What happened
The event is dated 3 January 2019 in Japan and Australia. In UTC, the decisive move happened late on 2 January. This difference is a time-zone conversion, not a second crash. All times below follow the Reserve Bank of Australia (RBA); the conversion to Tokyo and UTC is explicit.
Apple cut its quarterly revenue forecast earlier that evening and global risk sentiment weakened. The RBA reports orderly movements initially: the yen gained modestly, some higher-yielding currencies weakened, and bid–ask spreads were little changed. The sudden break came later, without a new material announcement at that instant.
Within about 30 seconds at 09:36 Australian Eastern Daylight Time, the yen appreciated roughly 3% against the US dollar. Approximately 30 seconds later the Australian dollar started to fall, losing about 3% against the dollar within roughly two minutes and briefly touching US$0.6715. Against the yen, the Australian dollar was around 7% lower. Prices retraced around half of the move within minutes.
One event, three clocks
Times in UTC unless marked otherwise. Approximate values retain the RBA’s precision.
The event minute by minute
- 01
Risk sentiment weakens
Apple’s guidance cut was already public. Yen strength and weakness in some risk currencies began in an orderly way.
- 02
Liquidity deteriorates
Currency markets became less orderly while US and Japanese cash and futures markets were shut.
- 03
The yen surges
The RBA reports roughly 3% appreciation versus USD over about 30 seconds; spreads jumped from roughly 2 to 100–300 pips.
- 04
AUD follows
AUD/USD then lost about 3% in roughly two minutes. AUD/JPY fell about 7% and AUD/USD briefly hit 0.6715.
- 05
Order returns
Prices recovered about half of the initial shock within minutes; spreads returned to normal within roughly an hour.
Observed short-window moves
These figures are observations reported by the RBA, not independently reconstructed tick quotes. The currencies and denominators differ, so the bars must not be added together.
Read each bar separately. A lower AUD/JPY means one Australian dollar bought fewer yen; a stronger yen does not mean every yen cross moved by the same percentage.
Why the move accelerated
Background news is not the instant trigger
Apple’s warning explains why traders were cautious. The RBA found no material news in the 30-second interval of the sharp yen move. Timing alone does not prove that one corporate announcement caused that specific print.
A thin trading window
The event struck between the US close and Tokyo opening, shortly after New Year. Japanese banks were closed for a holiday. Fewer market makers meant that a given imbalance could move quotes further than it would in a deep session.
Forced carry-trade reductions
Public positioning showed Japanese retail investors were generally long higher-yielding currencies against yen. Stop-loss and margin liquidations may have generated more yen buying as the yen strengthened. The RBA treats this as a likely contributor, not a confirmed single culprit.
Automated liquidity can retreat
Some market-making algorithms can stop quoting when conditions become abnormal, while momentum or hedging orders can intensify demand for yen. The RBA says these behaviours may have amplified the move, but fragmented trading data rule out a complete causal reconstruction.
Why three exchange rates moved together
Currency crosses are linked: in ordinary quote notation, AUD/JPY equals AUD/USD multiplied by USD/JPY. If AUD/USD falls while USD/JPY also falls, both changes push AUD/JPY lower. That is why a smaller move in each dollar pair can coincide with a much larger move in the Australian dollar against the yen. It does not require a separate, simultaneous announcement about the Australian economy.
The reported numbers cannot be multiplied to reproduce the exact 7% cross move. The RBA describes approximate extrema over slightly different seconds and venues, while executable bid and ask spreads were changing quickly. Multiplying two rounded 3% changes would suggest about 5.9%, but the reported cross decline was around 7%. This difference is a reason to preserve the source’s precision and timing, not to “correct” its historical observation.
A carry trade financed in yen typically sells yen to buy a higher-yielding currency. When that currency weakens and yen strengthens, the position loses value in yen terms. Automatic loss limits can then require the trader to buy yen back. If many accounts face the same constraint at the same time, their exits push in the same direction as the original move. This is a plausible feedback loop; the RBA does not claim to have measured every forced order.
During the brief disruption, the quote was a moving target. An order could have been triggered by one venue’s displayed price and filled against available liquidity at another price. In that setting, comparing a retail account statement with a later one-minute chart is insufficient to prove an execution error. The useful evidence is the exact order type, trigger side, server timestamp, available bid/ask, broker venue and fill record.
