Market makers without myths: algorithms, volume, and spread control

When the market moves sharply, price pushes through levels where two-sided quotes are expected. At that moment, the spread widens, order book depth shrinks, and execution quality deteriorates.

Market-making algorithms post bid and ask to earn the spread. Risk is bounded by quote size, inventory, and execution conditions. Visible order book volume is not a target variable. The spread changes as the price of risk across different market regimes.

⚙️ Who market makers are in real market mechanics

A market maker is an algorithmic participant that maintains two-sided limit quotes as long as execution risk is compensated by the spread and does not create an uncontrollable inventory imbalance.

🧩 Myths about market makers and the nature of liquidity

Distortions arise from projecting directional trading logic onto algorithms. Algorithms manage risk through the spread, quote size, and how quickly limits are pulled when matching conditions deteriorate.

💧 What liquidity is from the perspective of execution

Liquidity is the probability that an order will execute at a price close to what is expected, without sequentially crossing multiple book levels.

📐 Why a market maker doesn’t need a price forecast

A price forecast does not make the market-making model more robust. Shifting quotes based on expectations breaks neutral quoting. Under one-sided order flow, inventory accumulates faster.

1. Quotes shift due to an expected move.
2. One side of the book is filled more often.
3. The position grows via a series of fills.
4. Inventory risk grows faster than spread compensation.

🤖 Market-making algorithms: how liquidity is maintained

Market-making algorithms maintain liquidity through continuous bid/ask quoting. Execution parameters are set by the spread, size, and the speed of pulling orders.

The algorithm operates as a loop: read the order book → place quotes → a limit fills → recalculate the position → change spread and size.

↔️ Spread control as the primary protection mechanism

The spread compensates for unfavorable execution risk. The spread changes faster than the mid-price.

The spread is the difference between the best buy and sell price. Algorithms set it as the minimum income that covers the expected price drift and inventory risk.

Each fill of a limit order increases PnL sensitivity to mid-price movement until the next quote update.

Spread width sets fill frequency and the price at which the next size is accepted.

The spread widens as volatility rises, order flow becomes more asymmetric, and quote-update delays increase.

In impulse phases, the spread widens together with lower size at the best prices. The reason is that the risk of one fill rises faster than turnover income.

📉 Why the spread widens without news and volume

The spread can widen without news when unfavorable execution risk rises faster than expected compensation from turnover. This can happen even with a low number of trades.

Spread width depends on the probability of adverse selection. After a limit order fills, price moves against the market maker.

Adverse selection probability rises when the front line of the book thins out. Opposing limits are pulled faster than new quotes appear.

With low traded volume, a single market order crosses multiple levels if the book loses continuity.

1. Limit orders are pulled from best bid and best ask as order-flow structure deteriorates.
2. The front line loses density without a rise in trade count.
3. One market order moves price several ticks at once.
4. Algorithms widen the spread to reduce fill frequency and the risk of a series of executions.

A thinner front line worsens inventory rebalancing. Opposing fills appear less often. The position is held longer.

Compensation is achieved by widening the spread and shifting quotes away from the mid-price. Limit size is not increased.

Widening spreads without news points to fragile liquidity. The market absorbs small aggressive size worse.

📊 Market makers and order book structure

Order book structure shows the current distribution of market makers’ limit orders. It is not guaranteed executable size.

Market makers’ limit orders are temporary quotes. Parameters change as volatility rises, order flow becomes more asymmetric, and inventory risk increases.

Visible depth shows size by levels. It does not lock in that orders will remain until a market order arrives.

The practical question reduces to one thing: will the quotes remain active at the moment the order is matched?

When execution conditions deteriorate, size at best bid and best ask decreases. This limits the magnitude of one unfavorable fill.

After that, quotes move farther from the mid-price. Aggressive orders require more price distance. Matching frequency declines.

These changes occur without news and without higher traded volume, because canceling and moving limits does not require trades.

⚠️ When market makers stop being a source of liquidity

A market maker stops being a source of liquidity when a series of fills accumulates inventory risk faster than the spread compensates for adverse selection and price drift between quote updates.

From the outside, this looks like disappearing depth or a thinner front line. Mechanically, it is the cessation of quote support as the risk/compensation ratio deteriorates.

The trigger is directional market-order flow. Fills arrive from one side. A turnover model turns into position accumulation.

Each fill shifts the position. PnL sensitivity to mid-price movement rises until the next quote recalculation. Rebalancing cost increases. Execution robustness declines.

