Side Lesson 27: Building a Watchlist — Screens that Find Quality at Value

---

1. Why This Is Important

There are roughly 4,000 stocks listed on the major US exchanges (NYSE + Nasdaq, excluding ADRs, micro-caps under $50M, and the OTC ghost-mall). No retail investor — and very few institutional ones — can read a 10-K and an earnings call transcript on every one of them each year. The number you can actually own with conviction is not 4,000. It is somewhere between 5 and 20.

The job of a screener is to walk the funnel for you. You start with 4,000 names, eliminate 95% with three or four numerical filters, hand the survivors to your eyes and your spreadsheet, and end up with a watchlist of 10-15 names you actually understand plus a tighter buy-list of 3-7 you currently own. Doing this without a screener is like trying to find a wedding dress by walking through every store on Fifth Avenue. Doing it with one is like asking the saleswoman to pull only the white silk size-6 gowns.

Four reasons to take screening seriously:

Filter discipline beats stock-picking discipline. Alpha is rare.

The single highest-leverage decision you will make is which 30 names enter your due-diligence queue, not which one of those 30 you eventually buy. A bad shortlist makes good selection impossible; a good shortlist makes mediocre selection profitable.

It mechanises Buffett's "circle of competence." Every screen

has a thesis embedded in it. A GARP screen embeds *I believe ROIC above the cost of capital, bought below 1.2x PEG, compounds*. A deep-value screen embeds *I believe statistical cheapness in solvent businesses mean-reverts*. Writing the screen forces you to write the thesis.

Market caps do the heavy lifting. US-listed equities come with audited GAAP filings, deep secondary liquidity, and

the SEC EDGAR plumbing that makes all of this work. Run the same process on a Shenzhen B-share index and the data layer breaks. The $2B floor used in the GARP screen below is not a statistical convenience — it is Horace's no-penny-stock rule with a number on it: sub-billion names live in the casino tier (Side 14 makes the case in detail — wide spreads, manipulable float, thin disclosures, participants who are not investors), and the watchlist process exists above that line by design, not by accident.

It defends against narrative. The best filter against

"Cathie said it on CNBC" is not willpower; it is a numerical tripwire that says PEG > 1.5 -> no further work. Pre-committed filters convert the impulse buy into a process buy. (The market can stay irrational longer than you can stay solvent — and longer than your conviction, too.)

The funnel above is the canonical pipeline: **filter -> due-diligence -> watchlist -> position**. The numbers will not match yours exactly, but the shape should — most names die in the first two stages.

---

2. What You Need to Know

2.1 The Tool Stack — Free, Brokerage, and Paid

Three tiers of screener exist. The differences are not "feature count" but data depth, history length, and cross-sectional rigour.

Free tier (good enough for 90% of retail investors).

• Finviz (finviz.com) — fastest screener on the open web. Eighty-plus

fundamental + technical filters, real-time delayed quotes, sector heat-map. Free version capped at 20 screens; Finviz Elite at $39.50/month adds intraday data, backtesting, and email alerts.

• Yahoo Finance Screener — fewer filters, but free, ad-supported,

and integrated with the Yahoo data layer. Useful for quick one-shot checks.

• TradingView — strongest if you also chart. Built-in screener

with TradingView's technical indicators and 90+ fundamental fields. Free tier limited to one screen at a time.

• Schwab StreetSmart Edge / Fidelity Active Trader Pro — your

brokerage already includes a screener for free. Schwab's covers about 60 fundamentals; Fidelity's adds proprietary "equity summary score" composites.

Paid tier (worth it if you run weekly screens or hold >20 names).

• Stock Rover ($28-$40/month, $79/yr Essentials) — 600+ metrics

including 10 years of fundamentals, sector benchmarks, custom formulas, and GICS-aware ranking. The retail-investor gold standard for fundamental work.

• TIKR.com ($24-$30/month) — institutional-style data feed

(S&P Capital IQ behind it), 25-year financial history, transcripts, global coverage. Best if you also read 10-Ks.

• Roic.ai ($14/month) — clean ROIC-centric layout, useful for

quality screens; weaker on backtesting.

• YCharts / Koyfin (free + $39/month tier) — chart-first; Koyfin

is the closest free-with-a-paywall analogue to a Bloomberg Terminal that exists.

You do not need all of these. Pick one free + one paid and stop. The marginal hour spent learning a third platform is better spent reading a 10-K.

2.2 Screen 1 — GARP (Growth at a Reasonable Price)

GARP is Peter Lynch's screen, formalised. The thesis: *pay no more than 1x PEG for a high-ROIC business growing free cash flow*. It is the workhorse screen for tranche-1 (long-term compounders).

