What Is Inventory Skew?

Inventory skew is when a market maker adjusts its bid and ask quotes because its holdings are unbalanced, nudging trades that push the position back toward a target ratio.

This is a crypto market-making term, not the retail one. If you searched “skew” and landed on stock-keeping-unit (SKU) inventory pages, you are in a different aisle. Here, “skew” means tilting your quotes on an exchange order book. It has nothing to do with warehouse counts.

Inventory skew shows up in two places that look separate but are the same idea. It is a behavior a market maker performs by hand or by rule. It is also a setting inside bots like Hummingbot. This page connects both, then shows how the setting quietly loses money when it is tuned badly.

Key takeaways

• Inventory skew shifts a market maker’s bids and asks to rebalance an unbalanced position toward a target ratio.
• It is different from spread. Spread is how far quotes sit from mid-price. Skew is the asymmetry of those quotes.
• In Hummingbot it lives in three settings: inventory_skew_enabled, inventory_target_base_pct, and inventory_range_multiplier.
• Badly tuned skew loses money through adverse selection, one-sided quoting in a trend, and venues with no hedge.

What Is Inventory Skew In Crypto Market Making?

A market maker quotes both sides of a market at once. It posts a bid to buy and an ask to sell, hoping to earn the gap between them. To do that safely, it wants to hold a balanced position, often a rough 50/50 split between the base asset and the quote asset.

Inventory skew is what happens when that balance drifts. If the maker ends up holding too much of the base asset, it tilts its quotes to attract sellers and slow down buyers. That tilt is the skew.

The word “skew” trips people up because it sounds like SKU, the stock-keeping-unit tag used in shop inventory systems. They are unrelated. Crypto inventory skew is about the shape of your quotes on an order book, not a product catalog.

Think of a market maker’s job as staying near the middle of a see-saw. Every filled order pushes the position toward one side. Skew is the lean the maker adds to its quotes to walk itself back to center. This behavior is core to how any active market maker manages risk, and it is the same idea whether a human runs it or a bot does.

How Inventory Skew Works Step By Step

Start with a balanced maker holding an even split. Its bid and ask sit at equal distances on either side of the mid-price. Buys and sells arrive at a similar pace, and the position stays near target.

Now a run of buyers hits the ask. The maker sells and sells until it is net short the base asset. Its inventory has swung to one side. Left alone, it now carries directional risk it never wanted.

Skew corrects this. Because the maker is short the base asset, it wants to buy some back. So it shades both quotes upward and makes its buy order larger than its sell order. A higher, bigger bid attracts sellers. A higher, smaller ask discourages more buying. Trades that follow tend to refill the maker’s base holdings.

The mirror case runs the other way. If the maker is net long the base asset, it shades both quotes down and makes its sell order larger than its buy order. That pulls in buyers and slows sellers, bleeding off the excess.

The reference point for all of this is the mid-price and the target ratio. The further the position drifts from target, the harder the skew. As inventory returns to center, the skew relaxes and the quotes go symmetric again.

The pendulum swings both ways, and it never fully stops. A live market maker is always slightly off-target and always leaning back. That constant correction is the whole point.

One warning lives inside this mechanism. Skewed, one-sided quotes are exactly what fast liquidity hunts look for. When your bids sit lower and larger during a sell-off, a sharp move down can sweep them before the trend reverses, leaving you long into more weakness.

Inventory Skew Vs Bid-Ask Spread

People conflate spread and skew constantly, so pin them apart. Spread is the distance between your bid and your ask around the mid-price. It is how a market maker earns. A wider spread means more profit per round trip but fewer fills.

Skew is not distance. It is asymmetry. It is how far your quotes shift to one side and how you size your two orders relative to each other. Spread sets the width. Skew sets the tilt.

The table below lines up the two levers so the difference is obvious.

Spread	Inventory skew
Distance of quotes from mid-price	Sideways shift and sizing of those quotes
Controls profit per trade and fill rate	Controls how fast inventory returns to target
Symmetric by default on both sides	Deliberately asymmetric when unbalanced
Widens to protect against volatility	Tilts to protect against directional exposure

A working market maker uses both at once. It sets a spread wide enough to earn and survive volatility, then skews within that spread to keep inventory near target. Widening the spread is not a substitute for skewing. One manages the cost of quoting, the other manages the direction of your holdings.

For those who cannot run a market maker themselves, copy trading strategies offer a different route to market exposure without touching an order book.

Inventory Skew In Hummingbot: Target Base Percent And Range Multiplier

Hummingbot is an open-source market-making bot, and it turns this behavior into three settings. The first is inventory_skew_enabled. Switch it on and the bot starts sizing its two orders asymmetrically based on how far your position sits from target. Leave it off and the bot quotes symmetric sizes regardless of your balance.

The second setting is inventory_target_base_pct. This is the ratio you want to hold, expressed as the base asset’s share of your total value. Set it to 50% and the bot aims for an even split between base and quote. Set it lower and it wants to hold less of the base asset.

The third is inventory_range_multiplier. This widens or narrows the tolerance band around your target. A small multiplier makes the bot react hard to tiny imbalances. A large one lets inventory wander further before the skew bites. It scales how aggressively order sizes shift as you drift.

