No trading session is universally the most profitable. For many liquid FX pairs, the London–New York overlap is a sensible first window to test because two major dealing centres are active. For US index futures and stocks, activity often concentrates around the US cash-market open and close. For crypto, “Asia,” “London,” and “New York” are analysis buckets rather than market openings. Higher activity can improve execution, but it does not create a trading edge.
The practical answer is therefore conditional: choose the window that matches your instrument and setup, then compare net expectancy by entry time using the same strategy, risk unit, cost treatment, and market regime. If the result does not survive a later sample, you have a hypothesis—not a session edge.
London vs New York vs Asia: Which Trading Session Wins?
For a first test, not a universal verdict: start with the London–New York overlap for EUR/USD and other heavily traded USD majors; start with the US cash open for ES or NQ; use the relevant exchange’s core session for individual US stocks; and define crypto windows from your venue’s own timestamps. Then let your net results decide.
| Window | What is actually active | Reason to test it | What it does not prove |
|---|---|---|---|
| Asia | Asia-Pacific dealers and local cash markets; crypto remains open | JPY- and AUD-linked FX, Asia-sensitive futures, range or lower-tempo setups | That every pair, contract, or range strategy has positive expectancy |
| London | The UK and European business day | EUR/GBP-linked FX, European catalysts, breakout or trend setups | That “London” is one stable volatility regime every day |
| London–New York overlap | European and North American dealing hours overlap | USD majors and event-driven FX when participation can be broad | That more turnover produces higher trader profit |
| New York / US core | US data, cash equities, and the North American business day | US stocks, ES/NQ and USD-sensitive markets | That the whole afternoon has the same liquidity or edge as the open |
The labels above are useful buckets, not universal exchange rules. Wholesale FX has no single venue or official session clock: the BIS describes a market spread across many venues and centres that moves from Asia to London and then New York. Your broker may use different boundaries, server time, or daylight-saving logic.
What the Market Data Can—and Cannot—Tell You
Forex: location share is not overlap volume
The 2025 BIS Triennial Survey reports that UK sales desks handled about 38% of global FX turnover in April 2025. The Bank of England’s UK release gives the more precise 37.8% figure and explains that the measure is based on sales-desk location.
That supports London’s importance as an FX centre. It does not show that 38% of trading occurs during “the London session,” and it cannot be added to the US location share to claim that a four-hour overlap contains a particular percentage of daily liquidity. The earlier version made that category error.
The Bank of England’s April 2025 London FX survey found EUR/USD was the most traded pair in its London sample, representing 25% of reported turnover. That makes EUR/USD a defensible instrument to test during London hours; it still says nothing about a retail strategy’s profitability.
US index futures: open nearly around the clock, active unevenly
CME’s current Micro E-mini contract FAQ lists MES, MNQ, MYM, and M2K trading on CME Globex from Sunday evening through Friday afternoon, with a daily halt. Availability is not the same as equal liquidity at every hour.
In a 2025 CME study of E-mini S&P 500 trading, volume followed a U-shaped curve: heavier around the 9:30 a.m. ET cash open and 4:00 p.m. ET close, lighter in the middle of the US day. The study is evidence about ES activity in its sample—not proof that the first or last hour is more profitable, and not a direct result for NQ, CL, or GC.
US stocks: distinguish core from extended hours
As checked September 7, 2026, the NYSE core session remains 9:30 a.m.–4:00 p.m. ET. Some venues and brokers offer earlier, later, or overnight access. The SEC’s extended-hours risk guide warns that lower liquidity, wider spreads, fragmented quotes, and higher volatility can affect execution outside regular hours. Check the exact venue and broker route instead of treating “New York session” as one uniform market.
Crypto: no opening bell, but time still matters
Crypto venues operate continuously; Coinbase’s current market-hours guide describes crypto as available 24/7. A crypto “London session” is therefore a timestamp segment, not an exchange opening. Compare venue-specific volume, spread, funding, slippage, and your own results rather than importing an FX session ranking.
Use Local Market Time, Not a Frozen GMT Table
Hard-coded GMT windows drift when the UK and US change clocks on different dates. In 2026, US daylight saving began March 8 and ended November 1, while UK clocks changed March 29 and October 25. During the gaps, a London–New York overlap moves by one hour in UTC.
Store the original timestamp in UTC, but derive labels with named time zones such as Europe/London and America/New_York. Keep the conversion rule with the analysis. The UK government clock-change page and NIST’s US daylight-saving rules document the different calendars.
