Key takeaways
The relationship preceded the firm. Thirty institutions followed Marks to Oaktree before it had a single day of track record.
Performance and writing worked together. Without 19% net returns, the memos were context. Without the memos, the returns were just numbers.
Public distribution was never the point. Baupost's LP base is just as sticky, built entirely through private letters no one outside the fund ever read.
The most valuable memos were between raises. Marks wrote when nothing was being asked for, which is exactly why LPs kept reading.
One letter, written to no one in particular
October 12, 1990. Howard Marks is running the distressed debt group at TCW in Los Angeles. He is good at his job: analytically rigorous, well-regarded, with a solid track record in high-yield and distressed credit. He is also one of dozens of credit managers competing for the same institutional capital, with no particular reason for an allocator to choose him over anyone else.
No fundraise is open. He sits down and writes a letter to clients — not about performance, not about the quarter, but about how he thinks about risk, market cycles, and the psychology of investors who confuse momentum with insight.
He calls it a memo. Nobody told him to write it. Nobody told him to keep writing. But he did, for the next 35 years. What that decision compounded into is not quite the story most people tell about it.
The $1.5 billion that followed a mind, not a track record
When Marks and six partners left TCW in April 1995 to found Oaktree, the departure nearly ended before it began. TCW refused their request to retain management of the client funds they had been running. The new firm had no track record, no brand, and no client base. By any conventional measure, it should have spent years rebuilding from scratch.
Within three months, more than 30 institutional clients transferred $1.5 billion in assets to Oaktree. The Pennsylvania State Employees Retirement System was among them. Its CIO at the time was direct about why: they were following the team, and specifically Marks's ability to find value in specialised market niches.
There was no Oaktree fund record to point to. What those allocators were acting on was a way of thinking they had been reading for 5 years.
Why institutional capital moved on day 90
Pension funds don't transfer assets to a 90-day-old firm on instinct. They move when the IC conversation has already been prepared, when enough internal groundwork exists to defend a decision that would otherwise look reckless.
The memos had done that preparation. Not by marketing Marks, but by demonstrating across a full market cycle (including the S&L crisis and the 1990 recession) how he reasoned under pressure. By April 1995, the relationship had already been built. Oaktree was simply its next address.
Five years of memos before the firm existed had given institutional allocators a worked example of how Marks thought when markets turned difficult. No pitch deck replicates that. The rival managers watching $1.5 billion walk out of TCW to a firm that didn't exist yet faced a problem with no quick fix:
Intellectual output can't be manufactured retroactively.
A five-year analytical relationship with LPs can't be compressed into a fundraising timeline.
The asset Marks had built was non-transferable in the short term, and most of his competitors failed to understand why.
How Marks solved LP doubt before it existed
Every fund manager eventually faces the same conversation: a loss, a miss, or a quarter that needs explaining to an LP. Most of those conversations are reactive, assembled under pressure, after the fact. Marks had been having his in advance and in a register that was unusual for the time.
Most LP communications in the 1990s were performance summaries and quarterly letters. Marks was writing about market psychology, the structural mechanics of cycles, and the recurring mistakes investors make when memory is short and greed is running. He was treating LPs as intellectual peers, not as capital sources to be managed.
An IC member at a pension fund receiving one of these memos wasn't reading investor relations. They were reading the kind of analysis they weren't getting from anyone else in their manager universe, which is precisely why the memos circulated far beyond the LP base, and why Warren Buffett famously said they were the first thing he opened.
Published in November 2001, as the dotcom collapse confirmed what Marks had been describing for years, "You Can't Predict. You Can Prepare." wasn't assembled in response to the crisis. The explanation was already in their hands. The crisis was confirming it.
What gave this credibility wasn't just the accuracy. Marks was also honest when he was wrong. During the late 1990s, he was conspicuously cautious while the S&P 500 delivered five consecutive years of 20%+ returns from 1995 to 1999, an unbroken run with no precedent in the index's recorded history. LPs reading those memos and comparing their Oaktree returns to the S&P were holding underperformance with an explanation, not a result.
That honesty about being out of step, maintained in writing through the difficult periods, signalled something most managers never demonstrate: that the analytical register didn't change depending on whether performance was convenient. As Collateral has written on how allocators assess credibility, the signals managers send between formal conversations are often more revealing than the ones they send during them.
Would the memos have worked without the 19% returns?
A 30-year allocator would say: Oaktree's 17 distressed debt funds averaged 19% net annual returns over 22 years. You could have sent LPs a blank page every quarter and they would have stayed. The memos are a good story that strong performance made look meaningful.
That allocator isn't entirely wrong. A memo programme without those returns would not have produced the same outcome. Any manager who reads the Oaktree case as primarily a communications lesson is extracting the wrong variable.
