Universe Selection as Hypothesis Testing

If you’ve ever stared at a 200-ticker watchlist and felt your IQ dropping by the second, this one’s for you.

The way most traders “do” universe selection is backwards: they assemble a logo soup, add a few sorting categories, then hope the market hands them a reason to care. Better yet: treat your universe as a testable claim about where your edge most likely lives and when it tends to appear.

Your watchlist is a falsifiable hypothesis about where your edge actually lives.

Hypothesis & Tagging

A watchlist built from logos is just a lazy scrapbook. But a watchlist built from a falsifiable claim is research. Your hypothesis names the behavior you monetize, the conditions it prefers, and where it should fail. If you cannot state an invalidating condition, you don’t have a claim—you have a feeling.

Test the claim with three questions:

1. What must be true in the environment for this to work (liquidity, volatility, catalysts, etc.)?

This is the validation framework for adding a ticker to your watchlist. Intuitively, you’ll learn to recognize tickers in this bucket at a glance. Each ticker must meet these criteria. But this part is easy.

2. What must not be true (e.g., forced flows, structural price floors/ceilings, cross‑asset stress)?

Here you will filter out tickers that make your edge harder to monetize. You might decide not to chase tickers with heavy insider selling or low daily volatility as it’s simply not worth your time and risk.

3. What would quickly invalidate a ticker (a specific behavior that, if seen, tells you to stand down)?

These could be a framework for setting stop losses, closing early, or taking partial profits. The general goal here is to decide what emergent signals would change the analysis. Blow-off price action, sentiment extremes, losing key levels are all examples of behaviors which would invalidate a ticker.

These answers are universal across products; they don’t require a preferred indicator or style, only clarity about conditions.

From hypothesis to tags you can actually see

Make the language of your claim the language of what’s on your actual screens. Keep the terminology tight and consistent so you can build theses with it. For example, you may decide that your watch list has the following tags based on the available filters in your stock screener:

• Structure: accumulation / manipulation / distribution
• Regime: trend / range / transition
• Catalyst: earnings / macro‑CPI‑NFP‑FOMC / regulatory / narrative
• Price Action: above MAs / high relative volume / leader vs laggard
• Risk high beta / event‑vol / illiquid
• Grade A / B / C / Archive

Tag hygiene

Keep tags orthogonal (each tag answers a different question). A name shouldn’t carry two tags that mean the same thing. When in doubt, choose the tag that governs risk.

Finviz is a top-tier stock screener for getting started. Tradingview is a solid runner up, but finviz has a treasure trove of filtering mechanisms for dialing in your watchlist. The closer you can align your screeners to your hypotheses, the more consistent your universe selection process.

Operating in Regimes (the muscle)

Regimes are the context that make the same setup pay or fail. Treat them as the first decision on every trading day. When you decide the regime first, the rest of your process involves necessary choices, not discovery. You stop scanning for “anything that moves” and start loading a prepared basket built for today’s tape.

Start with the environment

Begin by classifying conditions before opening your playbook. Ask: is liquidity abundant or thin? Is implied/realized volatility relaxed or stressed? Are there forced flows—OPEX, index events, a major macro print—likely to pin current prices or create ranging price action?

Your answer routes you automatically: trend tools for calm tapes, range tools for pinned weeks (ugh), capital preservation failsafes for stress. Put a small macro data layer on your primary screen. The choice is yours based on your ability to connect your understanding of the data to its impact. Again, the only macro data that belongs on this layer is data you know how to act on. Rates/curves, broad USD, energy/industrial complexes, credit markets, oil/gold behaviors are all feasible options. Otherwise, you’re just distracting yourself.

Design your attention, not just your coverage

Most discretionary P&L concentrates on a few names on a few days. The edge is seeing those days on time and being sized appropriately for them. Build out your screens and alerts to ensure you end up with a compact universe that you can trade cleanly. Nobody is watching 200 tickers. Keep a mix (such as core, rotational, exploratory, etc.), and explicitly blacklist traps: illiquids, halt-prone narratives, symbols that hijack your decision-making. It’s easy to come back to a ticker and forget why you avoided it in the first place. The goal isn’t more tickers, it’s fewer missed A-setups under the right regime.

Separate A from B, then defend it

Develop your workflow to discriminate between “leading ticker in a favorable regime” and “lagging ticker in a favorable regime.” The A/B line matters more than most other distinctions below it; noise lives in the B bucket. If you respect this cut, you’ll take fewer trades and hold better ones longer—because you’re aligned with leadership and conditions, not just a pattern that happened to print.

Keep a small time budget for exploring new markets or methods so your edge evolves, but defend the bulk of your attention for execution. Edges decay toward zero; your process has to keep pace.

Beware the correlation trap

Correlation is a tax you pay in bad times. A 30-name watchlist can turn into a five-factor bet during stressful market conditions.

When analyzing your watchlist, cluster tickers by behavior and treat each cluster as one unit of risk when sizing. Many traders simply do this by sector (energy, staples, healthcare, etc.). If positions will move for the same story, size them as one. You’ll like this discipline most when conditions flip and everything you “liked” suddenly trades as one chart.

Portability Across Instruments

Aim to have a selection process that lets you keep the thesis while changing the vehicle for expressing your trades. When the regime shifts, the cleanest expression of a trade might be an index future, a sector option, an FX pair, or the single name. A compact dashboard with both macro and cross-asset data (index, rates, USD, energy/commodities, credit) helps you rotate expression without getting stuck trading the same names that are in style one day, and out the next. process.

What Survives a Full Regime Cycle

Like a good carry-on, a durable edge travels well. As an experiment, take the same setup/entry logic you use in your bread and butter sectors across unrelated clusters and observe how much you need to adjust: parameters, hold times, exit behavior. Small, explainable drift is acceptable; entire process rewrites are not. Cross-validation is a common statistical process to test the robustness of your process. During testing, look at metrics like distribution changes—wins compressing, average hold expanding, or other red flags. If the observed friction consistently taxes your edge in particular markets, stop expressing the thesis there even if the pattern still “passes” all checks. The edge exists in the net, not on paper.

After the Watchlist

You’ve shifted from collecting names to making a claim about the tape. The question now is whether that claim holds up when conditions move.

Keep the loop simple: 1) Define the claim, 2) Define what would invalidate it. Many traders keep that condition visible while risk is on. When it appears, the claim is done; when it doesn’t, time in the position is justified.

Look at what’s actually paying. Results often compress to a few drivers—index beta, duration, USD, energy, credit, or a one-off flow. If most P&L comes from a singularity, note it and decide whether a different ticker would express the same idea with less noise.

Test portability. Take the same entry logic into another cluster or instrument and see how little you need to change. Small, explainable adjustments suggest durability; rewrites suggest a narrow, conditional edge.

Budget attention. Archive the themes that waste time or whose stories have played out. Promote ones that travel well across markets. If metrics like spread, slippage, borrow, or halts keep eating away at your edge, the thesis might be fine, but your expression of it is not.

Every trading day, leave the desk with one thing earned: a sharper question than you had at the open. If you can do that daily (claim stated, failure tested, driver understood) your intellectual P&L will compound whether the account is up or not.

Keep curiosity funded, guard attention like capital, and let patience carry the edge between regimes. Tomorrow’s tape is uncertain; your process isn’t. Show up and keep asking questions.