04 · Event-Driven Trading
Event-driven trading (Event-Driven Trading): positioning around high-impact events at known times (earnings, NFP, CPI, central bank decisions, elections, geopolitical conflicts). Its biggest difference from other styles: you don't need to predict direction — you prepare before the event happens and execute your plan after — "buy the expectation, sell the fact" is the core of this logic.
But it is also the window where retail traders lose money fastest: the gap at the moment of a data release can erase a month of profit in one second. This article covers how to prepare, how to participate, how to control position size, and the most common ways to die.
1. Event Taxonomy: Which Events Are Worth Trading
| Event type | Typical examples | Instruments affected | Predictability |
|---|---|---|---|
| Earnings season | A-share earnings previews/reports, US quarterly reports | Individual stocks, related sectors | Timing certain, outcome uncertain |
| Macro data | NFP, CPI, GDP, PMI, unemployment | Equity indices, FX, gold, Treasuries | Timing certain, market expectations exist |
| Central bank decisions | Fed/ECB/BoJ rate meetings | Global risk assets, FX rates | Timing certain, path expected |
| Supply-side events | OPEC+ meetings, cut/boost decisions | Crude oil | Timing roughly certain |
| Political events | Elections, referendums, policy releases | The country's assets | Timing certain |
| Geopolitical conflicts | War, sanctions, supply cutoffs | Safe havens (gold), energy, shipping | Timing uncertain, most sudden |
| Crypto project events | Token upgrades, mainnet launches, large airdrops | The specific tokens | Timing roughly certain |
Classification (determines your response):
| Type | Definition | How to handle |
|---|---|---|
| Certain time + uncertain outcome | Earnings, CPI, rate meetings | Position in advance via an event calendar, trade the expectation gap |
| Uncertain time + huge impact | Geopolitical conflicts, black swans | Cannot pre-position; rely on position control and "never ride naked" |
| Certain good/bad news | Forced delisting, unambiguous policy | Usually already priced in — beware sell-the-news |
2. Preparing in Advance: Event Calendar and the Expectation Gap
2.1 The event calendar
Build your own event calendar 1-2 weeks ahead (sources: economic calendar sites/apps, exchange announcements, company investor-relations pages):
## This Week's Event Calendar (example)
| Date | Time (Beijing) | Event | Affected instruments | Market expectation | My plan |
|---|---|---|---|---|---|
| Monday | 21:30 | US Nonfarm Payrolls | Gold/USD/equities | +200k jobs | Trim before data, watch reaction after release |
| Wednesday | 02:00 | Fed rate decision | Risk assets | 80% chance of +25bp hike | No adds, wait for dot plot |
| Friday | 09:30 | Company X earnings | That stock | EPS expected 1.00 | No gambling before earnings, follow trend after |2.2 Expectation and expectation gap: why "buy the expectation, sell the fact" works
Price reflects "expectation", not "fact". When good news is fully expected by the market, it has already been bought; at the release moment, unless the result beats expectations, price has no reason to keep rising — profit-takers cash out instead. That is "buy the expectation, sell the fact".
Numeric example (gold vs NFP):
Event: US Nonfarm Payrolls, market expects +200k jobs
Before: consensus sees strong employment → USD index and Treasury yields rise early,
gold falls from 1980 to 1930 (gold-negative expectation, already priced)
Release: actual +180k, below the 200k expectation
Reaction: expectation gap = actual − expected = −20k (USD negative)
gold spikes from 1930 to 1960 (buying the fact: below expectations = gold positive)Three outcomes of the expectation gap (historical statistics, not predictions; defer to actuals):
| Case | Result vs expectation | Typical move |
|---|---|---|
| In line with expectation | Gap ≈ 0 | Reversal or drift after release — the news was priced |
| Beats/misses materially | Clearly better/worse than expected | A run along the surprise direction (but the first leg often fakes) |
| Double whammy | Data bad itself + worse than expected | Large one-sided move (e.g. CPI beat + prior revised up) |
💡 The core concept of event trading
The market trades the gap versus expectations, not whether the number itself is "good". A "good" print that merely matches expectations carries no new information.
