Non-Consensus and Right
Excess returns only come from being both contrarian and correct — and you don't want to stay contrarian for long.
- Difficulty
- Moderate
- Time to result
- ~months to results
- Steps
- 5
- Confidence
- 88%
Borrowed from Howard Marks and repeated across Benchmark's generations, this decision rule holds that market-beating returns require a position that differs from consensus AND turns out to be correct — being contrarian alone (or right alone, since 'right' is often already priced in) isn't enough. A key addendum surfaced in the episode sharpens the rule further: you want your thesis to be non-consensus at the moment you make the bet, but you do not want to remain non-consensus for long — rapid convergence toward your view, driven by real evidence like growth data, is the signal that you were right rather than simply wrong and stubborn.
Origin
Coined by investor Howard Marks and cited repeatedly inside Benchmark by partners like Bill Gurley and Andy Rachleff as the core justification for bets the rest of the market dismissed, most famously eBay.
Core principles
- 01Consensus-priced opportunities cannot generate outsized returns because the upside is already reflected in price.
- 02A contrarian view is only valuable if it is also correct — being wrong-but-different destroys capital just as fast as being consensus-but-wrong.
- 03The best signal that you were right is fast convergence of the market toward your view.
- 04Staying non-consensus for years without validation is a warning sign, not a badge of conviction.
How to run it
- 1
Find where your view diverges from the market's
Identify an opportunity most investors are ignoring or actively dismissing.
- 2
Test the divergence against real evidence
Look for concrete, hard-to-dismiss data — revenue growth, retention, unit economics — that supports your contrarian read rather than relying purely on narrative.
Pro tip Prioritize investments where the traction is already visible but the market hasn't re-rated it yet.
- 3
Size the bet for genuine asymmetry
Commit capital proportional to the size of the mispricing and the strength of your evidence, not just conviction alone.
- 4
Expect and watch for convergence
Track whether the market starts moving toward your thesis relatively quickly as new information becomes available.
Watch out If the market never converges toward your view, reassess whether you were simply wrong.
- 5
Exit the non-consensus phase deliberately
Once your thesis becomes consensus and reflected in price, treat that as validation and plan your exit rather than waiting indefinitely for further re-rating.
In the wild
Most of Silicon Valley dismissed eBay as an ugly site for selling Beanie Babies with no real business model, but Benchmark saw already-profitable, self-funded growth with checks flooding Pierre Omidyar's mailbox.
→ Wall Street quickly converged toward Benchmark's view after the 1998 IPO, turning a $6.7M investment into roughly $4B within 18 months.
Benchmark backed Louis Borders' hyper-ambitious, capital-intensive online grocery vision when most VCs thought it was too early or too risky, sizing the bet at under 5% of the fund.
→ Webvan reached an $8B market cap before the dot-com crash; even though the business ultimately failed, the venture math on the initial bet was sound.
Common mistakes
Staying Non-Consensus Too Long
If a contrarian thesis never gains market validation within a reasonable window, it's more likely a sign you were wrong than a sign of unrecognized brilliance.
Chasing Consensus Deals for Prestige
Later growth-stage bets made mostly because a deal looked prestigious (rather than genuinely non-consensus) diverge from the discipline that produced Benchmark's biggest wins.
Is it for you?
Not ideal for
Investors without access to differentiated data or judgment — contrarian positioning without genuine insight is just noise, not edge.
From the transcript
“in order to make money in investing you have to be both non-consensus and right”
“you want that to be true when you make the investment but you don't want to be very non-consensus for long”
From the episode
Benchmark Part I