The more precise read: strong, crisis-validated returns retained LPs. The memos made those returns interpretable across cycles. Both mattered, but they answered different questions:
Returns answered: should I be here?
The memos answered: do I understand what's happening here?
Both need answering across a ten-year fund life. That said, the memos only compounded because they were right. Marks's caution ahead of the dotcom collapse, his distressed positioning before 2008 — the writing built credibility because events validated it, repeatedly, over decades.
Why Oaktree kept raising more than it could take
By 2010, Oaktree was turning away capital due to self-imposed fund size limits. The 2021 close came in at $15.9 billion, exceeding its target. In February 2025, the largest distressed debt fund ever closed at approximately $16 billion. Each successive close drew from a LP base that had never been allowed to go cold.
That kind of demand doesn't accumulate during a fundraise. But it very nearly didn't happen at all. Between 1998 and 2000, Marks was sending out consistently cautious, cycle-aware memos while the market was delivering returns that made defensive positioning look like a failure of conviction.
LPs holding Oaktree through the late 1990s were underperforming the S&P with a written explanation for why. If the dotcom crash had been mild, or had not happened when it did, those memos would have looked like a manager talking himself into defensiveness rather than demonstrating foresight. The strategy required the cycle to turn. For nearly 18 months, it was genuinely unclear whether the memos were building credibility or quietly eroding it.
The cycle turned. But that period — not 2008, not the founding — was the moment of highest risk in this story.
When it comes to LP engagement between fund cycles, firms that only surface when capital is needed send a signal allocators recognise immediately. The Oaktree model was the opposite: present, analytical, and asking for nothing through the periods that tested it most.
What Baupost's closed-door letters reveal about the Marks model
The instinctive lesson most managers draw from Oaktree is that public, widely-distributed writing builds LP relationships. Seth Klarman at Baupost is the useful pressure test.
Klarman's annual letters to LPs are explicitly private, marked for LP recipients only, not for broader circulation. Baupost is routinely oversubscribed. Its LP base has demonstrated stickiness comparable to Oaktree's across multiple cycles, including periods of meaningful underperformance. Klarman has no public platform, no podcast, and has never sought institutional distribution beyond his existing LP base. The relationship holds anyway.
Public distribution is a consequence of Marks's memo quality, not its cause. What produced LP retention at both firms is sustained analytical engagement with existing LPs, maintained across market cycles, independent of whether a fundraise is active.
Ray Dalio's Principles sits at the structural opposite: a single foundational document, published once, front-loading the entire evaluative framework rather than compounding it across years. Different format, same underlying function: ensuring LPs hold a framework for interpreting a manager's behaviour before that behaviour needs explaining.
The memo is the least important part of the Marks story
Marks published irregularly, when he had something analytically distinctive to say, not on a schedule. The infrastructure for doing what Marks did in 1990 now exists at a fraction of the cost.
Substack, LinkedIn, and podcast platforms have made distribution a commodity. That accessibility is precisely why it's less effective. When everyone can publish, publishing is no longer a differentiating signal. In 1990, a distressed debt manager writing analytical letters to institutional clients was doing something almost no one else was doing. That novelty was part of what gave the output its weight.
Replicating the Marks model requires more than a publishing habit. Three things made it work, and all three are harder than they look:
Analytical accuracy, demonstrated over time. The memos built credibility because events validated them. Consistency without accuracy erodes LP confidence just as steadily as silence.
Output when nothing is being asked for. The memos written between fundraising cycles carried no commercial agenda. That absence of an implicit ask is what gave them weight, and that’s what most LP communications, written under fundraising pressure, can never replicate.
An existing LP base to engage. Marks began at TCW with an established client base. This mechanism works for managers maintaining LP relationships, not building them from scratch through content alone.
For managers thinking about their own investor reporting and LP communications, the question worth sitting with is not what to write. It's whether the output you're producing treats LPs as intellectual partners or capital sources. Consider whether that register is consistent enough and honest enough to hold across the cycles that will eventually test it.
Bottom line
Howard Marks didn't set out to build an LP retention mechanism. The relationship that resulted was consequence, not strategy, which is precisely what made it credible.
Most GP-LP relationships are transactional by design. They activate at fundraising, go quiet between cycles, and require reconstruction at each new close. The Oaktree model was something structurally different: a continuous intellectual relationship that gave LPs the context to stay when staying was difficult.
That asset was built before Marks had anything to sell. Which is why, when he finally needed it, it was already there. If you're thinking about how your firm's communications hold up across market cycles, not just during a raise, Collateral Partners works with private markets managers on exactly this.
