How a Position Earns Its Way Into the Book

Nothing enters the portfolio without clearing five gates. None of the five is original — screening, diligence, timing, stress testing, and sizing are the oldest checklist in the business. The work is in running all five, every time, in writing, and letting the process tell you no.

Gate 1: Screening

There are more than five thousand investable U.S. equities. Evaluating them by feel is a recipe for recency bias and narrative-chasing, so the first gate is fully systematized: a factor engine I built scores the full universe nightly, computing z-scores across value, growth, and technical dimensions with the comparisons done industry-relative — an industrial and a software company never get judged on the same absolute metrics.

The output is deliberately not a buy list. It's a research list: names where cheapness, cash generation, and improving fundamentals happen to coincide, which is the combination worth an analyst's hours. The screen decides what deserves attention; it never decides what gets bought.

Gate 2: Due diligence

Numbers can look great while the underlying business deteriorates, so this is where most of the time goes. Three questions, answered from filings and capital-allocation history rather than earnings-call language:

Who's running it? What management has actually done with capital over time — acquisition track record, insider buying, buyback discipline — tells you more than any forward guidance.

What protects it? Durable positions have a structural answer to "why can't someone else do this?" — switching costs, network effects, scale economics. No clear answer, no durable business.

Where is it headed? Whether each new dollar of revenue is getting easier or harder to earn — operating leverage building or eroding.

The gate's output is a written thesis: why this is mispriced, and specifically what evidence would kill the idea. If I can't write the kill condition, I don't understand the position yet.

Gate 3: Entry and risk/reward

Believing in a business doesn't make every price a good price. Every candidate entry gets mapped against where the stock has historically found buyers and sellers, and against prevailing momentum — and every entry carries a pre-committed exit. Not because I expect to be wrong, but because deciding where I'm wrong before committing capital removes the emotion from acting on it later.

The requirement is asymmetry: projected upside a meaningful multiple of the defined downside, with the downside enforced by a level, not a feeling.

Gate 4: Stress testing

Position-level analysis gives you a best guess; the stress test asks how the guess behaves inside the whole book. I model the portfolio across strong markets, weak markets, and outright crisis scenarios, because the uncomfortable empirical fact is that diversification breaks down exactly when it's needed — names that look uncorrelated in calm tape move in lockstep in a selloff.

The point isn't precision; the models are always wrong in the details. The point is being forced to look at the tail before it looks at you.

Gate 5: Sizing and construction

Sizing is where most of the money gets left on the table — a great idea with the wrong allocation is a mediocre outcome. Three inputs set the weight: conviction (higher-confidence positions earn more capital, inside hard caps), risk (volatile names in crowded sectors get sized down), and differentiation (if the finished book looks like the index, the fees and effort bought nothing).

Every position ends up with a defined role, a defined risk contribution, and a written reason it's there. Nothing is held by default.

What the gates look like in practice

A composite example — the shape is real even though the name is illustrative. A mid-cap vertical-software company clears the screen: top-decile industry-relative value score, free-cash-flow yield well above sector median, technical score turning up after a long base. Diligence finds a founder-CEO who has bought stock in the open market twice in eighteen months, gross retention above 95%, and a competitor whose product cycle has visibly stalled — thesis written, kill condition defined (two consecutive quarters of decelerating net revenue retention). Entry waits three weeks for price to return to a level that had absorbed selling twice before, with the stop set just below it — risking roughly 1 unit to a target of 3. The stress test flags that the book already carries two names exposed to the same enterprise-IT budget cycle, so the position starts at two-thirds of what conviction alone would justify. That's the whole system: five gates, each one allowed to shrink or kill the idea, none allowed to fall in love with it.

The system is the edge

The edge isn't any single gate — it's that all five run for every position, without shortcuts, with the reasoning written down where it can be audited later. This framework isn't built to find moonshots. It's built to surface a steady stream of adequately mispriced businesses, size them survivably, and let compounding do the rest.