Adverse Selection for Market Makers
Market makers face a persistent, quantifiable threat: that every trader who approaches them knows something they don't. A trader who suddenly wants to buy 100,000 shares might be acting on material nonpublic information about an upcoming acquisition. A trader who suddenly wants to sell might have advance knowledge of disappointing earnings. If the market maker executes at quoted prices before this information becomes public, they'll be at a severe disadvantage. This risk—that the market maker is trading with someone better informed—is called adverse selection. It's one of the most important forces shaping bid-ask spreads and market maker behavior. Understanding adverse selection reveals why spreads widen around earnings announcements, why options spreads are so wide compared to stock spreads, and why market makers obsess over detecting informed trading.
Quick definition: Adverse selection is the market maker's risk of trading with counterparties who possess superior information, potentially leading the market maker to execute at prices that are immediately disadvantageous as the market reprices based on that information.
Key Takeaways
- Adverse selection cost is often the largest component of bid-ask spreads
- Market makers adjust spreads and order preferences based on signals that informed trading is occurring
- Large orders, unusual order timing, and bid-ask imbalances are red flags for informed trading
- Earnings announcements, corporate events, and regulatory news create periods of extreme adverse selection risk
- Options markets face higher adverse selection costs than stock markets due to complexity and leverage
- Market makers use sophisticated algorithms to detect and adjust for informed trading
- Zero-informed-trading scenarios (like mechanical index rebalancing) see near-zero adverse selection costs
What Is Adverse Selection?
Adverse selection is fundamentally about information asymmetry. In a typical financial transaction, both parties believe they're getting a fair deal. A buyer pays what they think is a fair price; a seller accepts what they think is fair. But if one party has better information than the other, that equilibrium breaks down.
Consider two scenarios:
Scenario 1: Informed Trader A trader learns (through superior analysis or access to information) that a company will announce better-than-expected earnings tomorrow. They approach a market maker and buy 50,000 shares at the current ask price of $100. The market maker sells at $100. The next day, earnings are announced as expected, and the stock jumps to $105. The trader who bought at $100 now has a $5 per share gain ($250,000 total). The market maker, who sold at $100, has lost $5 per share ($250,000 total). The market maker traded with an informed counterparty and lost.
Scenario 2: Uninformed (Noise) Trader A retail investor has decided to diversify their portfolio and buys 50,000 shares at $100, the current ask price. The stock doesn't have any upcoming catalysts and trades randomly, moving up and down based on broader market movements. The market maker who sold has no special disadvantage—they're on equal informational footing with the buyer.
The adverse selection problem is that market makers can't always distinguish Scenario 1 from Scenario 2 in real time. They see an order for 50,000 shares and must decide: is this an informed trader about to take advantage of me, or just a noise trader? If it's informed, they should widen their spread to protect themselves. If it's noise, a wide spread is unnecessary and loses them trading volume.
This distinction—informed vs. uninformed—drives market maker behavior more than almost any other factor.
Historical Origins of Adverse Selection Theory
The formal study of adverse selection in market microstructure traces to research by economists including George Akerlof (famous for the "lemons problem" in used cars), followed by financial economists like George Stiglitz and others who applied information economics to markets.
In the context of securities markets, the seminal work came from Thomas Ho and Hans Stoll in the 1980s and subsequent refinement by researchers including Charles Jones, Yakov Amihud, and others. The key insight was that market makers must set spreads wide enough that even if they trade with informed traders some percentage of the time, they break even on average.
The mathematical model expresses this as:
Adverse Selection Cost = (Probability of Trading with Informed Trader) × (Informed Trader's Information Advantage)
If the probability of informed trading is 30% and the average information advantage is $0.05 per share, then the adverse selection cost is approximately $0.015 per share. To break even, spreads must be at least this wide (accounting for both buy and sell sides).
Detecting Informed Trading: The Signal Problem
Market makers can't observe whether a trader is informed. Instead, they observe signals and patterns that suggest information asymmetry. The key signals include:
Order Size: Large orders are red flags. A trader placing a 100,000-share order when typical orders are 1,000-5,000 shares is revealing something about their intensity or confidence. This intensity might signal informed trading.
Order Direction: When orders arrive heavily skewed toward buying, it suggests positive information is known. Heavy selling suggests negative information. Market makers track the balance of buy vs. sell order flow and adjust their spreads accordingly.
