Actionable insights — the cheap lead-gen marketplaces

1. The "double-cheap" cyclical screen — low price-to-sales AND low PE at once

1. In a cyclical industry, don't lead with the PE — earnings swing too much, so a high PE at the trough and a low PE at the peak are both traps. Anchor on price-to-sales instead, which is far steadier across the cycle.
2. The classic buy signal is a low price-to-sales while the PE is optically high (depressed trough earnings). The rarer, stronger signal — the "double win" — is when BOTH price-to-sales and PE are low at the same time, which usually means the market has simply abandoned the name.
3. Treat valuation as the "cherry on top," not the thesis. Get excited about the fundamentals turning first; let the cheapness be the margin of safety, not the reason.

• Cyclical names where price-to-sales sits near a multi-year low while the PE looks high or normal — and especially the rare case where both are low.

2. Buy operating leverage at the cyclical trough — but only with a secular tailwind underneath

1. Inside a recovering cycle, find the operating-leveraged way to play it — the smaller, asset-light supplier whose profits swing far more than the big incumbent's when volumes return (more upside than owning the incumbent directly).
2. Demand a second, non-cyclical growth driver underneath it, so you're not just renting a cycle. The ideal is a multi-decade secular shift that keeps compounding regardless of where the cycle sits.
3. Time the entry to the cyclical trough (depressed stocks, brakes-on demand) while the secular driver is visibly accelerating — both winds at your back at once.

• Suppliers/marketplaces to a cyclical industry that also ride a one-way adoption curve (online migration of a previously offline transaction); the cycle bottoming as the catalyst.

3. The AI-victim / AI-neutral / AI-winner triage — find the crushed-but-fine companies

1. Start from the observation that AI/LLM fear has indiscriminately crushed a long list of fine companies — so the inefficiency is in sorting them, not in the fear itself.
2. For each beaten-down name ask: is it an AI victim (disintermediated/disrupted), AI neutral (AI neither helps nor hurts), or an AI winner (AI makes it more valuable)? Buy the neutral and the winners that the market has mis-sorted as victims.
3. Cross-check the company's own words: read the recent earnings calls where management answers the AI-threat question directly, and weigh whether the threat is real or a narrative spillover from an unrelated scare (e.g. the "software/SaaS" de-rating).

• Stocks down on an AI narrative that doesn't actually touch their economics; management's call-by-call answers to the AI-threat question as the tell.

4. The regulatory / proprietary-data moat test — can an LLM actually disintermediate this?

1. When the fear is "LLMs will disintermediate them," check whether the critical input is actually obtainable by an LLM. If the value sits in proprietary, closely-guarded data or pricing, the LLM can't replicate it.
2. Look for a structural lock: state/federal regulation, trade secrecy, or a counterparty (here the carriers) with a strong incentive to keep its "secret sauce" off any public utility — and regulators who agree.
3. Then flip the question: if the LLM can't replace the chokepoint, does the LLM instead feed it? A new top-of-funnel source of demand is a tailwind, not a threat.

• Businesses whose moat is regulated/proprietary data an LLM can't access; whether the LLM ends up as a new demand source rather than a substitute.

5. Hunt orphan stocks by mapping who covers them (and who doesn't)

1. For an unloved small-cap, list the sell-side analysts covering it and what desk they sit on (tech/internet, business services, financials, insurance).
2. An "orphan" is a name covered by the wrong specialty — IPO'd as one kind of business, but its real economics belong to another sector the covering analysts don't understand (and the right-sector analysts ignore).
3. That coverage mismatch is the edge: come at it with the correct analytical frame the consensus lacks, and accept that sub-$1B caps with three-to-five analysts will stay mispriced longer.

• Sub-$1B caps whose covering analysts are from the wrong sector; an IPO-era classification that no longer fits the business.

6. Score the capital allocator — and ride a disciplined anchor shareholder

1. Among similarly cheap names, rank by capital management: who is actually buying back stock at a low multiple versus who is hoarding cash with no plan.
2. Prefer the one with a disciplined, buyback-minded anchor shareholder on the board — a large, proven serial-repurchaser whose track record signals the cash will be returned at accretive prices.
3. Where management is hoarding, treat engagement as part of the thesis: push them (directly) to return capital; size the position partly on the odds of that change.

• Cheap stocks with a serial-buyback anchor holder; rising buyback authorizations relative to cash; cash-hoarders where activism could unlock value.

7. Long/short ambidexterity buys patience

1. Run a genuine short book (here ~50%) alongside the longs — own the undervalued/underappreciated, short the overvalued/overappreciated.
2. Use the opposite book as psychological ballast: holding a multi-year cyclical long is easier when an offsetting short portfolio means a market drawdown isn't pure pain — which lets you wait out the re-rating instead of being shaken out.
3. Keep the macro footprint small: don't need a market-direction call if the long and short legs are each right on company-specific fundamentals.

• Whether your sizing/temperament lets you add to a working thesis during a drawdown; an offsetting book as the enabler of patience.

8. Sell discipline — exit when a cyclical re-rates to a "normal" multiple

1. Define the exit before you own it: a cheap cyclical is held until the market stops treating it as fully cyclical and re-rates it to the double-digit EV/EBITDA "most companies" get.
2. The signal to sell is the re-rating, not a price target — once you "can no longer steal them," the asymmetry is gone.
3. For carriers/incumbents, use a "mission-accomplished" rule: scale out as the cyclical thesis fully plays out (a double from the trough) and rotate proceeds into the names with more upside left.

• A formerly-orphaned cyclical re-rating toward peer multiples; a position that has doubled on the cyclical thesis as the trim trigger.

9. Validate EBITDA as a proxy only when it ≈ free cash flow

1. EBITDA is a flawed shortcut (Buffett's critique) — it adds back real costs. Only trust it as a valuation proxy when the business is asset-light with little interest, depreciation, or amortization.
2. Confirm by lining up the last reported quarter's EBITDA, free cash flow, and GAAP earnings — if they're within a few million of each other, EBITDA is a fair stand-in for owner earnings.
3. Watch for the gap re-opening after an acquisition (new amortization), and adjust.

• Asset-light models where EBITDA, FCF, and GAAP converge; a post-deal amortization gap that breaks the equivalence.

10. The multi-year breakout from a tight range (David Hay's chart screen)

1. Screen for stocks emerging from a long, tight trading range into a multi-year breakout — the "dream scenario," strongest when paired with a clear improving fundamental story.
2. Require the fundamental driver to explain the breakout (e.g. assets/revenue growing fast against a fixed cost base, so margins expand) — chart + story together, not chart alone.
3. Systematize it: scan for names making multi-year new highs, then hand the shortlist to a sector specialist to pick the ones with the best fundamentals.

• Tight multi-year bases breaking out with a margin-expansion story underneath; new-multi-year-high scans filtered by a domain specialist.