2.3 Where to find expectations (executable checklist)
| Event type | Expectation source | Notes |
|---|---|---|
| Macro data | Economic calendars' "consensus estimate" column, Bloomberg/Reuters surveys | Institutional forecast medians exist weeks ahead; track revisions too |
| Fed rates | CME FedWatch tool (implied probabilities from rate futures) | Hike/cut probabilities are a direct reading of the expectation gap |
| Central bank path | Dot plot (FOMC), officials' speeches | Wording (hawkish/dovish) often matters more than the decision |
| Earnings | Analyst consensus (EPS/revenue median) | Compute the gap as "actual vs consensus" |
| Oil/OPEC+ | Reuters surveys, producer officials' trial balloons | Cut-size expectations revise dynamically with the news |
Usage discipline:
- Track "expectation revisions", not just the final figure. If consensus revises from 250k down to 180k a week before release, an eventual 200k may be "below the original estimate" but "above the latest expectation" — completely opposite directions;
- The more unanimous the expectation, the more dangerous. With the whole market aligned (e.g. 100% hike odds), any deviation produces a big move;
- Decide on the difference between "my view vs market expectation", not on "good or bad data" — your personal opinion is worthless against market pricing.
3. Earnings Trading
3.1 A-share earnings previews / US quarterly reports
| Scenario | Description | Typical play |
|---|---|---|
| A-share earnings preview | Fixed schedule; results quality knowable weeks early | Preview games usually complete before the announcement; gap-up-then-fade on preview day is common (sell-the-news) |
| US quarterly reports | Released pre/post-market; first trading day often gaps 5%-15% (historical statistics; defer to actuals) | Gambling before vs chasing after earnings — see below |
3.2 Gambling direction before vs chasing trend after: win-rate common sense
| Approach | Win-rate common sense (historical statistics) | P&L profile | Main risk |
|---|---|---|---|
| Gamble direction before earnings | Near coin flip (~50%) | Big gap profits when right | Equally big gap losses when wrong; implied volatility expensive (options carry high Vega) |
| Chase trend after earnings | Slightly higher (55%-60%, historical statistics) | Clear information, follow through | The gap has eaten most of the profit, risk-reward worsens |
| Fade after earnings (bet sell-the-news) | Depends on "whether expectations were stretched" | More effective the fuller the expectation | Judging "fullness" is hard; easy to catch falling knives |
Practical advice:
- Don't gamble direction before earnings. Coin-flip odds + rich implied volatility makes long-run negative expectation near-certain (especially buying options: post-earnings IV crush can lose money even when direction is right — historical statistics; defer to actuals).
- After earnings, only trade "pullbacks after breakout" or volume-confirmed trends — never chase the first jump. Wait for direction to clarify 15-30 minutes after the open and for confirmation signals.
- Cap per-trade size (see Part 5); gaps can jump straight over your stop-loss level — use "stop-loss level + gap contingency" as double protection.
4. Macro Events: Volatility Patterns Around Releases
4.1 The 30 minutes before release: the calm before the storm
| Trait | Behavior | Response |
|---|---|---|
| Volume shrinks | Big players watching, volume well below normal | Don't open new positions in this window |
| Narrow oscillation | Price bounces in a tiny range | Existing positions face "two-sided stop hunts" |
| Stops swept | Institutions hunt stops during thin liquidity | Move stops away / trim before release |
4.2 The release moment (first 1-5 minutes): the first wave often fakes
- Price lunges one way instantly, then often gives back sharply or reverses within minutes — the first wave is liquidity vacuum plus programmatic front-running; its direction isn't necessarily sustainable (historical statistics; defer to actuals).
- Response: don't chase the first wave. Observe 5-15 minutes after release, wait for the second wave's direction (in most cases only the second wave reflects the real expectation gap).
4.3 The 30 minutes after release: direction clarification window
| Period | Typical pattern (historical statistics) | Trading window |
|---|---|---|
| 0-5 min after | Instant gap + first lunge | Chasing forbidden |
| 5-15 min after | Giveback/reverse probe, then re-selecting direction | Watch and confirm |
| 15-60 min after | Trend unfolds along the true direction | Can enter with the trend; widen stops by ATR |
⚠️ The first wave at release often fakes — don't chase it
The first wave right at data release is often a fake-out. Price lunges one way instantly, then often gives back sharply or reverses within minutes — the first wave is liquidity vacuum plus programmatic front-running, and its direction isn't necessarily sustainable. Watch 5-15 minutes after release and wait for the second wave's direction before acting.
