Odd Lots: Hundreds of Independent Bets, One Order to Leave

Notes on the Odd Lots episode published 11 November 2026 and recorded on 10 November, where Tracy Alloway and Joe Weisenthal talk with Bloomberg's chief hedge fund correspondent Nishant Kumar and macro strategist Michael Ball about the October drawdown that hit multi-strategy funds together: Millennium fell 2.3% in a month against an annualised return near 14%, with Balyasny, Point72 and ExodusPoint losing in the same month, while Pivotal Path puts the ten largest firms at roughly $2.2 trillion of positions backed by $300-350 billion of capital. Kumar calls it a working hypothesis: they have no portfolio-level detail and do not know what Citadel and D.E. Shaw did differently. Every number is a reading from the recording date. Educational listening notes only — no investment advice, no price targets, and the judgements here may be wrong.
Contents

The cunning hare keeps three burrows, and that is all that saves it from death. You, my lord, have one burrow, and cannot yet sleep with your head high on the pillow.
—— “Strategies of Qi IV”, Zhanguo Ce (compiled by Liu Xiang, Western Han, from Warring States sources; my own translation)
The Odd Lots episode published on 11 November 2026 was recorded on 10 November, with Tracy Alloway and Joe Weisenthal hosting Bloomberg’s chief hedge fund correspondent Nishant Kumar and macro strategist Michael Ball on the October drawdown that hit multi-strategy funds all at once. Millennium fell 2.3% for the month, with Balyasny, Point72 and ExodusPoint losing in the same month, and since Millennium annualises around 14%, that 2.3% means something different landing on them than it does on an index. Kumar’s explanation is crowding plus leverage: the research firm Pivotal Path calculates that the ten largest firms it studied back roughly $2.2 trillion of positions, borrowed money included, with $300 to $350 billion of their own capital. He also stated the limit plainly — this is a working hypothesis, they have no portfolio-level detail, and they do not know how Citadel and D.E. Shaw produced a positive month.
I used to treat diversification as a counting exercise
Multi-strats — one firm housing hundreds of small teams that trade independently, known in the industry as pods, a word Kumar says sounds like an insult to many of the people in it — were something I understood roughly like this: one team trades US equities, one macro, one credit, one commodities, nobody talks to anybody, the return streams are independent, so the whole thing is sturdier than one star manager. What I counted was the number of bets and the labels attached to them, treating “different desk, different market” as “different risk”.
As for a month down 2.3%, my reaction was that this hardly counts as news. I compared it to the index and asked whether the money had beaten the market.
This episode took both of those apart.
Correlation stayed inside its old band — what cracked was one trade wearing several coats
Start with the gauge that should have gone off. Kumar cites Goldman’s work showing that correlations between multi-strategy funds themselves have sat inside a 0.2 to 0.3 band for years, and October did not push them out of it. By that reading, correlation had not deteriorated. What broke has another name: crowding, with leverage underneath it.
He separates crowding from momentum carefully. Ten sophisticated firms can each independently come to like the same stock — it is liquid, the earnings are strong, it fits their risk framework — and that does not turn them into momentum traders. The trouble arrives on the day they all change their minds at once, which looks like what happened last month.
Joe’s follow-up is the best question in the episode. Suppose one PM is long chips, another is long memory, and a third is short Treasuries because all that capex will drive inflation. Those can be three coats on the same AI trade, so if I am the risk manager and I failed to spot it, is that my fault? Kumar answers on two levels. First, correlation is not visible all the time; it moves up and down, and in a month like October it spikes — and putting interconnectedness, leverage and everyone’s positioning into one calculation is very complex mathematics. Second, he defends the platforms: they do care about correlation and they do manage it, but asking a risk manager to cut the Magnificent Seven while it is still climbing, and then to underperform, is a hard decision to make.
Then there is the sequence. October arrived week after week: AI stocks cracked, portfolios were rebalanced, bonds fell, credit markets ran into trouble, a surprise election in Brazil, French bonds sliding too. Kumar says there was nowhere to hide last month — the macro manager Chris Rokos lost almost 5%, and another macro fund running billions lost 6.4%, its worst month since it started trading. Concentrated single-strategy funds looked worse: one fund whose holdings overlap heavily with situational awareness lost 33% in October, and Kumar’s killer statistic is that it is still up for the year, which tells you how much it had made first. Whale Rock was down 12%.
The most concrete stretch of the conversation is the risk machine’s rulebook. A team gets $100 to manage; lose $5 of it and at many firms your capital is halved on the spot; lose another 2.5%, for 7.5% in total, and your book can be liquidated and you can be fired immediately. Kumar says that ruthlessness is precisely what makes the machine work, because it stops anyone losing enough to become the whole firm’s problem.
