Foreign exchange looks simple on a screen: EUR/USD at 1.08, USD/JPY at 155, a two-way price and a button to trade. Underneath that quote sits one of the most complex market structures in global finance, a fragmented network of bank dealers, electronic communication networks, hedge funds, asset managers, corporates, high-frequency market makers and retail brokers. The Bank for International Settlements measured average daily FX turnover at $7.5 trillion in April 2022, but that headline number hides the real issue for traders: not all liquidity is equal, and not every participant sees the same market.
Microstructure matters because it determines execution quality. It explains why EUR/USD can trade with a 0.1 pip spread in London morning but gap violently during a Federal Reserve press conference, why USD/JPY liquidity thins at the Tokyo fix, why emerging-market currencies often trade in non-deliverable forwards rather than spot, and why retail stop-loss clusters can become fuel for institutional algorithms. In an era of central bank divergence, JPY-funded carry trades and episodic dollar squeezes, the plumbing of FX is not a technical footnote. It is the market.
The FX Market Is a Dealer Network, Not a Central Exchange
Unlike equities, where liquidity is anchored to exchanges such as Nasdaq or the Tokyo Stock Exchange, spot FX is predominantly over-the-counter. Prices are made by dealers and market makers across bilateral relationships, electronic platforms and prime brokerage arrangements. BIS data show spot transactions accounted for roughly $2.1 trillion of daily turnover in 2022, while FX swaps reached $3.8 trillion, confirming that the market is as much about funding and balance sheet management as directional currency speculation.
The hierarchy begins with the largest dealer banks: JPMorgan, Citi, UBS, Deutsche Bank, Barclays, HSBC and Goldman Sachs remain central nodes because they internalize client flow, warehouse risk and stream prices to thousands of counterparties. Their franchise is not just balance sheet; it is information. A bank that sees Japanese life insurers hedging Treasuries, macro funds buying USD/JPY options, and corporates selling dollars into month-end has a better read on flow than a trader looking at a single chart.
Interdealer venues such as EBS and Refinitiv Matching historically set the reference price for major pairs. EBS has been especially important for EUR/USD, USD/JPY and EUR/JPY, while Reuters Matching has carried weight in sterling and commodity currencies. Over the last decade, however, liquidity has fragmented toward multi-dealer platforms such as FXall, 360T, Currenex and Cboe FX, as well as single-dealer platforms operated by banks. The result is a market where the best price may not be public, and where execution depends on credit, technology and relationship tiering.
Banks Internalize Flow Before It Hits the Street
The most misunderstood feature of FX microstructure is internalization. When a client sells EUR/USD to a bank, that trade does not automatically go to the broader market. The bank may match it against another client buying euros, hold the position temporarily, or hedge only part of the risk. High internalization ratios reduce market impact for clients, but they also mean visible market volumes can understate true activity.
This is why the same economic headline can produce different price action across pairs. In EUR/USD, deep internalization and dense algorithmic liquidity often absorb moderate surprises. In USD/JPY, where positioning can be more one-sided around carry trades and intervention risk, the same dollar move may travel further because dealers are less willing to warehouse yen risk near official pain thresholds. When Japan’s Ministry of Finance intervened in 2022, the market learned that visible liquidity in USD/JPY can disappear precisely when the quoted depth looks comfortable.
Dealer balance sheets are also cyclical. Around quarter-end, year-end and regulatory reporting dates, banks become more selective with balance sheet-intensive trades, especially FX forwards and swaps. Cross-currency basis can widen not because investors suddenly changed their view on the euro or yen, but because dollar funding capacity tightened. For macro funds, confusing liquidity premium with fundamental valuation is an expensive mistake.
Funds Trade Information, Liquidity and Time Horizons
Institutional funds do not all interact with FX in the same way. A global macro hedge fund may express a Federal Reserve versus Bank of Japan divergence view through USD/JPY spot, options and rates futures. A real-money asset manager may hedge a portfolio of U.S. equities back into euros at month-end. A CTA may buy the dollar mechanically when trend signals flip. A sovereign wealth fund may rebalance quietly over several weeks to avoid moving the market.
These differences create predictable liquidity patterns. The 4 p.m. London WM/Reuters fix is a major execution window for passive equity and bond hedging. Ahead of the fix, banks and execution algorithms estimate imbalances from client orders, and market impact can be material in less liquid pairs. Academic and regulatory scrutiny after the 2013-2014 FX benchmark investigations changed behavior, but the fix remains one of the few moments when time-specific demand concentrates in a market that is otherwise continuous.
Macro funds care less about the tightest spread and more about adverse selection. If a fund buys $1 billion of USD/MXN after a hawkish Fed surprise, it needs liquidity that will not vanish after the first $100 million. This is where algorithmic execution matters. TWAP, VWAP, implementation shortfall and liquidity-seeking algos slice orders across venues, but in FX they must navigate last look, skewed bank streams and variable credit limits. The best algos are not simply fast; they are venue-aware and toxicity-aware.
Carry traders introduce another layer. When global volatility is low and real yield differentials are wide, leveraged investors borrow in low-yielding currencies such as the yen or Swiss franc to buy higher-yielding currencies like the Mexican peso, Brazilian real or Indonesian rupiah. The trade is profitable until volatility rises and funding currencies rally. Microstructure turns a macro unwind into a cascade: stop-losses trigger, dealers reduce risk limits, options hedges force spot buying, and liquidity providers widen spreads simultaneously.