1. Market orders dominate on one side of the book.
2. Fills concentrate without regular opposing fills.
3. Inventory imbalance grows faster than spread compensation.
4. The algorithm reduces size, widens the spread, and if pressure persists, pulls quotes.

After leaving the front line, the market moves into thin books. Nearby levels are sparse. A small aggressive order crosses several levels in a row.

Price movement can look like it happens “without volume.” The reason is the absence of opposing limits at the moment the next order is matched.

🧠 Algorithmic liquidity vs “human” liquidity

The difference between algorithmic and “human” liquidity is tied to how synchronously limit orders are pulled. It is not tied to a general willingness to take execution risk.

Algorithmic liquidity is formed by strategies with formal risk models. The same inputs about volatility, order flow, and adverse selection lead to the same decisions for many participants.

“Human” liquidity is created by manual and discretionary participants. Decisions are not tied to shared thresholds. Pulling limits is not synchronized in time.

This is visible in the book. When algorithms dominate, limits vanish in a coordinated way. When manual traders participate, quotes remain in fragments and unevenly.

With a high share of algorithmic liquidity, the market is sensitive to collective risk-model decisions. Regime shifts are abrupt.

When “human” liquidity is present, changes are more gradual. The reason is that different participants pull limits under different execution conditions.

📦 How market makers manage quote size

Limit order size caps the magnitude of one fill and the speed of inventory risk accumulation. It does not show available liquidity.

Each fill forms an inventory position. The position becomes sensitive to mid-price movement until the next quote update.

Order size sets the risk of one fill. It defines the potential mark-to-market loss under an unfavorable price drift.

Quote size is computed after setting an acceptable inventory loss. It is not computed from visible market size.

1. Set the maximum acceptable loss from one fill.
2. Estimate volatility and market-order frequency.
3. Compute order size so one fill does not create a position imbalance.
4. Update quotes when execution tempo or the price range changes.

When market-order flow accelerates, the same size fills more often. Inventory accumulates faster.

When volatility rises, the same size creates higher risk between fills even without higher traded volume.

In unstable conditions, quote size compresses. In calm environments, quote size expands.

Book sparsity during uncertain periods is more often tied to smaller quote sizes, not market makers disappearing. In such a structure, a single aggressive order more often crosses multiple levels in thin books. The mechanics of pacing execution and its link to slippage are covered in VWAP and TWAP: how to enter and exit without slippage.

👀 Typical trader mistakes when watching market makers

Mistakes arise when the book is read as a static snapshot and pulling limits and execution risk are not accounted for.

Limit size in the book is perceived as a commitment to take a trade. A limit order can be pulled before matching.

Depth changes are treated as a directional signal. In practice, they reflect a recalculation of adverse selection risk after previous fills.

Support/resistance expectations are built without considering inventory risk. Inventory risk limits the lifetime of quotes at a level.

1. A large limit order appears at a level.
2. The size is interpreted as an intent to hold price.
3. The order is pulled as execution risk rises.
4. A market order passes the level without an opposing fill.

The spread is read as a direction indicator. The spread reflects compensation for unfavorable execution risk.

When the spread widens, the market enters a lower-liquidity phase. Price reacts to smaller aggressive size.

🧭 How to correctly understand the role of market makers

A market maker places limit orders as long as execution risk is compensated by the current spread.

A market maker does not hold price and does not “defend levels.” A quote exists only within acceptable execution risk.

Each fill changes the inventory position. PnL sensitivity to mid-price movement increases.

As adverse selection probability rises, size is reduced or quotes are pulled. Price direction is not a condition of this decision.

Market-maker liquidity is probabilistic: a limit can be pulled before it is matched with a market order.

Under one-sided order flow, each trade accelerates inventory imbalance accumulation.

In this phase, quoting stops because risk rises faster than spread compensation.

🛠️ A practical model of market-maker behavior

A market maker places limit orders as long as execution risk is compensated by the current spread and does not create an uncontrollable inventory imbalance.

A limit order is a request for a potential fill. Execution forms or increases an inventory position. PnL sensitivity to price movement rises until the next quote update.

The spread compensates for adverse selection and losses from price drift against the formed inventory between quote recalculations.

Pulling limits begins under one-sided order flow. A series of fills from one side accelerates inventory imbalance. Rebalancing cost increases.

1. A market order fills the market maker’s limit.
2. The fill increases inventory and PnL sensitivity to mid-price movement.
3. The algorithm recalculates adverse selection risk and the required spread compensation.
4. Quotes shift or are pulled when execution risk exceeds the acceptable threshold.