How to Build an Investment Thesis You Can Actually Score Against

An investment thesis is a written statement of what you back and why, specific enough that it sorts real candidates rather than just describing your taste. Most theses fail not because they are wrong, but because they stay a paragraph on a website and never become criteria that rank a cohort. This guide shows how to turn "what we look for" into weighted criteria that actually decide.

Ask ten funds for their thesis and you get ten well-written paragraphs: a stage, a sector, a geography, a worldview. Ask the same funds how that thesis scored their last hundred applications, and most cannot answer, because the thesis lived on the About page and the scoring happened by feel. A thesis that cannot rank a candidate is a marketing statement, not a decision tool.

This guide covers the difference between a fuzzy thesis and an operating one, the three axes any thesis should resolve into, how to translate them into weighted criteria, why an explicit anti-thesis matters, how to test the thesis on a real cohort, and how to let it evolve without becoming a moving target.

Fuzzy thesis vs operating thesis

A fuzzy thesis reads well and decides nothing: "We back mission-driven founders building the future of work." Every founder claims to be mission-driven, every product claims to build the future of something, so the sentence excludes no one. It feels like a position. It functions as a mirror.

An operating thesis is one you can run a candidate through and get a different answer for different candidates. It names what would make a company a fit, what would make it a pass, and how much each factor weighs. The test is simple: if two plausible applicants score the same under your thesis no matter who they are, the thesis is not operating yet. It is decoration.

The goal of this guide is to move you from the first to the second, without losing the conviction that made the thesis worth writing.

The three axes any thesis resolves into

However you phrase it, a workable thesis has to answer three separate questions about a candidate. The Deckwise Method names them, and they are the frame here because keeping them apart is what makes a thesis scoreable:

• Is it true? Your thesis assumes the claims in front of you are real. If a candidate's traction or market is inflated, no amount of fit matters. This axis is not usually written into a thesis, and it should be, because a perfect-fit company built on a false claim is a trap.
• Is it a good company, objectively? The parts of your thesis about quality: team, traction, market, product, moat. These hold regardless of who is investing.
• Is it a good fit for us? The parts that are specific to you: stage, geography, sector, values, and the things you will not do. Two funds can rate the same company identically on quality and oppositely on fit, and both be right.

A thesis that blurs these together produces one vague impression. A thesis that separates them produces three signals you can weigh, and a candidate that passes two and fails one becomes visible instead of averaged into mush.

Translate the thesis into weighted criteria

This is the step that turns a paragraph into a tool. For each of the three axes, list the specific criteria your thesis actually cares about, then decide how much each weighs, relative to the others.

The weighting is where the thesis becomes real, because it forces a choice you were making implicitly anyway. An early-stage fund that says it backs teams should weigh team above financials, and see that reflected in the score. A growth fund that says the opposite should weigh financials above team. A grant program should weigh mission fit and eligibility above commercial traction. If your weighting does not match what you say you believe, one of the two is wrong, and finding that out is the point.

Two disciplines keep the weighting honest:

• Decide it once, in advance, and apply it to everyone. Weights re-tuned per candidate are not a thesis. They are a way to justify a decision you already made.
• Write down the anti-thesis explicitly. The things you will not back, however good, deserve to be criteria too. A strong company that violates your anti-thesis is still a no, and if that rule lives only in your head, it will get overridden the day a shiny exception shows up.

The anti-thesis is half the work

Most theses describe what a fund wants and stay silent on what it refuses. That silence is expensive. The clearest funds can tell you in one sentence what they will not do: a stage they will not touch, a business model they distrust, a market they have decided is not for them, a red-flag pattern they have been burned by.

An explicit anti-thesis does two things. It speeds up every screen, because a candidate that trips it can be set aside without a long debate. And it protects you from your own weakness for a good story, because the rule was written when you were calm, not in the room with a founder who is very good at pitching. Put the anti-thesis in the criteria, weighted like everything else, so it actually pulls a score down rather than living as a vibe you forget under pressure.

Test the thesis on a real cohort

A thesis is a hypothesis until it meets data. Before you trust it, run it backwards over a cohort you already know: last cycle's applicants, or a set of companies whose outcomes you can see. Score them under your new criteria and weights, and look at what the thesis surfaces and what it buries.

You are checking two things. Does the ranking match your considered judgment on the companies you know well? And where it does not, is the thesis wrong, or was your gut wrong? Both happen. Sometimes the thesis over-weights something that felt important but does not predict. Sometimes it correctly flags a company you liked for reasons you cannot defend. Either way, the disagreement is the most useful output. A thesis nobody stress-tested is just a confident guess.

Let it evolve, without becoming a moving target

A thesis should change as you learn, but there is a failure mode: a thesis so fluid it means whatever the current deal needs it to mean. The fix is to version it. When you change a weight or a criterion, change it deliberately, date it, and keep the old version, so a past decision can be reproduced under the thesis that actually made it.

Versioning turns "the thesis evolved" from an excuse into a record. It lets you look back and ask whether last year's thesis would have made this year's decision, and it stops a thesis from being quietly rewritten to fit whatever you already wanted to do. Conviction and discipline are not opposites here. The version history is what keeps them together.

Frequently asked questions

What is an investment thesis? An investment thesis is a written statement of what a fund or program backs and why, specific enough to sort real candidates: the stage, sector, geography and qualities it looks for, and, done well, the things it explicitly refuses.

How is a thesis different from criteria? A thesis is the conviction; criteria are how that conviction scores a candidate. A thesis becomes usable when it is translated into weighted criteria applied identically to everyone. Without that step it describes taste but decides nothing.

Why include an anti-thesis? Because what you refuse is as decisive as what you want, and it protects you from a good pitch. Written and weighted, the anti-thesis pulls a score down on purpose, so a strong company that violates it is still a no.

How often should a thesis change? As often as you learn something real, but always deliberately and with the old version kept. Versioning lets the thesis evolve without becoming a moving target that means whatever the current deal needs.

How do I know my thesis is actually working? Run it backwards over a cohort you already know. If the ranking matches your considered judgment on the companies you understand well, and disagrees only where your gut was weak, it is operating. If it ranks everyone the same, it is still decoration.

The bottom line

A thesis earns its keep only when it can rank a candidate you have never seen. Resolve it into three axes, is it true, is it a good company, is it a good fit, translate each into weighted criteria set once and applied to all, write the anti-thesis down, and test the whole thing on a cohort you already understand. Conviction is the input. A scoreable thesis is the tool.

How this shows up in Deckwise

In Deckwise, your thesis is not a paragraph on a slide, it is the scoring basis the Selection pillar applies to every candidate. You set the criteria and their weights across the three questions, keep an explicit anti-thesis that pulls scores down, and the weighting is versioned so a past decision can be reproduced under the thesis that made it. Deckwise also sources candidates against that thesis in the Prospection pillar, so the same conviction that scores your cohort is the one that finds it.

Related: The Deckwise Method · The Complete Guide to Startup Sourcing · The Complete Guide to Startup Screening