A worked example makes it concrete. Say you run BTC-USDT with 10 BTC of total value, a 50% target, and a base order size of 1 BTC. If your position drifts to 70% BTC, the bot shrinks your buy order and grows your sell order, so more BTC leaves than arrives. As you fall back toward 50%, the two order sizes converge again. The adjustment is roughly linear across the tolerance band.

The target does not have to be dollar-denominated. On a pair like ETH-BTC, the base is ETH and the quote is BTC, so your target ratio is measured against BTC, not USD. That matters for anyone quoting crypto-to-crypto pairs where neither leg is a stablecoin.

How Inventory Skew Connects To Avellaneda-Stoikov And Reservation Price

Many arrive at this term through the Avellaneda-Stoikov model, published by Marco Avellaneda and Sasha Stoikov. It is the academic backbone for a lot of modern market making, and skew lives inside it in a subtler form.

The model does not quote around the public mid-price. It quotes around a private number called the reservation price. This is the maker’s own adjusted fair value, and it drifts away from mid based on current inventory, volatility, and how much time is left in the trading horizon.

Inventory is the key driver of that drift. When the maker is long the base asset, the reservation price sinks below mid, pulling both quotes down to encourage selling. When the maker is short, it rises above mid. That shift is inventory skew, expressed as a moving fair value rather than a config percentage.

So the two framings are the same idea at different altitudes. Config-based skew in a bot is the plain-English cousin of the reservation-price shift in Avellaneda-Stoikov. One nudges order sizes by a target percent. The other nudges a private fair value by a risk-aversion term. Both answer the same question: my inventory is off, which way should I lean my quotes?

Because the reservation price leans away from your holdings, a heavily skewed quote also signals directional pressure. A maker quoting far off mid is telling the order book which way its inventory is straining.

When Inventory Skew Goes Wrong (And Loses Money)

Skew is a risk tool, but a badly tuned version is its own risk. The first failure mode is adverse selection. If your spread is too tight, informed traders pick off your quotes right before the price moves against you. Skew keeps rebalancing your inventory, yet you still bleed on every fill because the fills themselves are bad.

The second failure mode is trend. Skew assumes prices mean-revert, so it leans into the move that just happened. In a strong one-way trend, that means it keeps buying an asset that keeps falling. Your one-sided quotes get run over, and rebalancing toward target just means catching a falling knife with a bigger buy order.

The third failure mode is venues with no hedge. On some prediction-market platforms like Kalshi, you cannot offload an unbalanced position with an offsetting instrument. Skew can rebalance within the book, but if the whole book moves against you, there is nowhere to lay off the risk. A too-aggressive range multiplier can also force you to unwind a position at a loss just to hit a target that no longer makes sense.

Before you blame skew for a bleeding bot, run a short check.

• Is your spread wide enough to survive adverse selection, or are informed traders picking you off?
• Are you quoting into a strong trend where mean reversion is not happening?
• Does the venue even have a hedge or offset, or are you stuck holding the position?
• Is your range multiplier so aggressive that it unwinds positions at a loss?

A maker whose skew fails ends up holding an unwanted pile of the base asset. That is the same trap the bagholder mental model describes, and it connects directly to exit liquidity risk when you finally try to sell and no one is buying. On thin no-hedge venues, a comparison of the best prediction market platforms helps you pick venues where inventory is easier to manage.

Frequently Asked Questions About Inventory Skew

What does inventory skew mean in market making?

Inventory skew means a market maker tilts its bid and ask quotes because its holdings are unbalanced. It shades both quotes and resizes its two orders to attract the trades that push its position back toward a target ratio, usually a rough even split between the base and quote assets.

How is inventory skew different from bid-ask spread?

Spread is the distance between your bid and ask around the mid-price, and it sets how much you earn per trade. Inventory skew is the asymmetry of those quotes, the sideways tilt and order sizing you use to steer inventory back to target. Makers use both at the same time.

Is inventory skew the same as inventory risk?

No. Inventory risk is the danger of holding an unbalanced position that moves against you. Inventory skew is one of the tools you use to manage that risk. Skew adjusts your quotes to rebalance, while inventory risk is the underlying exposure that skew is trying to reduce.

Can inventory skew make a market-making bot lose money?

Yes. Skew can lose money through adverse selection on a too-tight spread, through one-sided quoting into a strong trend, and on venues with no hedge instrument where inventory cannot be offloaded. An over-aggressive range multiplier can also unwind positions at a loss chasing a target.

Is inventory skew the same as SKU inventory?

No. SKU stands for stock keeping unit and refers to retail product tracking in shop and warehouse systems. Crypto inventory skew is a market-making term about tilting your quotes on an exchange order book. The two only share a similar sound and have nothing else in common.

Where To Start

Inventory skew is easiest to internalize by watching it, not just reading it. A few practical next steps make the concept stick.

• Load a paper-trading Hummingbot instance and toggle inventory_skew_enabled to see order sizes shift.
• Set inventory_target_base_pct to 50% and let a volatile pair drift, then watch the bot lean back to center.
• Compare a wide-spread run against a tight-spread run to feel adverse selection firsthand.
• Test on a pair with a real hedge before ever risking capital on a no-hedge venue.
• Track your inventory ratio in a log so you can tell whether skew is helping or fighting the trend.

The Avellaneda-Stoikov paper by Marco Avellaneda and Sasha Stoikov, “High-frequency trading in a limit order book,” is the deeper reference once the plain-English version clicks.