How Sovereign Wealth Funds Create Alpha w/Peter Madsen

Madsen thinks market structure has changed, and is unsure today’s small caps resemble those of past decades. That doubt matters because a cheap index is not necessarily a basket of young, growing companies.

Weisburd relayed CIO Brad Conger of Hurdle Callahan’s view that small public companies differ from a decade ago. Conger pointed to SPAC targets, fallen angels, and listings that should not have gone public. Madsen replied that he agreed, adding that strong companies now tend to list at massive scale.

For example, Madsen still considers concentrated microcaps that might be acquired and biotech hedge funds within small-cap exposure. But he no longer treats broad small-cap cheapness as enough for a long-term overweight.

Madsen does not reject passive exposure. SITFO keeps some for liquidity, but he says benchmarks tend to grow more concentrated over cycles.

He calls choosing passive an active decision. It may reduce tracking error and management work. But that does not necessarily diversify risks the investor wants to control.

For example, SITFO uses rules-based public-equity strategies that weight value, momentum, and quality, then hold selected names. Madsen describes these as broad exposure with a beta of one, not as a hedge against every benchmark risk.

This middle ground still carries tracking error. It substitutes explicit rules for either pure benchmark ownership or discretionary manager selection.

SITFO began with an equal-weight perspective, then used quantitative work to stress-test its choices. Madsen says the initial portfolio came out fairly close to equal-weighted.

The approach was not mechanical. The team considered expected returns, available resources, and land-based revenues, mostly oil, gas, and real-estate development. It then reduced small-cap exposure and directed it toward private equity over about five years.

Madsen calls Sharpe optimization a perverse indicator when inputs are uncertain, returns are non-normal, or mean reversion matters. It can favor investments near a historical peak rather than the return the institution needs.

He says accepting volatility can be reasonable when the portfolio needs higher returns. But SITFO’s perpetual capital still has cash-flow obligations, with distributions going directly to Utah public schools. Land-revenue inflows also require forecasting so distributions stay manageable.

SITFO groups investments by purpose, then tests their factor exposures against benchmarks. A growth category can include public and private equity without making legal structure the investment thesis.

The test is whether asset-class labels hide shared risks. Madsen said private debt and real estate can carry equity beta, while infrastructure and real estate can carry interest-rate exposure.

Take data centers inside real estate. A holdings list can show the sector, while factor analysis can test whether its risks move with other exposures.

Madsen describes the approach as cheaper and easier to implement than a holdings-based risk system. Still, results depend on chosen factors and benchmarks, so the model makes assumptions visible rather than settling judgment.

SITFO initially set venture aside because its trustees shared a concern about access to famous, capacity-constrained firms. Madsen later built a philosophy around earlier-stage funds, after local seed investors presented their case.

He sees a different return equation in pre-seed, seed, and sometimes Series A deals. When valuations are arguably cheaper, smaller checks can buy more ownership for a given amount of capital. As an example, Madsen said a $100 million fund kind of needs five unicorns for five times. That is a scale illustration, not a forecast of repeated outcomes.

He also likes collaborative checks from solo GPs with strong networks. But Madsen offers no hit rate for these funds, so fund size alone does not establish an edge.

In October 2026, Gavin Ezekowitz on How I Invest said networked small funds can stay close to founders. The site’s summary says allocators should test whether that sourcing edge will last.

In November 2026, Jack Selby on How I Invest called mega-funds “a different asset class” and “a different bucket.” The site’s summary, not Selby directly, says LPs may need distinct benchmarks.

Matthew Harris’s October 2026 How I Invest moment describes Bain holding annual investment at $500 million to keep pacing even. That is a pacing comparison, not evidence for SITFO’s selection approach.

• GEGavin Ezekowitz · How I Invest · Oct 28Small funds can turn a focused founder network into an edge
• JSJack Selby · How I Invest · Nov 4Mega-funds and early-stage funds need separate benchmarks
• MHMatthew Harris · How I Invest · Oct 26Steady venture pacing helps limit cycle-driven risk

SITFO’s commercial partnerships exchange commitments for combinations of research, database access, software, revenue share, or discounted fees. Madsen says outside underwriting can extend a small team’s capacity.

He sees a particular case in private income, debt, and real assets, where he says upside is often limited. Co-invest accounts can avoid management fees, carry, and fees on committed capital.

Weisburd cited Steve Kaplan’s estimate that two and twenty reduces annual IRR by six percentage points. In the host’s hypothetical, splitting exposure evenly between traditional funds and zero-fee co-invests implies three percentage points across the portfolio.

That is the host’s illustration, not SITFO’s reported result. Madsen agreed with the fee argument, but the episode gives no SITFO performance comparison to verify that uplift. Partners receive commitments and other value in exchange, and terms vary.

SITFO is implementing AI to gather and organize manager diligence, including ownership, fees, and legal terms. Madsen said confidentiality concerns initially kept manager materials out of public ChatGPT tools.

Madsen describes a workflow that applies SITFO’s own criteria to meeting notes, manager documents, and legal materials. For example, it can surface ownership structure or flag terms in an LPA.

Madsen says the tool does not make investment decisions. Its intended return is analyst time for reading, discussion, and judgment.

He also said it was early, with no real success stories yet. The episode reports no measured time savings or accuracy, leaving its promised leverage unverified.