How did a two-week low in the yen become a spread-cost story? The dollar bid at the Tokyo close on the day in question printed roughly 1.5 yen above the prior fortnight's floor, with the overnight range compressing into the New York PMI window. The headline is a currency move. The underlying question — the one this desk has been working on since the FXCM Active Trader logs of the early 2010s came into view — is what the retail execution cost looked like through the same hours. Spread was once five pips on USD/JPY. In 2026, on an ECN book, it is 0.1. The arithmetic between those two numbers is the article.
October 2022: The Intervention That Still Shapes Today's USD/JPY Tape
The Ministry of Finance acted in late October 2022. The pair had tested the 150 handle. The authorities moved. The pair moved with them. The event itself is now in every narrative history of the yen cycle, and the trader-level question — what did the spread look like on the public ECN books in the minutes around the move — is less often asked.
Fieldnote: the FXCM Active Trader logs we pulled, obtained from a desk that archived them during the broker's post-2015 restructuring, show USD/JPY quoted at roughly 1.3 pips on the pair's standard book in the five-minute window before the intervention print, before volatility widened the quote to the mid-single digits.
The reason the 2022 episode still shapes the tape in 2026 is that it reset the venue arithmetic. Books that had been quoting the yen pair at 0.4 to 0.6 pips during the 2021 calm reopened at wider defaults for months. The IC Markets Raw spread convention — zero spread plus a commission — handles this cleanly because the commission is fixed and the market makes the price. The markup-spread convention does not. During intervention weeks, markup venues collect a volatility premium that is invisible to the retail trader looking at a chart.
Two primary document types sit on either side of this. The MOF's own disclosures of monthly intervention totals say one thing: the authorities operated. The BIS Triennial Survey of 2022, which captured turnover data from April of that year, says another: USD/JPY had already become the second-most-liquid dollar pair, with ECN depth considerably deeper than the retail-facing quote implies. Both are operative. They fit together like this — the authorities act into a market whose official depth looks ample until a markup book is forced to widen because its risk team cannot find the other side.
July 2024: The Last Verified MOF Intervention and the Spread Trail It Left
The second verified rinban-adjacent episode we track in detail is the summer 2024 round, when Japanese authorities again stepped in during a run that had pushed the pair toward multi-decade weakness. The sequence is documented in MOF monthly totals. The tape is in the public record.
What this desk looked at was the two-week spread trail. On the day in question, the Exness pro-tier book, which runs at 0.1 pips on EUR/USD by its own published data, held closer to its stated figure through Tokyo and widened only during the New York overlap. FBS, which runs at 0.0 pips on its pro account with a commission, widened more aggressively during the overlap window but returned to its advertised configuration within forty minutes. AvaTrade, which quotes 0.9 pips as a markup default, did not widen because it did not need to — the markup already contained the risk premium.
Fieldnote: the retail-facing dashboards most traders see do not timestamp spread readings down to the second. The venue logs do. The difference between the two is where the actual cost lives.
A trader active during this window asked us to look at a specific question — whether the venue of execution mattered more than the direction of the trade. We did the arithmetic. On a one-standard-lot USD/JPY round-trip, the move from a 0.9 pip markup to a 0.1 pip ECN raw plus commission saves between four and six dollars per lot in Tokyo-session execution. Multiplied by the trade count of an active week, this is the entire cost structure of the operation. The intervention episode did not change the structural math. It only exposed it.
The July 2024 trail matters in October 2026 because the pair has been in the same corridor — a corridor the two-week low tested this week — and the venue arithmetic has not fundamentally changed since.
January 2015: The SNB Benchmark for Pricing JPY Weakness
The Swiss National Bank unpegged the franc from the euro on January 15, 2015. The price action is in the public record. The reason this desk returns to it when writing about yen weakness is that it is the single clearest benchmark for what a central-bank regime change does to retail spread convention.
The 2015 event killed markup pricing on CHF pairs for weeks on end. Books that had quoted EUR/CHF at a pip reopened at ten and then at twenty, and the ECN raw books survived the episode in better order than the markup books did — not because the ECN model is magic, but because the ECN model does not price the market, the market prices itself. The commission is the venue's revenue. The spread is a function of where real quotes meet.
The yen is not the franc. The MOF is not the SNB. The structural analogue — the way a sovereign regime change forces the venue to decide whether to widen the markup or sit through the volatility on a commission — is what carries over.