Retail Traders Are Small Individually but Important in Clusters
Retail FX is often dismissed because ticket sizes are tiny compared with interbank flow, but aggregated retail positioning can matter at the margin, particularly in high-beta pairs and during illiquid hours. Retail brokers typically operate under two models. In an agency or straight-through-processing model, client trades are passed to liquidity providers. In a principal or market-making model, the broker internalizes client flow and may hedge net exposure externally.
The economics are clear. Retail traders usually pay through wider spreads, commissions, financing charges and negative slippage during volatility. Leverage magnifies the issue. Regulators in Europe capped leverage for major FX pairs at 30:1 for retail clients under ESMA rules, while some offshore venues still advertise much higher leverage. A 50:1 account can be wiped out by a 2% move, which is routine in GBP/JPY, USD/ZAR or USD/TRY during stress.
Retail order placement can still influence short-term price dynamics because stops cluster around obvious levels: round numbers, prior highs and lows, moving averages and breakout points. Institutional execution desks know this. Algorithms may not "hunt stops" in the conspiratorial sense, but liquidity-seeking behavior naturally gravitates toward areas where resting orders are expected. In thin Asian afternoon trading, a break of 155.00 in USD/JPY or 1.2500 in GBP/USD can attract outsized activity because both retail and systematic strategies react to the same levels.
The rise of social trading and mobile platforms has also compressed reaction times. Retail communities can crowd into dollar-yen carry trades when interest differentials are obvious, but they are usually late to changes in policy reaction functions. The Bank of Japan’s slow exit from yield curve control, for example, was not just a rates story; it altered the microstructure of yen shorts by increasing the cost of assuming official tolerance for one-way depreciation.
Last Look, Prime Brokerage and the Uneven Price of Liquidity
FX pricing is not uniformly firm. In many bilateral streams, banks retain a short window known as last look, during which they can accept or reject a trade request. Defenders argue last look protects market makers from latency arbitrage and stale prices. Critics argue it creates uncertainty and potential information leakage. For clients, the practical question is measurable: what is the fill ratio, rejection rate, hold time and post-trade markout by venue?
Prime brokerage is another key gatekeeper. Many hedge funds access bank liquidity through prime brokers that provide credit intermediation, netting and settlement. After the Swiss National Bank removed the EUR/CHF floor in January 2015, prime brokers tightened standards because client losses exposed the danger of offering credit into discontinuous markets. Smaller funds and high-frequency firms found that technology alone was not enough; balance sheet access became the scarce commodity.
Settlement infrastructure also shapes risk. CLS, the dominant payment-versus-payment settlement system, reduces principal risk in major currencies, but not every EM currency is CLS-eligible. That matters for currencies such as the rupee, rupiah, peso or yuan when global dollar funding tightens. In non-deliverable forward markets, including USD/CNY offshore hedges, USD/INR NDFs and USD/KRW NDFs, pricing reflects not just interest rate differentials but capital controls, fixing methodology and offshore dollar liquidity.
What Traders Should Watch in Real Time
The most useful microstructure signals are not mystical; they are observable if traders know where to look. First, watch spread behavior across sessions. If EUR/USD spreads widen during active London hours without a headline, it may signal balance sheet caution or a large order in progress. Second, compare spot moves with FX swaps and cross-currency basis. A dollar rally accompanied by wider basis is a funding stress signal, not just a growth story.
Third, monitor options markets. Risk reversals show demand for asymmetric protection. When USD/JPY topside risk reversals cheapen while spot rises, it can indicate carry exhaustion rather than bullish conviction. Fourth, respect fixing windows and calendar effects: month-end hedging, Tokyo fix flows, IMM futures positioning releases and central bank decision days can all distort intraday liquidity. Fifth, track EM opening times. USD/MXN behaves differently when Mexico City is open; USD/INR NDF liquidity changes around the onshore RBI reference environment.
The core rule of FX microstructure is simple: price is not just a macro opinion. It is the clearing level between balance sheet, information, regulation, credit and time-specific execution demand.
For professional traders, the actionable implication is to separate trade thesis from execution thesis. A bearish dollar view may be right over three months, but buying EUR/USD into a thin post-payroll spike can still produce poor returns. A yen carry trade may have positive expected carry, but if everyone owns the same position through leveraged accounts, the exit door is smaller than the entry door looked.
Conclusion: The Next FX Edge Is Execution-Aware Macro
Central bank divergence will remain a powerful driver of currencies, especially as the Federal Reserve, European Central Bank, Bank of Japan and emerging-market central banks move through different inflation and easing cycles. But the winners in FX will not be the traders with only the cleanest macro narrative. They will be the traders who understand how that narrative is transmitted through fragmented liquidity.
Banks will continue to internalize more flow, funds will rely on smarter execution algorithms, retail traders will remain vulnerable to leverage and clustering, and EM currencies will keep reflecting the interaction between offshore dollar liquidity and local policy credibility. The foreign exchange market is not one market but a hierarchy of access. Knowing where you sit in that hierarchy is the first step toward improving execution, reducing slippage and surviving the next dollar squeeze.