ExpertsFM20VC: How LPs Allocate to Venture in 2026: What They Want, What They Do Not Want | Why Fund Multiple Does Not Matter Without a Timeline | Why Velocity of Cashback is the Most Important Thing with Davi
The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch · September 14, 2026
5 Selected Lessons:
framework · 2 reasoning_pattern · 2 heuristic · 1
Private allocation bounded by liquidity/forced-selling avoidance
Source
The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch — 20VC: How LPs Allocate to Venture in 2026: What They Want, What They Do Not Want | Why Fund Multiple Does Not Matter Without a Timeline | Why Velocity of Cashback is the Most Important Thing with Davi (September 14, 2026) · at 9:50 ▸
When this applies
when setting an illiquid-vs-liquid capital split for any pool of capital with periodic cash obligations
The takeaway
Decide your illiquid-capacity ceiling before anything else, because that constraint is the hardest one to change later and the costliest one to violate.
Why it works
Illiquid assets can't be sold quickly in a denominator-effect downturn; setting a hard ceiling on the illiquid share ensures you're never forced to sell at a bad price.
One level up
Rank risks by reversibility: avoid irreversible catastrophic risks (fraud, forced selling) before optimizing for returns.
Where it stops holding
This stops working when the institution has no recurring cash obligations (e.g., distributions) or when secondary markets are liquid enough to exit privates quickly without penalty.
Detail
procedure:
- Determine the liquidity environment you can tolerate on the private side first
- Set an allocation range (e.g. 35-55%) designed so the upper bound doesn't trigger forced selling during a denominator-effect downturn
- Lock the box and let publics/privates float within that band rather than re-deciding structure each time
rationale: Private allocations are very hard to move at the margin (only secondaries help), so they must be sized conservatively up front.
In their words
Like the number one thing to avoid is fraud, and the number two thing to avoid is forced selling. That's a disaster.
Allocator's two-lane model: delegate expertise, own asset allocation direction
Source
The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch — 20VC: How LPs Allocate to Venture in 2026: What They Want, What They Do Not Want | Why Fund Multiple Does Not Matter Without a Timeline | Why Velocity of Cashback is the Most Important Thing with Davi (September 14, 2026) · at 24:15 ▸
When this applies
when deciding the division of labor between a principal allocator and specialist operators/managers they fund
The takeaway
Keep decision rights split along the axis of actual comparative advantage: macro/thematic calls stay with the allocator, micro/specific-asset calls stay with the domain expert.
Why it works
The allocator retains comparative advantage in top-down judgment (which themes/behaviors to bet on) while managers retain comparative advantage in bottom-up domain expertise (which specific names within a theme); mixing these roles correctly captures both edges.
One level up
Good capital allocation is a layered system of judgment, not a single binary choice between 'trust the experts fully' and 'do it all yourself.'
Where it stops holding
This stops working when the allocator lacks a genuinely differentiated top-down edge, in which case inserting their own thematic view just adds noise to the expert manager's process.
Detail
procedure:
- Identify a macro/thematic edge (e.g. human-behavior based insight on disruption)
- Direct additional capital toward managers/sectors aligned with that edge
- Let the manager apply their domain expertise to pick specific names within that directive
rationale: I'm making a decision based on human behavior... but I'm relying on the manager to be expert in their individual field and give me the correct perspective.
In their words
So I'm like, I'm going to give you more money, but I want you to go through your list with me and tell me all the things that are least likely to be interdicted by AI and then like own those.
Human-behavior edge over technical edge when assessing disruption panic
Source
The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch — 20VC: How LPs Allocate to Venture in 2026: What They Want, What They Do Not Want | Why Fund Multiple Does Not Matter Without a Timeline | Why Velocity of Cashback is the Most Important Thing with Davi (September 14, 2026) · at 20:52 ▸ · at 24:15 ▸
When this applies
when a market panics that an incumbent business model will be wiped out by a new technology
The takeaway
When markets panic about a technology replacing an entrenched system, test the thesis by asking the actual decision-makers who would have to make the switch, not just by reasoning about the technology's capability.
Why it works
Incumbent decision-makers (business owners) weigh switching costs, trust, and risk of operational failure heavily; calling real operators directly reveals that trusted, working systems are rarely torn out for unproven alternatives, even if theoretically cheaper/better.
