Bid-ask spread mikrostruktura — głębiej niż sądzisz

Glosten-Milgrom + Stoll modele decompose spread w 4 komponenty: (1) Order processing cost (~10% spread): technical infrastructure dealer/broker. Servers, connectivity, employee salaries, regulation compliance. Fixed cost amortized across trades. Mini niż przed elektronifikacja (90s = 50%+). (2) Inventory holding cost (~20-30%): market maker holds inventory long/short. Position risk while waiting opposite-side counterparty. Volatile markets = higher inventory risk = wider spread compensation. (3) Adverse selection (~30-50%): BIGGEST component. Market maker trades with mix random + informed traders. Informed = those z information edge (institutional, central bank intervention insiders). Każda transakcja z informed = market maker loses. Spread compensates expected losses. Economics: information asymmetric → spread wider. News events = info edge higher → spread widens dramatically (e.g. NFP +20 pips EUR/USD). (4) Market maker profit margin (~10-30%): pure profit. Competition reduces this. Single broker monopoly = high. ECN multi-LP = low. Sum: 4 komponenty = total spread. EUR/USD 0.5 pip = 0.05 processing + 0.1-0.15 inventory + 0.15-0.25 adverse selection + 0.05-0.15 profit margin.

Adverse selection = SINGLE biggest spread driver. Concept: market maker provides liquidity. NIE wie czy trader prze nim = random retail OR informed (central bank insider, hedge fund z research). Bayesian update: gdy trader buys, market maker thinks "maybe knows something". Updates probability informed trader. Implication: każda trade po widely informed period = market maker assumes informed. Pushes price away. Example NFP release: pre-NFP, EUR/USD spread 0.5 pips. NFP release 14:30: spread widens 5-15 pips (10-30× normal) dla 30-60 sec. Market makers afraid informed traders front-running data. Empirical evidence: spread positively correlated z trading activity (volume = informed activity proxy), volatility (information arrival), asymmetric information (announcements). Why ECN tightens spread: multi-dealer competition. Każdy dealer estimates adverse selection independently. Aggressive ones bid tighter. Implication trader: tight spread = pretty random, low informed presence. Wide spread = high informed presence (news, low liquidity). NIE trade w wide spread periods if NIE informed yourself. Pro insight: informed traders profit asymmetric information. Retail = uninformed. Trade w tight spread periods.

Predictable spread widening events. Major economic releases: (1) NFP (US Non-Farm Payrolls) first Friday: 14:30 GMT, +10-25 pips dla EUR/USD 30-60 sec. (2) FOMC decision: 18:00 GMT, +5-15 pips. (3) ECB decision: 11:45 GMT (rate) + 12:30 GMT (presser), +5-15 pips. (4) CPI inflation (US/EU): +5-10 pips. Low liquidity periods: (1) Asian session 23:00-06:00 GMT: spread 2-3× normal. (2) Holidays (Christmas, NY, Easter): spread 3-5× normal. (3) Friday close 21:00-22:00 GMT: gradual widening. (4) Bank holidays specific countries: pair affected (e.g. UK holiday → GBP pair widens). Risk-off events: financial crises, geopolitical (Ukraine war Feb 2022 EUR/USD spread 5-10 pips wider weeks). Trading strategy: (1) Avoid trading w obvious widening periods unless news strategy. (2) Reduce position size if entering near event. (3) Use limit orders (NIE market orders) dla protection. (4) Set wider SL: spread widening can stop you out unfair. Pre-event preparation: 30 min przed major release = wszystkie open positions reviewed. SL adjusted, size potentially reduced. Calendar check Forex Factory free.

FX market microstructure = academic field studying mechanics rynek wymiany walut. Pioneers: Lawrence Glosten (Columbia), Paul Milgrom (Stanford), Maureen O'Hara (Cornell), Carol Osler (Brandeis). Glosten-Milgrom 1985: foundational paper "Bid, Ask and Transaction Prices in a Specialist Market with Heterogeneously Informed Traders". Introduced adverse selection model. Stoll 1989: "Inferring the Components of the Bid-Ask Spread". Empirical decomposition NYSE. O'Hara 1995: "Market Microstructure Theory" book. Comprehensive review. Modern research: FX-specific (Cheung-Chinn-Marsh-Pinell), high-frequency dynamics, dealer behavior. Practical applications: (1) Optimal execution algorithms (institutional). (2) Spread prediction models. (3) Market quality measurement. (4) Regulatory policy (e.g. MiFID II Europe). Trader application: understanding microstructure helps explain spread variability beyond "broker greed". Spread widens because economic forces. Provides framework dla strategy timing. Academic resources: Journal of Finance, Journal of Financial Economics, Review of Financial Studies. Most papers behind paywalls. Free: NBER working papers, SSRN. Trader-friendly books: O'Hara "Market Microstructure", Hasbrouck "Empirical Market Microstructure".