Volatility traits of major events (historical statistics, not predictions; defer to actuals):
| Event | Typically affected instruments | Volatility profile |
|---|---|---|
| NFP | Gold, USD, equity indices | Extreme instant volatility; "spike then fade" common |
| CPI | Treasuries, USD, equities, crypto | Surprise = broad risk-off, effects lasting days |
| Fed decision | Everything | Two phases: decision instant + press conference (wording matters more than the rate) |
| OPEC+ meeting | Crude oil | Cut/boost magnitude versus expectations sets direction |
5. Position Discipline for Event Trading: The Half-Position Rule
Event moves are 2-3× normal volatility (historical statistics, not predictions; defer to actuals) — the same position doubles its risk in event markets. The answer is one rule: the half-position rule.
| Case | Normal position | Event-market position |
|---|---|---|
| Per-trade risk | ≤ 1% of capital | ≤ 0.5% |
| Daily risk | ≤ 2%-3% of capital | ≤ 1%-1.5% |
| Leverage | Normal | Halved or none |
| Positions held into data | Hold normally | Trim or hedge before data (reduce directional exposure) |
Additional discipline:
- Stops must be placed in advance — there is no time to act manually at the gap instant; resting orders are the only execution guarantee;
- No new positions in the 30 minutes before release (leave that knife fight to quant firms and market makers);
- One event, one direction — no "clever cross-instrument hedging" — event reactions routinely exceed your model;
- Two losing events in a row → stop and review. Event samples are small; consecutive losses mean your expectation model is wrong, not that luck is bad.
💀 The half-position rule
Event moves are 2-3× normal volatility — the same position doubles its risk. One answer: the half-position rule — per-trade risk from ≤ 1% of capital down to ≤ 0.5%, daily risk from ≤ 2%-3% down to ≤ 1%-1.5%, and leverage halved or dropped. Place stops in advance; open nothing new in the 30 minutes before release.
6. Ways Event Traders Die: Checklist
| Death | Script | Antidote |
|---|---|---|
| Front-running | Heavy bet 30 minutes pre-data, swept out by the fake first wave | No adds/new positions before data; wait 15 minutes after release |
| Chasing the first wave | Jumping into the gap at release, catching the exact reversal point | Wait for second-wave confirmation; enter 5-15 minutes later |
| Data double whammy | Beat expectations + prior revised up, losses amplified | Half-position rule — even a double whammy finds "no position to kill" |
| Sell-the-news distribution | News merely matches expectations, gap-up fades and traps you | Remember: in line = no new information; don't chase highs on release day |
| Gambling pre-earnings | Buying options/heavy before earnings; IV crush + wrong direction = double loss | Don't gamble pre-earnings (Part 3); follow confirmed signals after |
| No event calendar | Forgot CPI is tonight; gap blows through your stop | Build a weekly event calendar; check positions and stops before events |
| Riding naked into geopolitics | Fully invested, unhedged; one sudden move resets everything | Keep a cash buffer always; black swans can't be predicted but margin can be reserved |
| Over-modeling events | Believing "hikes must fall, cuts must rally" | Markets trade the expectation gap: a fully-priced hike rising instead is common |
7. Full Event-Trading Workflow (Action Checklist)
[1 week before] Build/update the event calendar; note affected instruments and market expectations
[1-3 days before] Fix your plan:
① What is the expectation? (calendar consensus value + recent revision direction)
② For each scenario — beat/in-line/miss — what exactly will I do?
③ Position: compute amounts under the half-position rule
[1 hour before] Check:
□ Stops placed for existing holdings? Away from pre-data sweep zones?
□ Any new-entry plan scheduled "15+ minutes after release"?
[Release instant] Do nothing; observe the first wave
[15-60 min after] Execute per plan:
Confirm second-wave direction → enter with trend at half position → widen stop by ATR
[After] Review: directional call, entry timing, position vs plan8. Quick Reference: Common Event-Trading Fallacies
| Fallacy | Reality |
|---|---|
| Bad data = must fall | Direction depends on "data vs expectation", layered with liquidity behavior at release; no guaranteed move |
| Positioning early beats chasing late | Early positions must survive pre-data volatility and stop sweeps; late entries give up some spread but get clearer signals |
| Hikes fall, cuts rally | A priced-in hike can flip to "bad news exhausted = good"; what's traded is the gap, not the event itself |
| Everyone makes money in earnings season | Earnings volatility is zero-sum against institutions and algorithms; retail's edge is discipline, not information |
| Event trading requires insider info | Public sources (calendars, consensus, probability tools) suffice; tips are mostly noise |
| One big event defines the year | Never double down after a single-event loss (stop after two straight); event trading is not a core strategy |
| Rush in as soon as data drops | First-wave fake-outs are the norm; wait 5-15 minutes and confirm the second wave |
⚠️ Risk Warning
Event markets are the most violent and least controllable by gaps of all regimes; history is full of traders who called direction correctly yet were badly hurt by oversized positions or missing stops. All win rates, volatility magnitudes, and timing windows here are historical statistics, not predictions; defer to actual market conditions; treat event data and consensus values as subject to official release channels and market consensus. Participate only with money you can afford to lose — leveraged trades in event markets can wipe principal in an instant or produce a negative balance.