Order Urgency: Orders that demand execution immediately (market orders) are less suspicious than orders patient enough to wait for better prices (limit orders). A trader willing to accept a worse price to wait might be less informed and more price-sensitive. A trader willing to pay market prices immediately is either desperate to execute or very confident in their price target.
Timing Pattern: Orders that arrive suddenly after news or catalyst events suggest informed trading. Orders that arrive on a regular, mechanical schedule (like daily portfolio rebalancing) suggest noise trading. For example, institutional investors rebalancing quarterly holdings on options expiration days can be distinguished from informed traders by their predictable patterns.
Volatility Context: Large orders during calm markets are more suspicious than large orders during volatile markets. During calm periods, if someone shows up with a massive buy order, why aren't they patient and waiting for better prices? The immediacy suggests private information. During volatile markets, urgent order flow is normal as many traders react to public events.
Related Securities: If a trader buys the stock but also shorts the call options (betting the stock will move less than expected) or engages in other hedging, it suggests they know something but are being careful. If a trader buys the stock and also buys call options (betting the stock will move more), they appear to be speculating, not informed about a directional move.
Bid-Ask Imbalance: When market makers observe that all incoming orders are one-sided (all buying or all selling), they infer that either positive or negative information is known. They widen spreads to protect against the risk that this imbalance reflects information.
Market Microstructure Theory and the Glosten-Milgrom Model
One of the most influential formal models of adverse selection in markets is the Glosten-Milgrom model (1985), developed by Lawrence Glosten and Paul Milgrom. This model shows how market makers rationally adjust spreads based on order flow in the presence of informed traders.
The model works as follows: A market maker knows that some fraction of incoming orders come from informed traders and some from uninformed traders. When they see an order, they use Bayesian updating to estimate whether it came from an informed or uninformed trader.
If the market maker sees a large buy order, they think: "This could be informed buying (because the trader knows good news) or uninformed buying (random). Given that I see this buy order, I should update my estimate of the probability it's informed." If they conclude there's a high probability of informed buying, they:
- Widen their ask price to compensate for the risk
- Tighten their bid price to discourage selling (which would mean they're long inventory while informed traders are buying)
- Possibly withdraw from the market entirely if the informed-trading probability is very high
The model shows that spreads naturally widen when order imbalances are extreme, because extreme imbalances indicate a high probability of informed trading.
The Earnings Announcement Period: Adverse Selection in Action
The clearest real-world manifestation of adverse selection risk is the widening of spreads around earnings announcements. Weeks before earnings, spreads in a typical stock might be 1-2 cents. As earnings approach, spreads widen to 3-5 cents. In the hours immediately before earnings are announced, spreads might widen to 10-15 cents.
Why? Because earnings announcements represent a binary event—the announced results will either be better than expected, worse than expected, or in line with expectations. Traders who have better analysis, data, or access to information may know the direction before public announcement. Market makers, aware of this, widen spreads to protect themselves against trading with traders who may know the earnings results.
Empirically, research has documented that this widening is entirely driven by adverse selection risk. Inventory risk (the risk of holding the stock) doesn't increase on earnings day. Volatility doesn't spike before earnings are announced. But information asymmetry—the probability of trading with informed traders—increases dramatically. Market makers respond by widening spreads proportionally to the increase in adverse selection risk.
After earnings are announced and results become public, spreads typically contract sharply. The information is now public, so the advantage that informed traders had is gone. The information advantage shrinks toward zero, and spreads revert to normal.
Other High-Adverse-Selection Scenarios
Earnings announcements are the most obvious example, but many other corporate and economic events create windows of high adverse selection risk:
Merger and Acquisition Announcements: Before a M&A deal is announced, traders with inside knowledge may begin buying the acquiring company's stock (if they believe the deal will be dilutive to earnings, the stock might fall, creating short-selling opportunities). Market makers, suspicious of sudden order imbalances, widen spreads. When the deal is announced, information becomes public and spreads revert.
Regulatory Decisions and Approvals: Industries subject to regulatory approval (pharmaceuticals, utilities, financial services) face cycles of regulatory decision points. Before a ruling, information about the likely decision may leak to insiders or be inferred from pattern recognition. Spreads widen.