What follows is my own inference rather than his: hundreds of independent decision makers sharing one stop-loss rulebook means that rulebook is the thread tying them together. The action triggers at the same percentage, so the selling stacks up at the same time. Tracy’s closing remark runs the same way — the defining feature of multi-strats is the focus on short-term returns, the environment flips daily, they chase the flip, and they add to the force of the selling.
Michael Ball supplies the mechanical layer. On the Thursday two days after the election, the first thirty minutes of trading brought the biggest single-stock selling since COVID, with no macro catalyst; Wednesday had people rebuilding positions and volatility falling, which made Thursday a sucker punch. Once the S&P broke the 7,600 to 7,700 range it had held for months, the rules took over: investors buying protection leave dealers short those puts, so dealers sell into a falling market to hedge; leveraged ETFs have to rebalance; vol control strategies and trend-following CTAs have to de-gross as soon as volatility picks up. Ball calls Thursday a perfect storm that broke every range of the past several months.
Joe closes the loop on why 2.3% is serious. A pension fund or a sovereign fund is inherently long risk assets — it has to be — so the allocation to hedge funds buys a cushion for bad months. The problem with October is that the money failed to do the job it was bought to do. Tracy puts it shorter: they paid for the hedge and they did not get the hedge.
Now I ask about the exit rule before I count the holdings
If you hold seven or eight names spread across different sectors, you have probably asked yourself the same thing: does this count as diversified? And when bad news comes and they all fall together, do I just have to sit through it? After this episode, diversification became three questions for me. Who wrote the trigger that forces these things out, and at what number does it fire? Whose money is holding them up, and can it be pulled? And are their customers the same group of people? The count comes last.
Tracy’s move of linking the two interviews is the thing I most want to keep. She asks Ball whether the neocloud problem has the same shape as the multi-strat one if all of their tenants are exposed to AI one way or another. She thinks of these companies as real estate and asks about tenant quality. IREN sits right there as the example: a $77 stock in November 2025, $21 on the recording day, and part of what hit it was Microsoft wanting a chip upgrade, because chips now move faster than projects can be finished. Ball’s summary is that it is all operational risk now.
I am keeping the uncertainties too. Nobody knows what Citadel and D.E. Shaw did in October, so I cannot tell skill from being in the right place. Kumar says this week has already brought mean reversion and dip buying, but only 40 days are left in the year, and dozens of large risk takers have been fired, poached, or are sitting out long non-competes on the beach — whether anyone has the courage to take risk over the next two months, he cannot answer. Ball draws his line between years: 2027 looks reasonably secure on backlogs and prepayments, while the question mark is 2028, because new projects need state-level approvals. The metric he watches is whether compute demand is still growing, and so far there is no flinch.
The capacity numbers are worth holding onto, because they explain why people keep launching these firms. Millennium set out to raise about $20 billion this year, saw demand above $30 billion and took $22 billion. Meanwhile D.E. Shaw, Point72 and Citadel have returned something like $35 to $40 billion between them, and that money needs a new home. On the other side of the ledger, Jain Global has had problems, Walleye cut parts of its equities trading, and Eisler shut down. Tracy’s line is the receipt for the whole episode: more teams, more pods and more money do not necessarily translate into more adjusted returns.
One more thing. Kumar took a screenshot of all his stories from last week and posted it on X, and it got around 750,000 views — in sixteen years of covering hedge funds he had never seen a week like it. Tracy’s response was to propose merchandise: one of those religious devotional candles, with mean reversion on it, so you can light one and pray to the god of mean reversion.
Worth a look
- The Odd Lots episode of 11 November 2026, the source for this piece (the episode mentions it was recorded on 10 November)
- Pivotal Path’s report from last week on hedge fund position overlap
- Goldman’s work on correlations between multi-strategy funds
- The early Odd Lots episode with Krishna Kumar on multi-strats, referenced back to in this one
- The Odd Lots episodes on structured products, leveraged ETFs and vol funds, which explain Thursday’s mechanical selling
The one thing to take with you
Whether the things you hold are diversified depends on whose order they obey when they are forced to leave.
Every thread in this episode points to the same place. Hundreds of pods each run their own book, and their stop-loss percentage is written in one shared rulebook. Dealers, leveraged ETFs, vol control and CTAs are four crowds who have never met, called out to sell by the same price level. The neocloud tenants have different names and their rent comes from one theme. The items on the list are independent; the order that moves them is single.
Here is something I have tried that works outside investing. Take the three “just in case” things you set aside for yourself this week — a backup plan, a second contact, that afternoon you kept free at the weekend. Write one sentence after each: under what circumstances would I cancel this myself. Then read the three sentences and see whether they describe the same circumstance.
This article is an educational discussion of investment method. It is not advice to buy or sell any individual security, offers no target prices, and does not analyze any current holding. Investing carries risk; make your own decisions or consult a qualified professional.
Comments
Loading comments…