This episode was severe but short-lived. The RBA describes a market that regained orderly pricing within minutes and spread conditions that normalised within about an hour, with no systemic spillover. That does not erase losses experienced by particular leveraged traders. A temporary market-wide recovery and an individual account’s realised outcome answer different questions.
Where did liquidity go?
The RBA estimates the USD/JPY bid–ask spread near 2 pips before the break and 100–300 pips during the brief disruption. In a yen pair, one pip commonly equals ¥0.01: the cited range is about ¥1–¥3 between bid and ask. This is an indicative market spread, not a guaranteed price offered by any particular broker.
A stop order becomes a market instruction when triggered; it does not promise a fill at the stop level. A screenshot of a chart normally shows only one price stream and cannot reconstruct the executable bid, ask, available depth and fill order at a particular venue.
The relative cost of being wrong can therefore grow far faster than the visible candle. Margin calls and forced reductions can meet an order book with little depth, while different brokers and platforms report different extremes.
The market recovered; the earlier spread was not a permanent new normal.
The thin yellow mark represents roughly 2 pips. The coloured band spans 100–300 pips on a scale ending at 300; it shows a published range, not one guaranteed executable quote.
Illustrative spread calculation — not a historical execution
Imagine a USD/JPY trade of 10,000 USD at an illustrative price of ¥106 per dollar. A 2-pip spread is ¥200, about $1.89 at that rate; a 100–300-pip spread is ¥10,000–¥30,000, about $94–$283. This arithmetic shows the order of magnitude only. It excludes commission, slippage beyond the spread, changing prices and broker-specific execution.
Observed facts versus likely mechanisms
The RBA could not definitively identify a unique cause because the FX market is fragmented. It lists likely interactions between position liquidation, low seasonal liquidity and algorithmic behaviour. Deutsche Bank analysts likewise linked the pattern of yen crosses to Japanese retail positions; that assessment is a market interpretation, not an official adjudication.
The RBA charts use market information available to the central bank. This article does not present a tick-by-tick replay or pretend that every broker offered the same bid and ask. The visual comparison below shows only the percentages and spread ranges explicitly published by the RBA.
The sharp recovery is part of the story: the RBA reported roughly half the price movement retraced quickly and spreads back to normal within about an hour. A dramatic low can still be a real traded price for some participants, but it should not be mistaken for a stable market-wide executable quote.
The evidence supports a rapid yen move, contagion into AUD and a temporary collapse in executable liquidity. It supports several plausible amplifiers, but no single proven trigger or repeatable trading signal.
Practical checks
- 01Check exposure to the same funding currency across every open position, not merely each pair separately.
- 02Treat holidays and session handovers as a change in available depth; a 24-hour quote does not mean 24-hour liquidity.
- 03Size positions for gaps and variable spreads. Test what happens if the exit price is worse than the stop.
- 04Separate a displayed mid-price from executable bid/ask and keep broker execution records after an unusual fill.
- 05Use the official time zone when comparing news, charts and statements; the event crosses the UTC calendar boundary.
Questions about the 2019 yen flash event
Did Apple cause the 30-second move?
Apple’s warning preceded the event and likely worsened risk sentiment. The RBA found no new material announcement at the exact break and did not establish a single cause. The responsible conclusion is that the news was part of the setting, with thin liquidity and order flow shaping the sudden move.
Why do some sources call it 2 January?
The key 09:36 AEDT observation converts to 22:36 UTC on 2 January, 07:36 in Tokyo and 00:36 in Kyiv on 3 January. The timestamps describe the same market event.
Would a stop loss have protected the account?
A stop can trigger an exit, but the final fill depends on quotes and depth. In a fast market with a 100–300-pip spread, it may execute materially away from the chosen level. Position sizing, leverage and broker protections therefore matter alongside the stop.
Was there proof of a single bad algorithm?
No. The RBA describes possible amplification by automated market makers withdrawing and by momentum or hedging orders. Fragmented FX data prevented a definitive allocation of responsibility to one program, platform or participant.
Primary sources
All observed percentages, timestamps and spread ranges come from the RBA’s February 2019 analysis. Apple’s 2 January letter establishes the timing of the guidance revision. Deutsche Bank’s contemporary analyst note is used only to describe one interpretation of Japanese retail positioning. The charts below are our own visual summaries of published figures, not a continuous price history.
How a stop-loss executes: trigger price, slippage and quotation gaps