Economic Data Releases: Major economic releases (jobs reports, inflation data, GDP) create windows where traders with better economic analysis or information might have advantages. Spreads typically widen just before major data releases and contract sharply after the data is released.
Corporate Actions: Stock splits, dividend announcements, credit rating decisions—all create information asymmetry risks. Market makers widen spreads around these events.
Insider Trading and Information Leakage: In rare but serious cases, material nonpublic information leaks to traders, creating severe adverse selection risk. The SEC investigates suspected insider trading, but the damage to market makers happens in real time, before the SEC intervenes.
Adverse Selection vs. Volatility: Disentangling the Costs
A key finding in academic research is that adverse selection and volatility are distinct drivers of spreads, and they operate independently. This is important because it clarifies how much of the spread is "bad" (compensation for trading with informed traders) versus "natural" (compensation for inventory risk in a volatile environment).
Volatility-driven spreads: When a stock's price becomes more volatile (perhaps due to upcoming earnings or external shocks), the spread widens because inventory risk increases. This is a natural, necessary adjustment. Investors benefit because more volatility means more uncertainty, and market makers need compensation for taking on that uncertainty. When volatility subsides, spreads contract.
Adverse selection-driven spreads: When information asymmetry increases (because a binary event is pending), spreads widen beyond what volatility alone would predict. This component represents the cost of trading with informed traders.
Research by Stoll (1989), Huang and Stoll (1997), and others shows that in high-information-asymmetry periods (like the day before earnings), the adverse selection component of spreads can be 2-3x the volatility component.
The Role of Large Traders and Block Trades
Large block trades—purchases or sales of millions of shares in a single transaction—face particularly high adverse selection costs. When a trader announces they want to buy or sell a massive block, the market immediately asks: why? Is this an informed trader with time-sensitive information? Or is it a mutual fund rebalancing or a pension plan adjusting its allocation?
Block traders address adverse selection risk through several mechanisms:
Negotiated Pricing: Instead of hitting the market's public quotes, block traders negotiate prices with market makers directly. This allows both sides to share information and come to a mutually acceptable price that compensates the market maker for adverse selection risk if warranted.
Search for Counterparties: Block traders use brokers to search for natural counterparties (traders on the opposite side of the trade). If a seller wants to sell 5 million shares and a buyer wants to buy 5 million shares, the transaction happens at a single negotiated price, eliminating the need for inventory accumulation by market makers. This reduces adverse selection risk significantly.
Dark Pool Transactions: Large traders often execute blocks in dark pools—private venues where orders are not visible to the public market. Dark pools reduce information leakage and give traders some protection against executing at public prices that embed large adverse selection costs. However, dark pools have their own complexities and risks.
Gradual Execution: Rather than execute a massive block at once, traders may execute gradually over hours or days, reducing the intensity of the adverse selection problem.
The Lemons Problem and Bid-Ask Spreads
There's a subtle connection between Akerlof's "lemons problem" (where adverse selection causes market breakdown because sellers of quality goods can't distinguish from sellers of "lemons") and bid-ask spreads in securities markets.
In the lemons model, when buyers can't distinguish quality goods from poor ones, sellers of quality goods are forced to accept lemons prices. Some sellers of quality goods exit the market, reducing supply and worsening the average quality of remaining goods. This process can spiral until the market breaks down entirely.
In securities markets, this would work as follows: If all traders were informed (trading on private information), market makers would require spreads so wide that uninformed traders couldn't execute economically. Uninformed traders would exit the market. With no uninformed traders, the adverse selection risk would be 100% (every remaining trader is informed), requiring even wider spreads. Eventually, market makers might exit entirely.
In practice, this catastrophic scenario doesn't occur because there are always some uninformed traders (noise traders, rebalancers, hedgers) mixed with informed traders. This mixture keeps the market functional. But during extreme stress (like the 2008 crisis), when the proportion of uninformed traders falls sharply, spreads can widen dramatically, and market makers do exit.
Detecting Adverse Selection: Algorithmic Approaches
Modern market makers employ sophisticated algorithms to detect adverse selection in real time. These systems monitor:
Order Flow Patterns: Machine learning models identify whether incoming order flow is consistent with random noise or exhibits directional bias suggesting informed trading.
Price Impact Analysis: The algorithm measures how much price movement follows large orders. If price movement is consistent with the informed trading hypothesis (large buy orders followed by price increases), the adverse selection risk is flagged as high.