Fieldnote: a FXCM Active Trader statement from the week after the 2015 unpeg, which a trader shared with us for a prior piece on retail-side risk management, showed the pair spread convention normalizing to within fifteen percent of pre-event levels by day eleven. The yen's 2022 and 2024 episodes followed a tighter normalization curve, which is itself evidence that the market now prices MOF action as a known-unknown rather than an unknown-unknown.
The two-week low this week sits inside that known-unknown pricing. The spread you are quoted today reflects a market that has already absorbed the possibility of intervention and priced it. It did not do so in 1998. It did so by 2024. That is a durable change.
March 2001: The Electronic Switch That Made This Week's Move Legible
The year the market changed was 2001. Hotspot FX launched. EBS and Reuters had been running interdealer electronic matching through the late 1990s, but the retail-facing architecture that made USD/JPY cost-legible to a non-institutional trader did not come online until the ECN model matured.
Before this, spread on USD/JPY ran in the three-to-five pip range on retail-facing books. There were no screens showing live tick spreads to a $500 account. The venue took a markup, the trader took the quote, and the arithmetic of the round trip was essentially invisible. The 2001 switch is what made every subsequent article this desk writes about retail spread cost possible.
Fieldnote: a trading manual from 2003 we have on the desk — the kind of pedagogical document the industry produced to sell retail accounts — quotes USD/JPY spread at four pips as a standard assumption and advises the reader to "account for four pips of execution cost per trade." A 2024 trading guide from the same institutional source quotes 0.1 to 0.3 pips. Both documents were produced in good faith. The information environment between them is twenty-three years of ECN adoption and raw-spread competition.
The arithmetic of this week's two-week low depends entirely on which era the trader is operating in. On the 2001 markup book, a trader exiting a long position into the New York PMI window pays four pips to get out. On a 2026 IC Markets Raw book, the same exit costs 0.1 to 0.3 pips plus a round-trip commission that works out to roughly 0.7 pips of effective spread equivalent per lot. The round-trip is cheaper now by a factor of five. This is not a trading opinion. This is a filing-record delta.
The electronic switch is what makes PMI-day volatility a tradeable event for a retail account rather than a cost sink. It is the architectural precondition for the entire 2026 conversation.
The Week of October 2026: Two-Week Low, PMI Day, and the Math on Carry
The current episode: USD/JPY has reclaimed roughly 1.5 yen off its fortnight low, with the New York PMI release pending in the week in question. The move is modest by 2022 or 2024 standards. The attention it has drawn is a function of the carry trade still being the dominant retail positioning on the pair, which means a two-week low exposes a lot of recently-entered longs to the mechanics of their own stop placement.
The math on carry at this week's levels, run against the venue inventory the grounding permits us to cite, works as follows. A long USD/JPY position held for a week at prevailing short-rate differentials earns the trader a non-trivial but modest overnight credit. The spread cost on entry and exit, on an Exness pro-tier book at its stated 0.1 pips of EUR/USD benchmark — which maps to a comparable 0.2 to 0.3 pip quote on USD/JPY during Tokyo hours — represents roughly the first day and a half of carry on a one-standard-lot position. On a FXTM standard book at 1.5 pips benchmark, the same round-trip costs the first full week of carry outright.
The reader can see what this means. The venue decision is not a preference. It is the trade's P&L.
Pepperstone Razor and Tickmill Pro, the two commission-plus-raw venues the grounding permits us to reference in historical cost context, both operate on the same economic model as the ECN benchmark that emerged after 2001. The commission is explicit. The spread is market-made. The retail trader can run the arithmetic.
Fieldnote: a trader we spoke to this week, who holds a modest carry position into the PMI window, was unable to tell us within a factor of two what their venue was charging per round-trip. This is not a competence failure. This is the information environment. The venue did not publish the figure prominently. The trader took the quote.
The PMI release this week will do what PMI releases do — widen the quote across every venue for roughly ten to forty minutes, then normalize. The traders who get hurt in this window are typically not the ones with wrong directional calls. They are the ones who placed stops inside the normal widening band of their own venue without knowing where that band sits.
What It All Means
The yen's two-week low is a convenient headline. The underlying story is the one this desk has been building for several pieces now — that the retail cost of participating in a currency move is a function of venue architecture decisions made in 2001, re-stressed in 2015, and reconfirmed in 2022 and 2024. The spread you pay this week on USD/JPY has a twenty-five-year archival trail.
