot, perp, options
• Massive capital deployment ($billions)
• Direct relationships with exchanges (better fees, faster API access)

The combination produces consistent returns, typically double-digit annual ROI on capital deployed, with relatively low drawdowns. Boring, profitable, requires infrastructure.

The retail MM equivalent is essentially an attempt to play the same game without any of the infrastructure advantages. The result is predictable.

When passive LPing makes sense for retail

Despite the cautions, some scenarios where AMM LPing is reasonable:

1. Stable-stable pairs. USDC/USDT and similar. Minimal IL. Fee income is the main return. Boring 1-5% annual yield. Comparable to high-yield savings; more complex but transparent.

2. Pegged-asset pools. ETH/wstETH (regular ETH against staked ETH). Pegged relationship, low IL, captures swap fees from people moving between the two forms.

3. You actively manage and accept the risks. If you understand IL precisely and are willing to do the work of repositioning concentrated liquidity, the returns can be better, but it's an active strategy, not passive yield.

4. You'd hold the assets anyway. If you'd hold both assets in your portfolio anyway, LPing them captures fees on top. The IL is the opportunity cost vs the optimal mix; if you'd hold the mix anyway, the IL is partially mitigated.

For most other scenarios, volatile pairs, casual LPing without active management, the returns disappoint.

A common mistake: chasing high-APR pools

DeFi platforms often advertise pools with massive APRs (50%+, sometimes 1000%+). The high APRs are usually:

• Token emissions (the protocol's own token paid to LPs; often crashes in price as everyone dumps)
• Unsustainable bootstrap incentives (will end)
• Compensation for extreme IL risk (you pay the IL with your other assets)

Net realized returns are usually much lower than headline APRs.

The fix: model net returns including IL, token-price risk, and time-decay of incentives. The realistic return is usually a small fraction of the advertised APR.

A common mistake: ignoring smart-contract risk

LPing requires depositing assets into a smart contract. That contract can have bugs, exploits, or outright rug-pulls. Many LPs over the years have lost significant capital to contract failures.

The fix: stick to audited, established protocols (Uniswap, Curve, Aave). New protocols offering high returns also have higher contract risk. The "extra yield" might be the risk premium for trusting unproven contracts, and the risk sometimes materializes.

Mental model, market making as running a small bank

A bank takes deposits from one party and lends them to another, capturing the spread between deposit rate and loan rate. Done at scale with good risk management, banks are profitable.

A retail trader who tries to "be a bank" without the risk management capabilities loses to better-managed banks. Same risks (default, duration mismatch, liquidity); fewer tools to manage them.

Market making is similar. Pros do it as a business with risk-management infrastructure. Retail doing it casually gets out-competed and adversely selected. Some forms (stable-stable LPing) are accessible; most forms aren't.

Why this matters for trading

Understanding market making clarifies the structure of the markets you trade in. The MM is the counterparty to most of your fills. Their economics are why spreads exist, why fees structure the way they do, and why liquidity behaves the way it does. For most retail, the right relationship to MMing is "understand it, use the parts that fit (basic LPing on stable pairs), don't try to compete with pros." Hex37's order-book mechanics mirror real exchange MMing dynamics, practicing limit-order patience builds the same intuition pros deploy at scale.

Takeaway

Market makers post bids and asks, earn the spread plus maker rebates. The challenge is inventory risk and adverse selection. Pros manage these with low-latency infrastructure, sophisticated risk models, and hedging. Retail attempts (manual CEX MMing, casual AMM LPing) usually underperform because they can't match the pro infrastructure. Stable-stable LPing and pegged-asset pools are the most retail-accessible forms. High-APR pools are usually misleading on net returns. Don't try to compete with pro MMs on liquid volatile pairs; you'll lose.