News and Event Monitoring: The algorithm continuously monitors news wires, SEC filings, social media, and other information sources for events that might trigger informed trading. When an event risk is detected, the algorithm flags the security as high-adverse-selection risk and adjusts spreads proactively.
Volatility and Volatility Forecasting: The algorithm distinguishes volatility-driven spread widening from adverse-selection-driven widening by explicitly forecasting volatility and comparing actual spreads to volatility-implied spreads.
Cross-Venue Information: Market makers can see orders and trades across multiple venues (exchanges, dark pools, ECNs). Unusual patterns across venues suggest informed trading.
Trade-Backs and Price Reversals: If a market maker buys at $100 from a trader and, moments later, has to sell at $99 to a different trader, they've suffered an adverse selection loss. The algorithm tracks these reversal patterns and increases adverse selection estimates when reversals are frequent.
Empirical Evidence on Adverse Selection Costs
Academic research has quantified the adverse selection component of spreads through various methodologies. Two main approaches:
Statistical decomposition: Researchers separate bid-ask spreads into components attributable to adverse selection, inventory risk, and operational costs using statistical models. A landmark study by Stoll (1989) estimated that adverse selection accounts for 25-50% of spreads in actively traded stocks.
Event studies: Researchers examine spread changes around events where adverse selection risk changes but other factors remain constant. Earnings announcements are ideal for this purpose. Spreads widen by 50-100% on earnings day, with most of this widening attributable to adverse selection rather than volatility changes.
Estimation from order flow: Researchers use the Glosten-Milgrom model to estimate the probability of informed trading based on observed order flow and spreads. In most stocks, the estimated probability of informed trading is 10-30%, with higher probabilities in stocks with binary events pending.
Cross-security comparisons: Securities with similar volatility but different adverse selection risk characteristics show different spreads. For example, an index fund and a single stock with the same volatility have different spreads because the fund faces lower adverse selection risk (it's a mechanical holder, not an active trader seeking information advantage).
Adverse Selection and Options Markets
Options markets face particularly severe adverse selection risks. This is because options are leveraged instruments, and small changes in the underlying stock can create large profits for informed traders. An informed trader who knows a stock will move 5% might not bother trading the stock itself (they'd need to deploy capital equal to the position size). But they'll be very interested in trading options, where they can achieve leveraged exposure with less capital.
For example, suppose an informed trader knows a stock will rise from $100 to $103. They could:
- Buy 100 shares for $10,000 and profit $300 (3% return)
- Buy 100 call options ($100 strike) for, say, $500 total and profit $2,000+ (400% return)
The options trade is far more profitable for informed traders. Market makers, aware that options attract informed trading disproportionately, set much wider spreads in options than in the underlying stock.
Empirically, options bid-ask spreads (as a percentage of option price) are 10-100x wider than stock spreads. Much of this difference is attributable to higher adverse selection risk in options.
Real-World Adverse Selection Examples
Tesla Before Earnings: In the hours before Tesla reports earnings, spreads in TSLA stock widen from 1-2 cents to 10-15 cents. After earnings are announced, spreads revert to 1-2 cents within seconds. The widening is driven entirely by adverse selection risk—traders with better models of Tesla's financials may know the direction of the surprise.
Pharmaceutical Stocks and FDA Decisions: A biotech company awaiting FDA approval of a drug faces extreme adverse selection risk. Spreads in the company's stock might widen to 5-10 cents (or more) in the weeks before the decision. Insiders or traders with good regulatory analysis might know the likely decision. When the decision is announced, adverse selection risk collapses and spreads contract sharply.
Interest Rate Decisions: Immediately before the Federal Reserve announces interest rate decisions, spreads in rate-sensitive stocks (financial services, utilities) widen. Market makers know traders might be trading ahead of the decision based on superior economic analysis. When the Fed announces, spreads contract.
Brexit Vote: In the days leading up to the June 2016 Brexit referendum, spreads in UK-exposed stocks widened significantly. Spreads in European bank stocks widened because traders might have information about the referendum's outcome. When the result was announced (narrowly favoring Leave), spreads contracted sharply.
Common Mistakes in Understanding Adverse Selection
Mistake 1: Assuming all spread widening is adverse selection. Volatility also widens spreads. On earnings day, both adverse selection risk and volatility spike. But the two are distinct, and spreads widen for both reasons.
