Happy Thursday and welcome.

After 8 months of BeyondTVPI, today we're going beyond IRR.

Every GP leads with IRR. It's the headline in quarterly reports, track record decks, and fundraising pitches. Most LPs treat it as the primary measure of returns.

And so did I.

But IRR is not a measure of wealth created. It's a measure of time-adjusted return efficiency. And it is acutely sensitive to the timing of cash flows.

And despite knowing this, I still underestimated just how sensitive it is.

That gap is what this piece is about.

On June 2nd, I'm speaking at European Emerging LPs - a community of institutional allocators from across Europe.

The topic: how capital actually gets allocated. Not the polished version. My version.

It's a closed forum, so I'll be naming names and going into some behind the scenes war stories.

If you're a member, I'll see you there. If you're not and want to be, check them out.

A subscription credit facility lets a GP fund an investment before calling LP capital. The LP receives the capital call later — repaying principal plus interest. The underlying investment does not change. The exit timing does not change. But the LP's clock starts later.

Let's take two identical underlying investments. We'll start with the real returns.

IRR is roughly 16%. TVPI is 1.80x.

Now add a one-year sub line at 8% interest. The LP doesn't get called until January 2021, pays $108M (principal plus interest), and still receives $180M at exit.

IRR jumps to roughly 18.6%. TVPI drops to 1.67x.

Most managers now report IRR both with and without the credit line effect, so dual disclosure is no longer the key thing to ask for. What matters is ensuring you're always comparing apples to apples — or at least knowing when you're comparing apples to oranges.

And always check TVPI alongside IRR: a fund with a rising IRR and a falling TVPI is usually paying for one with the other.

As we've already implicitly seen, IRR weights early cash flows more heavily than late ones. A dollar returned in Year 2 does more for IRR than a dollar returned in Year 7 — even if total capital returned is identical.

Of course, this is entirely fair. I'd rather have that dollar today than in six years.

But this creates a structural incentive: GPs who can generate early liquidity events report higher IRRs regardless of total value created.

The clearest example is a dividend recap. A portfolio company takes on debt. The fund receives an early distribution. The GP gets an IRR boost. The company carries more leverage. The LP may ultimately receive the same total value.

Consider two funds that each deploy $100M and return $200M over a seven-year hold period — identical 2.0x TVPIs.

Fund A: IRR of roughly 10.4%.
Fund B: IRR of roughly 12.9%.

In fact, for Fund A to equal Fund B's IRR, it would need to deliver an exit multiple of approximately 2.3x.

Identical IRR created.
Identical wealth created?

I'd say the jury is still out on that.

This one surfaces less obviously, and it's the one worth watching most carefully.

IRR treats unrealized NAV as a cash flow at the valuation date — effectively like a distribution. A fund with 80% of its TVPI in marked-up but unsold assets will report an IRR driven largely by those marks. The mark is a GP estimate. IRR presents it as a return.

Two funds can look similar on paper while representing completely different risk profiles:

Fund Alpha: 22% net IRR, 2.4x TVPI, 0.4x DPI. 83% of that TVPI comes from marks the GP set internally.

Fund Beta: 19% net IRR, 2.0x TVPI, 1.5x DPI. Only 25% of the TVPI is unrealized.

Fund Alpha's 22% may be entirely correct. It may also be a very good fund. But Fund Beta's 19% has been stress-tested by actual buyers at actual prices.

I've covered how to validate this at scale before. You can find the post here. For today, this is the key point: if a mature fund still carries large unrealized marks, dig deeper.

Effect 3 is IRR inflated by paper marks. This one looks different on the surface — but the underlying issue is similar.

A NAV facility is a loan the GP takes against the portfolio's net asset value, typically later in the fund's life. The proceeds get distributed to LPs. From a cash flow perspective, it looks like a real distribution — DPI goes up, the LP receives actual cash, and IRR gets a timing boost. But the distribution isn't backed by an exit. It's backed by the same unrealized marks discussed in Effect 3.

The LP has effectively borrowed against their own unrealized assets and received the proceeds as income. When the underlying companies eventually sell, that debt gets repaid first.

So where Effect 3 gives you IRR inflated by GP marks dressed up as returns, a NAV facility gives you DPI at a very real TVPI cost through interest expense.

It's pseudo-realized. The cash is real. The exit isn't.

There's a small silver lining. A bank willing to lend against the portfolio has done its own assessment of the marks - which is more validation than a fully unrealized fund can offer. But that comfort comes at a cost: the interest accrues, and it comes off your proceeds at exit.

One more thing IRR cannot tell you: whether the environment helped.

The same 15% net IRR can be first-quartile performance in a 2007 vintage — where the GFC hammered entire cohorts - and third-quartile performance in a 2016 vintage, where multiple expansion and easy financing gave most managers a tailwind. Same number. Completely different signal.

Comparing IRR without vintage context is like comparing two marathon times without knowing one was run at altitude in July and the other was a flat course in October. Same distance. Completely different conditions.

IRR is the entry point, not the verdict.

A few rules of thumb:

And that's part of the beauty of fund investing.

There is no single metric that predicts performance.

It's like pearls on a string - and no two pearls are quite the same.

Last week, I wrote about how Getting Things Done had changed my professional life.

It… didn’t perform. 5 likes on LinkedIn.

At least that’s what it looks like on the surface.

But an equal amount of people reached out privately. Two DMs on LinkedIn. Three emails. All of them had their own GTD story. A few of them took the time to recommend an app called Nirvana - which, as it happens, is exactly what I've used to run the system since 2018.

Five people isn't a big number. But five people took the time to write - because they’d had the same experience I had.

It may not be visible, but it’s real conversations.