Best-performing investors to target as a founder: not just the famous names

---

1. Key Themes

Theme 1: VC Returns Are Radically Concentrated — and Brand Is a Poor Proxy for Performance

The core premise of the article is that the famous names in VC are not reliably the best-performing ones, and that the gap between top and average is staggering.

"By my estimates, about 5% of venture capitalists have generated roughly 90% of the industry's profits. If you are choosing whom to take money from, the question that matters most is whether your investor sits in that 5%. Fame is an imperfect proxy for it."

The power law applies not just to deals, but to firms themselves:

"By rank 10 the score has dropped to roughly 3,000, less than a third of Sequoia's. By rank 100 it is 245. The top firm scores about 41 times the hundredth."

---

Theme 2: AI Is Already Reshaping Who the Best Investors Are — In Real Time

The ranking's time-decay mechanism means AI investments are immediately rewarded, and firms that made early concentrated AI bets have rapidly risen in the rankings.

"For 23 of the 100 firms, the single highest-scoring investment is a frontier-AI or AI-infrastructure company. That is nearly a quarter of the list, anchored to a wave of companies most of which did not exist, or were tiny, five years ago."

"Firms that placed early, concentrated bets on the leading AI companies, such as Thrive on OpenAI, Menlo on Anthropic and Lightspeed on Mistral, are rewarded immediately rather than years after the fact."

---

Theme 3: Capital Efficiency and Concentration Beat Volume and Logo Count

The methodology explicitly penalizes check-spraying strategies and rewards concentrated, board-level involvement.

"Turning $10 million into $2 billion is a different achievement from turning $1 billion into the same $2 billion. Subtracting the cost of every investment rewards capital efficiency and penalizes spraying large checks to manufacture a few headline wins. In our data, roughly three-quarters of investments returned negative net profits."

The unicorn count disconnect makes this vivid:

"SV Angel has backed roughly 139 unicorns yet ranks 31st. Insight Partners has invested in about 124 and ranks 28th... Meanwhile DST Global, with 62 unicorns, ranks 4th, and Thrive is 8th with 47."

---

Theme 4: Firm Brand and Individual Partner Quality Are Largely Decoupled

One of the most actionable structural insights: the firm name on your term sheet is not predictive of who will actually help you.

"One of the most striking facts in our data is that fully half of our top 100 firms have no partner in our individual top 100. Firm strength and individual strength are far from the same thing, which is precisely why our methodology splits credit between them. The brand does not sit on your board; a person does."

Credit-splitting methodology further confirms this:

"Most return variation traces to individuals rather than institutions... We split it, giving a quarter to the firm where the investment was made and three-quarters to the firm where the partner works now."

---

Theme 5: Specialist and Sector-Focused Firms Are Systematically Underrepresented in Mainstream Rankings

The article identifies clusters of life-science and crypto-native firms that rank highly but would never appear on generalist media lists.

"A cluster of life-sciences firms earns a place on therapeutics rather than software: OrbiMed (27), Atlas Venture (38), ARCH (49), Versant (61) and Sofinnova (75), all ranking on concentrated, capital-efficient bets. Crypto-native firms such as Paradigm (34), Pantera (76), Multicoin (84) and Polychain (94) rank on a different opportunity set again. The methodology does not privilege any sector; it measures value created, net of cost and decay, wherever it occurs."

"If you are building in a specific space, the best-performing investor for you may be a specialist who will never appear near the top of a generalist media list."

---

2. Contrarian Perspectives

Contrarian 1: The Forbes Midas List Is Largely Unreliable as a Performance Ranking

The article makes a pointed, data-backed case that the industry's default investor ranking is not reproducible and has low correlation with actual performance data.

"When we tried to reverse-engineer it from its own disclosed methodology, even our best-fitting replication left 49 of its own top 100 absent from the actual list. Among investors who appear on both our ranking and Midas, the correlation is only about 0.27."

This is a striking indictment: a correlation of 0.27 is barely better than noise, meaning founders who use the Midas List as a guide to investor quality are operating on a significantly flawed map.

---

Contrarian 2: Unicorn Count Is Meaningless — or Worse, Misleading — as a Quality Signal

The conventional signal most founders use to evaluate VCs (portfolio unicorn count) is shown to inversely correlate with quality in several cases.

"How can a firm with a fraction of the unicorn count rank far higher? Because the methodology rewards value actually captured rather than the number of top deals... Expressed as points per unicorn, the spread runs from around six for the highest-volume names to more than eighty for the most concentrated."

A firm with 139 unicorns (SV Angel, rank 31) trails a firm with 47 (Thrive, rank 8) — because small, diluted checks into crowded cap tables create the appearance of success without capturing the economics.

---

Contrarian 3: Private Unicorn Valuations Are Systematically Overstated by ~50%, Distorting Almost Every VC Comparison

This is a foundational point that undermines most public narratives about VC performance.

"Two companies can both be described as worth $100 billion when one figure is a public market capitalization and the other a private post-money valuation. These are not the same number, because the preferred stock VCs buy carries downside protections that common stock lacks. My work with Will Gornall put the average overstatement for unicorns near 50%."

This means roughly half of the "unicorn value" reported in press coverage is an accounting artifact, not real economic value — a fact that should change how founders, LPs, and journalists interpret VC performance claims.

---

3. Companies Identified

---

4. People Identified

---

5. Operating Insights

Insight 1: Screen for Partner, Not Logo — Then Reference-Check Their Worst Days

The most actionable fundraising insight in the article is to go one level deeper than firm brand and diligence the individual partner, specifically on how they behave when things go wrong.

"Ask who specifically will work with you, what else they are carrying and how long they have been at the firm. Then reference-check that person with founders they have backed, including the ones whose companies did not work out. How an investor behaves in a down round is the information you most need, and no ranking, ours included, can supply it."

---

Insight 2: Competitive Round Pressure Is Precisely When You Should Slow Down on Investor Selection

The article makes a structural point about misaligned incentives during hot fundraising moments that founders consistently get wrong.

"An investor is your partner for seven, ten, sometimes fifteen years. That is longer than many marriages and considerably harder to exit. You can sell your house, change your product or replace your team. You cannot easily remove an investor from your cap table or your board. That asymmetry should make you slow down at precisely the moment a competitive round pressures you to speed up."

---

Insight 3: Match Investor Involvement Model to Your Actual Needs Before Targeting by Rank

Different top-ranked firms operate on fundamentally different models, and choosing the wrong model — even from a highly ranked firm — can be costly.

"A founder who wants an engaged thought partner and a founder who wants capital plus autonomy should be targeting different firms from the same top 100. A high rank cannot tell you which of the two you are dealing with."

---

6. Overlooked Insights

Overlooked Insight 1: A Decade Is Now Sufficient to Build a World-Class VC Track Record

Conventional LP wisdom holds that VC franchise quality requires multi-decade proof. The data challenges this.

"Twenty-one of the top 100 firms were founded in 2015 or later, and several rocketed up on a single recent, fast-appreciating bet... Track record compounds, and a decade is enough to build one."

This has implications for both founders (younger, hungrier firms may be underrated targets) and LPs (the case for only backing established franchises is weaker than assumed).

---

Overlooked Insight 2: More Than Half of the Top 100 Individual VCs Don't Appear on the Midas List at All

The upcoming individual-level ranking data, briefly teased at the end, may be the most disruptive finding of the entire research program — yet it's mentioned only in passing.

"More than half of our top 100 individual VCs appear nowhere on the 2026 Forbes Midas List."

If true, this means the list most founders and journalists use to identify star investors excludes the majority of actual top performers — a systematic blind spot with direct fundraising consequences.