Filter	Threshold	Why
Market cap	> \$2 B	liquidity + audited Big-4 books
PEG (forward)	< 1.2	growth not already priced in
FCF growth (5y)	> 10%	thesis-confirming cash dynamics
ROIC (TTM)	> 15%	spread above ~10% cost of capital
Debt / equity	< 1.0	survives a credit cycle
Gross margin	> 40%	pricing power floor

A GARP screen run on the S&P 1500 in April 2026 returns roughly 80-120 names. Typical hits today (representative, not a buy list): MSFT, GOOGL, V, MA, ADBE, NOW, INTU, TXN, MCO, MSCI, AVGO, LIN, ANET, CDNS. Not all are cheap right now — the screen is the entry filter, not the buy decision. PEG below 1.0 is rare in 2026 because the index is trading at a 22x forward P/E vs a long-run mean of 16x. Loosen to 1.5 and you get 200+ names; tighten to 0.8 and you get 15 — which is the right number for a deep-dive queue.

2.3 Screen 2 — Deep Value (Statistical Cheapness, Solvent)

Deep value is the Graham screen, the academic HML factor, the Bogle "reversion to the mean" version. The thesis: *the market over-discounts boring, slow, or temporarily impaired businesses, and a basket of them mean-reverts.*

Filter	Threshold	Why
Market cap	> \$500 M	tradeable, audited
P/E (TTM)	< 12	bottom quintile
P/B	< 1.5	balance-sheet protection
Current ratio	> 2.0	survives 12 months without a roll
FCF (TTM)	> \$0	not a value trap
Debt / EBITDA	< 3	rate-cycle resilient

Deep value is the screen you run into market panics, not out of them. April 2026 is mid-cycle, the screen is sparse — energy mid-caps (DVN, MRO, OVV), some regional banks (FNB, RF, CFR), and a handful of consumer staples (KMB, K, GIS). In March 2020, the same screen returned 600+ names. The screen does not change; the market does.

2.4 Screen 3 — Quality Compounder (10-Year Track Record)

The Munger screen. The thesis: *a great business at a fair price beats a fair business at a great price*. This is the screen you keep open in the background and run quarterly.

Filter	Threshold	Why
ROE	> 20% in each of last 5 years	stable, not levered up
Gross margin	> 40%	moat indicator
Operating margin	> 15%	converts revenue to cash
FCF positive	8 / 10 years	proven through a cycle
Debt / equity	< 0.8	optionality, not leverage
Buybacks > issuance	net negative shares	management aligned

This is the most punishing screen of the four — it returns about 40 names in any reasonable market. April 2026 hits include AAPL, MSFT, V, MA, COST, ULTA, MCO, MSCI, ROL, PAYX, AZO, TROW, FAST. There is no PEG/PE filter here on purpose. This screen finds the universe; valuation enters at the watchlist stage. You wait for one of the 40 to fall 25-35% below its 5-year median EV/EBITDA, then you act.

2.5 Screen 4 — Recovery (52-Week-Low Quality)

The contrarian screen. The thesis: *good businesses occasionally trade near their 52-week low for reasons that have nothing to do with long-term economics. Catch them before the market re-rates.*

Filter	Threshold	Why
Price	within 10% of 52-week low	psychologically hated
Gross margin	> 25%	rules out commodity wreckage
Working capital	> 0	survives the next 12 months
Net debt / EBITDA	< 4	solvable balance sheet
Insider buying (90d)	> 0 net	management agrees
Short interest	< 8% of float	not a known short-seller target

Recovery screens are the easiest to mis-execute (the "falling knife" problem) and the most rewarding when right. Insider buying + short interest filters are the quality gates. April 2026 candidates: a small handful of regional banks post-2024 deposit-flight, a couple of consumer-discretionary mid-caps that missed Q1, and one or two healthcare names hit by tariff noise. Limit position size to 0.5-1% of the portfolio — these are tranche-4 (opportunistic), not tranche-1.

The scatter above is the visual frame. Anything in the lower-right quadrant is a candidate; anything in the upper-left is the "glamour-stock trap" (high P/E, mediocre ROE). Most US large-caps in April 2026 sit in the upper-right — quality at a premium. The GARP screen is, geometrically, *find the names heading into the lower-right.*

2.6 The Watchlist Pipeline — From Screen to Position

The screen is step 1 of a five-step funnel:

Screen (4,000 -> 250). Run the four screens monthly. Union the

hits.

Triage (250 -> 50). Eyeball each name for two minutes. Reject

anything with restated financials, going-concern language, SEC investigation, or sub-$200M average daily dollar volume.

Deep dive (50 -> 10). Read the latest 10-K (or three of them

for compounders), the most recent earnings transcript, and one short-seller report if one exists. Build a one-page thesis with five lines: *what they do, how they make money, what could kill it, what is it worth, what is it trading at.* (See side02_reading_10k.md.)

Watchlist (10 -> 3-7). Add to your tracked watchlist with

price triggers — the price at which you would buy, not the price you wish you had bought at. A watchlist without triggers is a daydream.

Position (buy when triggered). Use a 1/3-1/3-1/3 entry: buy

1/3 at trigger, 1/3 if it falls another 10%, 1/3 if it falls another 10% on no thesis-breaking news. The barbell says you are wrong sometimes; the staggered entry pays you to be wrong about timing.

The buy-trigger is the single most underrated discipline. Decide the price at the desk, in good light, with the 10-K open and CNBC off. By the time the price arrives, the market will be telling you not to buy.