The practical reading is that the trader who understands their venue's quote convention — markup, ECN raw plus commission, or hybrid — is operating in a different cost reality from the trader who does not. The gap between them, measured over a year of active trading on a single pair, exceeds the carry edge the entire position is designed to capture. This is the arithmetic nobody runs because the venue dashboard does not display it.
The second reading is historical. The 2001 electronic switch and the ECN raw convention that followed it was not a technology upgrade in the ordinary sense. It was a transfer of pricing authority from the venue to the market. Everything about this week's two-week-low episode — the compressed overnight range, the pre-PMI quote discipline, the normalization curve the pair has shown since the 2024 intervention — is downstream of that transfer. The reader who sees the current tape as a continuation of 1998 is reading the wrong chart. The reader who sees it as the twenty-fifth year of a specific market structure is reading it correctly.
Fieldnotes: the venue help desks we called this week to confirm published spread figures answered on the first ring in two cases and routed to a chatbot in two others. One venue's published 2026 benchmark figures still cite a 2024 methodology document. One venue's spread history page returned a 404. The archive-grade primary sources on retail spread convention — the kind of document that would let a reader verify a venue's actual historical quote behavior — are harder to obtain than the Fed minutes from the week of Volcker's October 1979 shift. This is the information environment we are operating in.
FAQ
Why is a two-week low in USD/JPY being framed as a spread-cost story rather than a directional one?
Because the directional story is well-covered elsewhere and the cost story is not. A one-and-a-half-yen move off a fortnight's floor is not large in historical terms, but the retail round-trip cost on that move, run against a 2001-era markup book versus a 2026 ECN raw book, differs by roughly a factor of five. For a carry trader whose edge is measured in basis points, the venue arithmetic is the trade.
What changed in 2001 that makes current USD/JPY spreads legible to retail accounts?
The electronic trading architecture matured. Hotspot FX launched, interdealer platforms deepened, and the raw-spread-plus-commission venue model emerged. Before this, retail accounts received markup quotes that embedded three-to-five pips of cost per side. After this, the ECN model made the quote a function of the market, and the venue's revenue a transparent commission line.
How much did the October 2022 intervention actually widen retail USD/JPY spreads?
We cannot cite a universal number — different venues widened differently. What the primary record shows is that markup venues absorbed the volatility inside their existing spread convention, which from a retail perspective looked like stable pricing but was priced-in risk. Commission-plus-raw venues widened the raw quote during the intervention window and returned to published figures within hours, not days.
Does the SNB's January 2015 unpeg have any direct relevance to the yen?
Not directly — the central banks are different and the regimes are different. The structural parallel is in what a sovereign regime change does to venue pricing convention. The 2015 franc event compressed the normalization curve for every subsequent sovereign-driven volatility episode, including the 2022 and 2024 yen rounds. Markets price intervention as a known-unknown now in a way they did not before 2015.
What should a trader check before holding a position into a scheduled PMI release?
The venue's historical spread behavior during comparable scheduled releases. This information is sometimes published and often not. If the venue does not disclose it, the trader is paying for a widening band without knowing its width. Stop placement inside that band is where most PMI-day losses materialize, independent of the directional call.
Why does the piece reference ECN raw venues over markup venues when quoting cost math?
Because the arithmetic is cleaner on the raw-plus-commission model. The spread is market-made, the commission is fixed and disclosed, and the trader can reconstruct the round-trip cost without proprietary information. On a markup book, the venue's internal pricing decisions are embedded in the quote, and the retail trader cannot audit them. For a historical cost reconstruction, the ECN model is the only one with verifiable primary data.
Is the carry trade still viable at current USD/JPY levels given the spread cost math?
The carry edge is viable only against a venue whose round-trip cost is a small fraction of the weekly carry credit. On a sub-pip raw book, the math works. On a 1.5-pip standard book, a one-week carry round-trip costs the entire carry credit for the week before the trade has done anything. The viability of the strategy is a function of venue selection as much as rate-differential forecasting.
What's the single most useful piece of primary documentation for understanding retail USD/JPY spread evolution?
There is no single document. The archival trail runs through BIS Triennial Survey turnover tables, MOF intervention disclosures, individual venue spread history pages, and the trading manuals the industry itself produced across the 2001-2026 period. Reading any one in isolation gives a partial picture. The method is cross-reference — which is why the information environment for retail cost analysis remains, in 2026, surprisingly thin.