EUR/USD: London Session or New York Session?
If you can trade only a few hours, test one repeatable block rather than chasing both sessions.
- Choose London morning first when your setup is built around the European open, EUR/GBP catalysts, or early-session range expansion.
- Choose the London–New York overlap first when your setup needs broad USD participation or scheduled US data, and you can execute without rushing.
- Choose New York morning outside the overlap only when your setup is explicitly tied to US events or the US cash open.
- Do not choose by average range alone. Record the spread at entry, slippage, stop distance, and result in R so a faster window is not mistaken for a better one.
A clean test uses the same EUR/USD setup, risk rule, and weekday mix in each block. If London gets breakout trades while New York gets mean-reversion trades, the comparison is about strategy selection as much as time.
Which Futures Trade During Asian and London Sessions?
“Trades during” is safer than “trades better.” CME Globex makes many contracts available across Asian and European hours, but the useful contract depends on which underlying market and catalyst are active.
| Window | Contracts worth measuring | Why they may be relevant | Required check |
|---|---|---|---|
| Asian hours | Japanese yen, Australian dollar, gold; MES/MNQ when global news is moving US risk | Local currency participation and Asia-Pacific macro events can concentrate attention | Your contract month’s depth, spread, and volume at the exact hour |
| London hours | Euro FX, British pound, gold, US index futures | European data and the opening of European cash markets can move linked products | Whether your setup survives before the US cash market opens |
| US cash hours | ES/MES, NQ/MNQ, YM/MYM, RTY/M2K | Their underlying US equity markets and cash auctions are active | Open, midday, and close separately; do not pool the entire session |
For NQ or MNQ specifically, London-hours trading is real, but the Nasdaq-100 cash constituents are not in their core session yet. A strategy that responds to European or global macro news may still find opportunity; a setup that relies on the US opening auction should be tested around the US cash open instead.
Why the Busiest Session Can Still Be Your Worst
Liquidity, volatility, and profitability are different variables.
- Liquidity affects execution. More resting interest can tighten spreads and reduce market impact, but only for the instrument and size actually traded.
- Volatility affects opportunity and error size. Larger moves may help a breakout system while increasing false starts or stop distance for another setup.
- Strategy selection changes by hour. If you take different setups in different windows, the apparent session result may be a setup result.
- Scheduled events cluster in time. A small number of announcement trades can dominate a session’s average.
- Execution quality is personal. Sleep, work schedule, decision speed, and prior losses can change your result even when market conditions are favorable.
This is why the article no longer calls NY afternoon a universal “value destroyer” or Asia universally low edge. Those may be testable hypotheses for a defined instrument and setup; they are not market-wide facts.
How to Measure Your Own Session Performance
- Freeze the question. Example: “Does setup A on EUR/USD have higher net expectancy in London morning or the overlap?” Do not start with “Which session is best?”
- Normalize timestamps. Store UTC, derive market-local labels with daylight-saving-aware time zones, and document the rule.
- Keep comparisons like-for-like. Same instrument, setup definition, risk model, account, and preferably similar volatility regimes.
- Use net results. Include commission, spread, fees, and slippage. Gross P&L can make a high-turnover window look better than it is.
- Measure in R as well as account currency. R normalizes trades with different stop sizes and makes windows more comparable.
- Inspect distribution, not just win rate. Track average and median R, average win and loss, trade count, drawdown, and the influence of the largest winner or loser.
- Separate event days. Mark scheduled macro releases, earnings, roll days, and exceptional volatility before attributing the outcome to a session.
- Validate later. Choose a rule on one sample, then test it on subsequent trades without changing the definition.
| Metric | Why it matters | Failure mode |
|---|---|---|
| Net expectancy per trade | Combines win rate and payoff after costs | One outlier can dominate a small sample |
| Median R | Shows the typical trade with less outlier influence | Can hide a strategy whose edge depends on rare large wins |
| Trade count | Makes uncertainty visible | A large count still fails if strategy rules changed |
| Average win / average loss | Explains whether payoff, not hit rate, drives the result | Open risk and partial exits may be classified inconsistently |
| Max drawdown and loss tail | Shows the risk cost of the window | Historical drawdown is not a future maximum |
| Costs and slippage | Tests whether apparent edge survives execution | Missing spread or fee data overstates active-session results |
How many trades per session are enough?