2.7 Avoiding the Common Pitfalls of Screening

Three traps eat 80% of screening returns:

Trap 1 — Optimising on the back-test. The temptation to keep adding filters until the screen returns "all winners" historically is irresistible and fatal. Every additional filter narrows your universe and adds in-sample bias. Stop at 6 filters. Ever. (See week46_backtest_validator.html.)

Trap 2 — Confusing screen output with research output. The screen tells you which 50 names are worth investigating. It tells you nothing about whether you should own them. The 50->10 stage is where your edge actually lives.

Trap 3 — Not running the screen on yourself. Once a year, run the four screens with your own current holdings as the input universe. If a position you currently own does not appear in any of the four screens, you are holding it on inertia, not thesis. Either re-justify it from scratch or sell.

---

3. Common Misconceptions

"More filters -> better screen." Past 6 filters, you are

curve-fitting noise. The marginal filter typically removes more future winners than future losers.

"The same screen should always return names." Deep-value

screens are supposed to return 0-10 names in a frothy market. An empty screen is a regime signal, not a tool failure.

"Free screeners are not good enough for serious work." Finviz +

your brokerage's screener will get you 90% of the way to Stock Rover for a passive long-only book. Pay up only when your hours start to outweigh your money.

"PEG < 1 means cheap." PEG uses analyst consensus growth,

which is wrong on average and very wrong at turning points. Treat PEG as a triage filter, not a valuation conclusion.

"ROE > 20% means quality." Not if it is leverage-driven. A

bank with 15x equity multiplier and 1.5% ROA has 22% ROE and is not a quality compounder. Always check ROIC alongside ROE.

"A screen finding 200 names is too noisy to use." That is

exactly the right number to triage down. The problem with screens is too few hits, not too many.

"Insider selling is a sell signal." Insiders sell for a hundred

reasons (taxes, divorce, diversification, 10b5-1 plans). Insider buying, especially open-market and outside an earnings window, is the asymmetric signal.

"Screening should be automated." The screen should be saved

and rerun-able. The interpretation must stay manual. The day you automate "buy anything that screens green" is the day you buy the next Enron at 8x earnings.

"My watchlist is full -> I am done." A watchlist is a queue,

not a museum. Every name on it should have a price trigger and a review date. Names with neither are dead weight.

"Backtested screens that worked for 10 years will keep working."

The macro regime changes every 30-40 years. A screen that worked 1995-2020 (cheap quality) might fail 2025-2040 (rates higher, cheap-quality re-rates faster). Re-validate annually.

---

4. Q&A Section

Q: How often should I run the screens? A: Monthly is the institutional default. Quarterly is fine for retail. Daily is overkill — the names that pass a 5-filter screen do not move daily.

Q: I have 60 names from the GARP screen. How do I get to 10? A: First, eliminate sectors you do not understand. Second, kill any name you cannot describe in one sentence. Third, sort the rest by ROIC descending and take the top decile. You will be at 6-8 names.

Q: Can I just buy an ETF that does the screen for me? A: Yes. AVUV (small-cap value), QUAL (quality), MTUM (momentum), COWZ (cash-flow yield) are factor ETFs that mechanise screens 1-3. They are a perfectly reasonable substitute for individual stock screening if you are time-constrained. (See week50_factor_tilts.md.)

Q: How is screening different from factor investing? A: Factor ETFs hold all 100-500 names that pass a single factor filter. Stock screens are an idea-generation step that is followed by concentrated manual research. Factors win on diversification; screens win on conviction-per-name.

**Q: My deep-value screen hits the same energy mid-caps every month. Should I just buy them?** A: No. The fact that a name has been cheap for 36 months running is information about why it is cheap (structural decline, governance, commodity cycle). Run the recovery screen on that universe instead; the names that pass both deep-value AND recovery have the asymmetry.

Q: How do I screen for moats? A: You cannot, directly. Proxies: 5-year gross margin > 40% AND stable; ROIC > 15% across a full cycle; market share gains over 10 years. Screen for the financial fingerprint, then verify the moat qualitatively.

Q: What about screens for short ideas? A: Same engine, inverted thresholds: PE > 80, P/B > 8, FCF < 0 for 3 years, debt/EBITDA > 6, accruals > 10% of assets. Short alpha is real but the borrow + dividend-pass-through + carry is brutal — most retail investors should let factor short ETFs do the work.

Q: How long does this whole process take per week? A: Steady-state, 2-3 hours: 30 min to run screens, 90 min to deep-dive 1-2 names, 30 min to update existing thesis pages. The first month is heavier (5-8 hours/week) while you build the infrastructure.

Q: What if I run the same screens as everyone else? A: You will. The edge is not the filter — it is the 50->10 stage, which depends on your judgement. PEG < 1.2 + ROIC > 15% returns the same 100 names for everyone. Picking the right 5 from those 100 is where the alpha lives. (Alpha sources are stable, but each requires manual judgement at the buy decision.)

Q: Can I use these screens in tax-advantaged accounts only? A: Recovery and deep-value screens generate higher turnover — better in IRAs. GARP and quality compounders work in either; quality compounders especially in taxable accounts because hold periods stretch into LTCG territory by year 2. (See side04 for tax-location maths.)