"History doesn't repeat… but it rhymes." — Mark Twain
◉ THE PRESENT
The yen moved seven percent against the dollar in three trading sessions. That is not a drift. That is a stampede.
The Bank of Japan's slow, almost apologetic tightening — dismissed for a year as too timid to matter — finally cracked the wall holding back the largest leveraged trade in modern finance, and when the yen carry started to unwind it took everything with it: Nasdaq megacaps, Mexican peso bonds, copper, bitcoin, high-yield credit. The VIX doubled in forty-eight hours. Somewhere in Greenwich and Mayfair, a lot of funds are on emergency calls with their prime brokers.
USD/JPY: 148.2 → 137.6 | VIX: 16 → 34 | HY OAS: +52 bp to 412
We have seen this movie before. October 1998. Same currency. Same leverage. Different decade.
◉ THE ECHO — OCTOBER 7, 1998
Three phones rang at the same time on the trading floor in Greenwich, Connecticut.
It was a Wednesday. John Meriwether was in his office at 600 Steamboat Road in Greenwich when the yen moved six figures in the time it took him to pour a cup of coffee. Long-Term Capital Management had a $1.25 trillion notional book, sixteen Nobel-adjacent PhDs on payroll, and leverage north of twenty-five to one. It had been returning twenty percent a year since 1994 and had never once posted a losing month. Until August.
Russia defaulted on August 17th. That should have been a rounding error. The entire Russian bond market was smaller than a mid-cap American bank. But when you lever a rounding error twenty-five times, it stops being a rounding error. Credit spreads started drifting the wrong way. Then swap spreads. Then swaption vols. Every uncorrelated trade LTCM held began moving in the same direction at the same time. The quants call that correlation going to one. Traders on the floor call it the building being on fire.
On the morning of October 6th the yen traded at 136. By the close on October 8th it was at 111. That is an eighteen percent currency revaluation in forty-eight hours for the third largest economy in the G7. The yen carry trade — borrow cheap in Tokyo, park the proceeds in anything yielding more, which in 1998 meant nearly everything — unwound all at once. Every leveraged desk that had funded itself in yen had to buy yen back. At the same time. With no sellers.
Bill McDonough ran the New York Fed then. On September 23rd, two weeks before the yen went vertical, he had called the fourteen CEOs of Wall Street's major banks into the boardroom on the tenth floor of 33 Liberty Street. The room was quiet in a way rooms with fourteen bank CEOs usually are not. McDonough explained that LTCM's book was large enough that if it was liquidated over a weekend, a handful of them in the room might not open for business Monday. They pooled $3.625 billion to take the fund over. Alan Greenspan cut rates on September 29th. Then again on October 15th. Then on November 17th. Three cuts in seven weeks by a Fed that did not move fast.
By December the panic was over. By March of 1999 the S&P was at an all-time high. The single currency that had ignited the whole thing, the one nobody wanted to talk about on CNBC, had quietly settled back near 115 and stayed there. Now it is back. And it is doing exactly what it did then.
◉ THE RHYME — WHAT'S IDENTICAL
The yen is once again the match. Leverage is the gasoline. What differs is only the size of the building on fire.
◉ THE DIVERGENCE — WHAT'S DIFFERENT THIS TIME
The pattern is close. It is not identical. Four things separate 2026 from 1998, and they all matter.
The Bank of Japan now owns roughly fifty-five percent of the government bond float, up from ten percent in 1998. That gives it tools the 1998 BOJ simply did not have — it can cap yields at the long end without a press conference by shifting its own purchase schedule. Whether it uses those tools fast enough is the open question.
There is no LTCM this time. The carry trade is spread across thousands of leveraged accounts, Japanese retail margin books, and corporate treasury desks. That is better for systemic risk, because no single failure takes the system down. It is worse for optics, because there is nobody for the Fed to call into a boardroom and fix.
The Fed is already in a cutting cycle. Markets priced most of it months ago. That leaves far less ammunition in the surprise barrel if the unwind gets worse — three emergency cuts in seven weeks, 1998-style, is not available when you started the quarter with a three-handle policy rate.
Stablecoin rails and around-the-clock crypto markets now transmit margin stress on weekends. In 1998 the weekend was a firebreak. In 2026 it is an accelerant. The first real test of that fact could arrive before the next Tokyo open.
◉ THE RECKONING — WHAT HAPPENS NEXT
After the LTCM rescue, the S&P bottomed on October 8, 1998, nineteen percent below its July high. Then it did something nobody in the room on Liberty Street would have predicted. By year end it closed up twenty-six percent for 1998. By March of 1999 it was at an all-time high. The whole panic, start to finish, lasted about eleven weeks.
But the real trade was not in the S&P. The real trade was in credit. High-yield bonds widened to 580 basis points over Treasuries at the worst of the October panic. Eighteen months later they were inside 300. The people who bought junk debt at the widest spreads doubled their money before LTCM's partners had finished winding down the fund. The people who bought emerging market dollar bonds did nearly as well. Japanese equities, hated and heavy, rallied forty percent off the lows once the yen stabilized.
The smart money did not try to call the S&P bottom. It bought what the forced sellers were dumping. That is the lesson that shows up in every leverage panic since 1929. The forced seller is a gift. The patient buyer on the other side of him is the one who looks like a genius on the next cycle.
The trigger for the end of the 1998 panic was not a Fed cut. It was the yen settling. Once USD/JPY stopped moving five percent a day, the carry players stopped getting stopped out, the forced selling stopped, credit spreads peaked, and equities had the floor under them. The tell was in currencies. Stocks came last.
Forced deleveraging feels like the end of the world for six to eight weeks, then the rebound is violent. Watch credit, not the S&P. When high-yield spreads stop widening, the unwind is ending. That was the signal the 1998 tape gave, and it is the tape we are watching now.
◉ TOMORROW’S WATCH
Keep one eye on the Swiss franc. It is the other quiet carry funding currency, and the last time the SNB lost control of a carry dynamic was January 15, 2015. That one took less than twelve minutes.
