The “VC Algorithm”: What Percent Is Sourcing vs. Picking?

I was reading a great piece by Dan Gray recently where he argues that deal quality (sourcing) is the single biggest determinant of VC returns.

His thesis is simple but powerful: you cannot pick a winner if it isn’t in your funnel. If your deal flow is average, your “picking” skill is irrelevant. You’re just optimizing mediocrity.

It’s a compelling argument, so I opened my analysis to test it against 13 years of my own data.

I wanted to reverse-engineer my own “VC algorithm.” If I had to assign weights to the variables that actually drove my returns, what would the equation look like?

My audit confirmed Dan is directionally right but it also reveals that “sourcing + picking” is an incomplete formula.

Here’s my holistic algorithm:

1. Sourcing / Deal Quality (35%) – “The Ceiling”

Dan is right: this is the prerequisite. My data shows my returns dropped meaningfully in years where I relied heavily on inbound versus years where I actively hustled for edge-case deals. Access sets the ceiling. You can’t pick what you don’t see.

2. Winning / Brand (20%) – “The Conversion Rate”

In my top three winners, I wasn’t the only term sheet. The founders had options. If you have high deal quality and decent picking but low winning, your return is zero.

3. Sizing & Entry Price (25%) – “The Multiplier”

This is the variable most people underweight. My data shows my “angel pool checks” (small, risky) often outperformed my lead checks. Why? Because price matters. Paying for certainty compresses your multiple. Sourcing the right company doesn’t help if you enter at a 100x multiple.

4. Liquidity Management (10%) – “The Realization”

We obsess over buying, but my audit revealed a number of companies that could have been 3x–5x outcomes if we had sold earlier. Knowing when to sell is a distinct skill from knowing what to buy and it matters more than we admit.

5. Picking / Judgment (10%) – “The Illusion”

This is the controversial one. Why so low? Because at the Seed stage, the difference between a crazy idea that fails and a crazy idea that works is often indistinguishable at Day Zero. We give picking too much credit because it feeds our ego.

The conclusion? I used to think this job was 90% picking. Now I realize the algorithm is far more complex.

How would you weight the algorithm? Am I underestimating picking?