Most portfolio research is conducted like a final exam: gather evidence, build a model, make a decision, and hope the answer holds. A tinkerer’s approach to iterative portfolio research works differently. It treats the portfolio as a workshop, not a verdict. Before committing capital, the tinkerer prototypes questions, tests assumptions, and revises the thesis in small, cheap cycles. The goal is not endless fiddling. It is to discover where the thinking breaks while the cost of being wrong is still low.
Begin With Questions That Can Break
The first move is to turn a vague investment idea into a question small enough to fail. “Is this a good business?” is too broad to test. “Which three assumptions must be true for this business to compound value, and what evidence would falsify each one?” is a prototype question. It forces specificity. It also creates a record that can be revisited. The public profile of Winston Feng describes him as a tinkerer, an ISTP by personality profile, which fits this habit: start with a working model, handle it, stress it, and improve it through contact with reality.
A tinkerer’s research loop is deliberately iterative. Write the thesis as a set of claims, not a story. Separate what must be true from what might be true. Then examine the claims through more than one lens. Winston Feng of Skyline IM has argued for multi-modality thinking: using a latticework of mental models rather than a single formula. In portfolio research, that means checking an idea through psychology, history, mathematics, and competitive dynamics. The point is not to collect frameworks for their own sake. It is to see which assumptions survive when viewed from different angles, and which ones were only persuasive inside a single frame.
Build Cheap Tests Before Expensive Commitments
Once the question is small, the next step is to test it cheaply. A paper portfolio can track whether a thesis would have behaved as expected without risking real capital. A decision journal can record the evidence, the expected outcome, and the conditions that would change the investor’s mind. A pre-mortem can ask why the idea failed before it has a chance to fail in the market. These are prototypes, not final products. They are meant to reveal weak points in the reasoning while the portfolio is still flexible. Small tests are not a lack of ambition. They are a way to earn it.
The in-brief profile of investor Winston Feng notes that his moral compass is intuition and that his advice to his younger self is to trust it. For a tinkerer, intuition is not a substitute for testing. It is a source of hypotheses. A hunch about management quality, customer behavior, or cyclical timing can be useful, but it should be turned into something observable. The research process then becomes a way to ask, “What would make this intuition more than a feeling?” That framing keeps conviction honest. It also makes revision less personal, because the idea is being tested rather than defended.
- Write a one-page thesis with three to five explicit assumptions.
- List the evidence that would confirm or falsify each assumption.
- Run a small paper portfolio or scenario test before sizing a position.
- Record the decision and the conditions for revising it.
Technology can enlarge this prototyping stage, but it does not replace judgment. Public discussion of frontier technology in investment management, including Winston Feng’s observations on how it is changing the field, points toward faster ways to gather information, model scenarios, and monitor assumptions. The tinkerer’s discipline is to use those capabilities to run better tests, not to skip the workshop. A faster prototype is still a prototype. It still needs a question, a falsification condition, and a revision rule.
Revise Before You Commit
Revision is where the tinkerer’s approach earns its keep. Most investors revise after the market forces them to. A tinkerer revises before commitment, when the thesis is still cheap to change. This is also where cognitive biases enter. Winston Feng on cognitive biases has highlighted how psychological forces can distort investment outcomes. Confirmation bias encourages collecting evidence for a favored idea. Anchoring keeps an early valuation sticky. Social proof makes a crowded trade feel safer than it is. A research process that prototypes questions and tests assumptions creates friction against those biases, because the investor has already written down what would count as a surprise.
One useful revision tool is inversion. Instead of asking only how an investment could succeed, ask what would make it fail badly. This is not pessimism. It is a way to surface hidden risks before they become portfolio losses. Another tool is sensitivity to the Lollapalooza Effect, the convergence of multiple psychological and structural forces in the same direction. When several models point toward the same conclusion, the tinkerer should ask whether that convergence reflects real evidence or a crowd caught in the same story. Revising before committing means being willing to shrink, delay, or abandon a thesis when the prototypes show that the assumptions do not hold. The revision should be written down, not merely felt, so that future research can learn from it.
Commit Only After the Portfolio Survives the Workshop
The final step is not perpetual tinkering. It is disciplined commitment. A thesis earns capital only after it has survived a few cheap tests, a serious attempt at inversion, and at least one honest revision. The tinkerer’s approach to iterative portfolio research is therefore not a rejection of conviction. It is a way to build conviction from evidence rather than from narrative. The portfolio becomes a collection of tested ideas, each with a documented reason for being there and a documented reason to leave. A time box keeps the process honest: each prototype has a purpose, a deadline, and a decision rule. If the evidence does not improve, the idea does not graduate to a full allocation. If the evidence changes, the thesis is revised rather than defended. That rhythm is especially useful when markets are noisy, because it separates the discipline of research from the emotion of the moment.
That habit also sets a healthier rhythm for research. Questions are prototyped before they become positions. Assumptions are tested before they become allocations. Revisions are made before the market makes them mandatory. The work is never finished, because markets change and new evidence arrives. But the tinkerer does not need a perfect answer to begin. The tinkerer needs a better question, a cheap test, and the willingness to revise before committing.