Mistake 2: Thinking adverse selection is always bad for market makers. In fact, while adverse selection creates costs, it also creates opportunities. Sophisticated market makers use order flow analysis to infer informed traders' intentions and adjust their own inventory accordingly, often profiting from this information.
Mistake 3: Confusing adverse selection with price impact. Price impact is the immediate cost a trader incurs from moving the market (executing a large order that moves prices). Adverse selection is the longer-term cost from trading with informed counterparties. They're related but distinct.
Mistake 4: Assuming informed traders always profit. Informed traders often do profit, but if their information is wrong or if they misunderstand the market structure, they can lose. Market makers sometimes out-trade the traders they thought were informed.
Mistake 5: Thinking adverse selection only occurs around major events. While it's most obvious around earnings, adverse selection is a continuous phenomenon. Traders with superior analysis or earlier access to information are trading every day, creating persistent adverse selection costs.
FAQ
Q: How can I tell if I'm trading with a market maker who suspects adverse selection? A: If spreads widen suddenly or unexpectedly, or if your market order incurs higher slippage than usual, adverse selection risk may have increased. Monitor the bid-ask spread and compare it to recent history.
Q: Is trading ahead of earnings a form of insider trading? A: Not necessarily. Trading based on superior fundamental analysis is legal. Insider trading involves trading on material nonpublic information obtained through a breach of duty. The distinction is legally complex, but the key is whether the information came through proper channels (public analysis) or improper channels (insider access).
Q: Can I profit from adverse selection? A: Retail investors can't directly. But if you trade against a market maker who's facing high adverse selection risk, you might face wider spreads, which is a cost to you. Institutional traders with speed advantages and better information can sometimes profit.
Q: How do market makers prevent themselves from being picked off by informed traders? A: They adjust spreads based on order flow signals, use hedging strategies, maintain algorithms that detect informed trading patterns, and withdraw from the market entirely if adverse selection becomes too severe.
Q: Are informed traders always profitable? A: No. Many traders believe they're informed but aren't. They may trade on incorrect information or misinterpret public information. Trading with what you think is an information advantage is risky if that advantage is illusory.
Q: How does adverse selection differ across market venues? A: Adverse selection risk is highest on public exchanges where information is visible to all traders simultaneously. Dark pools have lower adverse selection risk because information about trades is hidden. Registered market makers on exchanges face higher adverse selection than broker-dealers in private markets.
Q: Can spreads widen without adverse selection risk? A: Yes. Spreads widen when volatility increases, when inventory imbalances are severe, when liquidity providers are scarce, and when operational costs rise. Adverse selection is just one driver.
Related Concepts
- Moral Hazard: Related to adverse selection; the risk that one party will behave differently after a transaction, knowing they're protected against downside
- Information Advantage: The edge a trader has from accessing better or earlier information than competitors
- Volatility: Often confused with adverse selection but distinct; compensation for inventory risk rather than information risk
- Bid-Ask Bounce: The tendency for prices to oscillate between bid and ask, driven by the spread itself and order flow patterns
- Market Microstructure: The broader field studying how markets function at the transaction level
- Lemons Problem: Akerlof's foundational insight about adverse selection in markets where quality is hidden
External Resources
Research on adverse selection and market microstructure:
- SEC Office of Market Intelligence
- FINRA Market Integrity Resources
- Investor.gov Information on Inside Trading
Summary
Adverse selection—the risk of trading with informed counterparties—is one of the most important forces in financial markets. Market makers face this risk constantly and adjust spreads and quoting behavior accordingly. When information asymmetry is high (around corporate events, earnings announcements, or regulatory decisions), spreads widen dramatically. When information is public and symmetric, spreads tighten.
The magnitude of adverse selection costs varies dramatically across securities. Highly liquid, widely followed stocks face lower adverse selection costs because information leaks quickly to the broad market. Thinly traded, complex securities face higher adverse selection costs. Options face much higher adverse selection costs than their underlying stocks.
Understanding adverse selection is critical for understanding market structure, spreads, and the profitability of market making. It explains why market makers obsess over detecting informed traders, why they adjust spreads so rapidly in response to order flow, and why they sometimes exit the market entirely when adverse selection risk becomes overwhelming.