One level up
Ground abstract technological narratives in observed human decision-making before acting on them.
Where it stops holding
This stops working when the incumbent relationship/trust advantage is weak, when switching costs are genuinely low, or when the new technology is dramatically cheaper/better such that even risk-averse operators switch anyway.
Detail
approach: Call real-world business owners/operators to test whether they would actually rip out trusted systems for an unproven AI alternative, rather than trusting the market's panic narrative.
signals:
- Switching cost / trust in existing vendor
- Criticality of 100% accuracy (books must match) vs AI's ~93% accuracy ceiling
- SaaS vendors' incentive to integrate AI rather than be replaced, given their own enterprise value
rationale: I actually think as we've kind of like thought about it more, I actually think that in some of these vertical industries, that software is actually going to be the delivery mechanism for AI.
In their words
So, like, if you need your books to like match up and whatever, like, yeah, that's not going to happen. So, I actually think as we've kind of like thought about it more, I actually think that in some of these vertical industries, that software is actually going to be the delivery mechanism for AI.
Diseconomies of scale in large funds/endowments ('law of large numbers')
Source
The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch — 20VC: How LPs Allocate to Venture in 2026: What They Want, What They Do Not Want | Why Fund Multiple Does Not Matter Without a Timeline | Why Velocity of Cashback is the Most Important Thing with Davi (September 14, 2026) · at 53:06 ▸ · at 56:55 ▸
When this applies
when any capital pool or organization scales to the point where individual wins stop being material to the aggregate
The takeaway
Recognize that past a certain scale, the game changes from picking winners to deploying size, and strategies/structures must adapt (e.g. bigger checks, platform bets) or accept diminishing relative returns.
Why it works
As the denominator (total AUM) grows, any fixed-size position's contribution to the whole shrinks proportionally, so even outlier winners contribute less and less to headline performance, making it structurally harder to clear a meaningful return hurdle.
One level up
Scale itself is a regime-changing variable that can silently invalidate strategies that worked at smaller size.
Where it stops holding
This stops working when the fund/firm redesigns its structure around scale (e.g. platform model, much larger check sizes) specifically to counteract the dilution effect.
Detail
approach: Evaluate whether a check size or fund size, even if it returns an exceptional multiple, actually contributes materially to the total portfolio's performance.
signals:
- Check size relative to total AUM
- Whether a theoretical outlier outcome (e.g. 50x) would still be immaterial at scale
- Shift toward platform/large-check models (e.g. a16z) as a structural adaptation
rationale: When you have $40 billion or $60 billion and you can allocate $20 million to a fund, even if you're up like a real lot doesn't move the needle as much as it used to.
In their words
I think it just gets harder. I think it gets harder to have a return figure over the requisite period of time that actually pays you for the risk that you're taking. I get how they do it. I get why they do it, but I think it is the law of large numbers, right?
Infinite portfolio diversity makes marginal concentration low-risk
Source
The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch — 20VC: How LPs Allocate to Venture in 2026: What They Want, What They Do Not Want | Why Fund Multiple Does Not Matter Without a Timeline | Why Velocity of Cashback is the Most Important Thing with Davi (September 14, 2026) · at 37:44 ▸
When this applies
when any sub-portfolio within a much larger, highly diversified total portfolio becomes temporarily more concentrated
The takeaway
Evaluate the riskiness of a sub-portfolio's concentration relative to the total portfolio's actual diversification, not in isolation or against a narrow benchmark.
Why it works
Because the total portfolio already spans vastly more asset types and sectors than any single benchmark index, incremental concentration within one manager's sleeve is a small perturbation against an already enormous diversification base.
One level up
Context (the size and diversity of the whole system) determines whether a locally risky-looking move is actually risky at the system level.
Where it stops holding
This stops working when the total portfolio is not actually as diversified as assumed, or when concentrated positions across multiple managers correlate in a downturn, defeating the diversification benefit.
Detail
rule: We own everything under the sun... it's always funny to me when people like compare an endowment portfolio to the SP 500... We're infinitely more diverse than the SP 500.
rationale: So, no, if we get a little bit more concentrated on at the margin, like that's, that doesn't remotely change anything for us.
In their words
We own everything under the sun. So does every ENF portfolio, right? So like we own everything from like sunscreen to helium to like tech, like I don't know what, right?