There is no universal count that turns a noisy result into truth. Thirty trades can be a useful first screen, but it is not a reliability certificate. Required sample size depends on payoff dispersion, win probability, setup consistency, and how small a difference you are trying to detect. Show the count, confidence interval or bootstrap range, and result with the largest winner and loser removed. Then require a later holdout sample before changing meaningful risk.
If one bucket has 120 trades and another has 18, do not rank the headline averages as if confidence were equal. Keep collecting the smaller bucket or narrow the question to the window you can actually trade.
Worked Example Without a Fake Population Average
Suppose one trader reviews a single breakout setup on one instrument. These numbers are hypothetical and illustrate the method only.
| Bucket | Trades | Net expectancy | Median | Read |
|---|---|---|---|---|
| London morning | 64 | +0.18R | +0.05R | Candidate edge; check outliers and later sample |
| Overlap | 61 | +0.10R | −0.02R | Positive mean may depend on larger winners |
| New York afternoon | 19 | −0.08R | −0.04R | Too little evidence for a permanent ban |
The correct conclusion is not “London always wins.” It is: for this trader, setup, instrument, cost model, and sample, London morning is the strongest candidate. The next step is a predeclared holdout test—not immediately reallocating most of the risk budget.
What to Do After You Find a Session Pattern
- If one window looks stronger: keep the setup unchanged and test a modest session filter on the next sample. Do not jump straight to a 70–80% risk allocation.
- If one window looks weaker: review setup mix, costs, news days, and fatigue before declaring the clock causal.
- If all windows are negative: return to the edge measurement framework. Timing cannot rescue a setup with negative net expectancy everywhere.
- If only the first block you trade is positive: test screen time, prior P&L, and execution quality as alternative explanations.
- If you can trade only one window: optimize the setups and risk controls available there; a theoretically better session you cannot execute consistently is irrelevant.
Review the result alongside win rate versus risk/reward, trade quality versus P&L, and the broader performance-analysis workflow.
Run the Session Check on Your Own Trades
Trader’s Second Brain is our product. Its dashboard includes performance-by-session and hourly-performance views, while session settings let you normalize labels. Use those views to find a candidate window in your imported or recorded history, then apply the sample and holdout checks above. TSB does not know a universally profitable session and should not turn a small bucket into a trading recommendation.
If a spreadsheet already contains clean timestamps, setup labels, costs, and R results, it can answer the same question. If import coverage and repeatable filtering would remove manual work, verify your exact source in the live TSB import directory.
Session Analysis Mistakes to Avoid
Calling volume “profitability”
Exchange or survey volume describes participation. It does not include your setup rules, costs, timing, or mistakes.
Using fixed UTC labels through daylight saving
A hard-coded overlap can silently compare different local-market hours across the year. Recalculate with named time zones.
Pooling different instruments and setups
An Asia bucket full of AUD/JPY mean reversion and an overlap bucket full of EUR/USD breakouts cannot isolate a session effect.
Optimizing and validating on the same trades
The best historical bucket will often look better partly by chance. Freeze the rule and test later trades before relying on it.
Ignoring costs or the largest trade
Active windows may increase both opportunity and turnover. Show net results and rerun the comparison without the largest winner and loser.
Methodology and Source Boundaries
- Fact check date: September 7, 2026.
- Market structure: BIS 2025 global FX turnover; Bank of England 2025 UK and London FX surveys; current CME Micro E-mini hours and a 2025 CME ES liquidity study; current NYSE hours; SEC extended-hours risk guidance; current Coinbase crypto-hours guidance.
- Timezone handling: official UK and US daylight-saving calendars; session labels are analysis conventions, not a universal OTC FX rule.
- Profitability: no market-wide retail session ranking is asserted. The prior unattributed TSB cohort figures are withdrawn.
- Scope: examples are educational and hypothetical. They do not predict returns or recommend a particular instrument, session, or risk allocation.
No firm or funding program determines this educational decision, so a firm/program catalog card would be artificial here. If a prop rule later constrains a trader’s permitted hours, that exact program belongs in a program-specific guide with a verified date.
Final Verdict: Match the Window to the Trade
The London–New York overlap is the best general first test for liquid USD-major FX, not a universal profit winner. London morning may fit EUR/GBP catalysts and early breakouts; the US cash open is the defensible starting point for ES/NQ and individual US stocks; Asia may fit local currencies, Asia-sensitive contracts, or slower setups; crypto requires venue- and strategy-specific time buckets.
Your best session is the one with repeatable net expectancy for the same instrument and setup, after costs, across enough comparable trades and a later holdout sample. Market hours narrow the search. Your own evidence makes the decision.
