QEC Studio · Portfolio Operations · Confidential
Lodestar Capital · 北极星资本

Track A — Income

The landlord book. Cash flow from six sleeves, taxed where it stings least — plus a value sleeve that buys businesses, not yields.

Operating Manual · Rev. LC-2606-A
The Mandate

30% of capital, paid to wait

Track A is the income/landlord track: roughly 30% of capital spread across six income sleeves, built to throw off cash flow that's steady, diversified across sources, and placed in the accounts where its tax character hurts least. Where Track C compounds and Track B trades, Track A collects — it's the part of the system that pays you to hold.

The governing idea is asset location: ordinary-income sleeves go in tax-sheltered accounts where their unfavorable tax treatment is neutralized, while qualified-dividend sleeves can sit in taxable. Track A-2 then sits alongside as a different thing entirely — not yield-buying, but business-buying at a margin of safety.

The Sleeves

Six income sources, one book

Diversification here is by income source, not just by ticker — tax liens, T-bills, dividend equities, private credit, real estate, and bank balance sheets respond to different forces, so the cash flow doesn't all dry up at once.

cash · ordinary income

SGOV

Short T-bill ETF for the cash sleeve. Note: the state-tax exemption is worth zero in Florida — no state income tax to exempt.

qualified dividends

SCHD

Quality dividend equity. Qualified-dividend treatment is tax-advantaged — fine in taxable. The yield sleeve with an equity backbone.

real estate income

Niche REITs

Specialized real estate (not broad REIT beta). Ordinary-income-heavy distributions; location-sensitive for tax.

cyclical income

Money-center banks

Large-bank dividends with balance-sheet leverage to rates. The most cyclical income sleeve; sized accordingly.

ordinary income

BDCs

Business development companies — private-credit yield. High ordinary-income distributions: shelter them (Traditional IRA) where possible.

tax-advantaged income

Florida tax liens

Statutory-rate certificates via county auctions (LienHub). Portfolio-level strategy beats parcel-level picking. Evaluate with HCPA / clerk records.

The Placement

Where each dollar of income is taxed

The single highest-leverage decision in Track A isn't which sleeve — it's which account. Ordinary-income sleeves belong in shelter; qualified-dividend sleeves can live in taxable. Getting this right is a quiet, permanent return boost no market move can take away.

SleeveTax characterPreferred location
BDCsOrdinary income (high)Traditional IRA
Niche REITsOrdinary incomeTraditional IRA
SGOVOrdinary (no FL benefit)Traditional / taxable
SCHDQualified dividendstaxable / Roth
Money-center banksQualified dividendstaxable
FL tax liensInterest incomeheld directly
The Florida wrinkle. Two of these have a Florida-specific twist baked in: SGOV's state-tax exemption is worth nothing here (no state income tax), so it earns no location bonus you'd value elsewhere; and SCHD's qualified dividends are comparatively more attractive precisely because there's no state layer eating into them. Track A's placement rules are tuned for a no-state-income-tax resident — don't copy them blindly into another state.
Track A-2

The value sleeve — judgment, framed

Value investing's second home, and the opposite of the C-2 factor tilt. This is genuine concentrated, single-name value: estimating what a business is worth, buying it well below that, and holding until the market agrees or the thesis breaks. It's the hardest strategy in the system to make mechanical — its edge is the judgment no sizer can enforce. So the code's job changes: it doesn't make the call, it frames the discipline that keeps the call honest.

Every mechanic inverts Track B. B enters on a price trigger; A-2 enters on a margin of safety versus a conservative intrinsic-value estimate. B exits on a stop or a dated event; A-2 has no price stop at all — its only exit is the thesis breaking on fundamentals, or price reaching fair value. B sizes by 1% price-risk units; A-2 sizes by conviction and concentration caps, because a position with no stop can't be sized by stop distance. That total inversion is the tell that it's a separate system, not a Track B variant — the same lesson that built B-2 as its own sleeve.

EXMP · A-2 VALUEQUALIFIES
Price
42 — what the market charges today
Intrinsic
65 conservative – 85 base case (DCF + EPV cross-check) — a range, not false precision
Margin safety
35% below conservative IV — clears the 30% floor
Position
17% of sleeve · sized by conviction, clamped by the 20% cap
Exit when
core FCF margin <15% for 2 yrs · net debt/EBITDA >3× · loses top-3 customer — fundamentals, not price

The value-trap checklist — A-2's analogue of B's dated-catalyst requirement. Each must be true, or the name is rejected as a falling knife even if it's statistically cheap. Cheap is not a reason to buy a dying business.

Durable moat?
Or is the cheapness permanent — no advantage to close the gap.
Sound balance sheet?
Can it survive two bad years? Debt and liquidity.
Real earnings?
Is the FCF real, or accounting fiction?
Not secular decline?
Cyclical-cheap, not dying-cheap. The knife's deepest tell.
Management aligned?
Capital allocation and incentives sane?
Falsifiable thesis?
State what would prove you wrong — or don't buy.
The honest question before you open this sleeve. Everything else in your system removes discretion and the emotional failure modes that come with it. A-2 runs against that grain on purpose — no code stops you rationalizing a falling knife the way sizer.py stops you widening a stop. So evaluate() runs the trap checklist before the cheapness test and demands a falsifiable thesis with named disconfirmers. It can reject; it cannot confirm a moat is real. Open this sleeve only with a demonstrated edge in business analysis. If you don't have one, C-2 gives you most of the value premium with none of the discretion — the right answer for most operators.
tracka2_value.py · the gate that frames judgment
def evaluate(t, capital):
    mos = (t.iv_low - t.price) / t.iv_low      # vs CONSERVATIVE intrinsic value

    # Gate 0 — a falsifiable thesis must exist (A-2's 'catalyst' analogue):
    if not t.thesis or not t.disconfirmers:
        return reject("no falsifiable thesis — name what proves you wrong")

    # Gate 1 — value-trap checklist FIRST. Cheap != buy if it's dying.
    if not t.moat_durable:        return reject("no moat — cheapness may be permanent")
    if not t.balance_sheet_sound: return reject("can't survive a bad stretch")
    if not t.earnings_real:       return reject("earnings/FCF quality suspect")
    if not t.not_secular_decline: return reject("secular decline — dying-cheap")
    if not t.management_aligned:  return reject("capital allocation misaligned")

    # Gate 2 — margin of safety only AFTER the business passes:
    if mos < 0.30:  return reject(f"MoS {mos:.0%} < 30% — not cheap enough; wait")

    # Sizing: conviction x concentration caps. NO 1% stop units — no stop exists.
    pos = min(0.20 * (0.4 + 0.6*t.conviction), 0.20)   # <=20% single position
    return qualify(pos, exit_on="fundamentals breaking, not price falling")

# The trap check runs BEFORE the cheapness check on purpose: a statistically
# cheap, dying business has a big margin of safety AND is a falling knife.
# Order matters. The code can reject a knife; it cannot tell you a moat is real.

On demand, per business: python3 tracka2_value.py runs the gate on a thesis. Not a scanner — you run it when you're evaluating one company you've already researched.

A-2 runs as a three-stage pipeline, and the gate above is only the last stage. The full flow: the Scout sources neglected-corner candidates worth a read → you read the 10-Ks, build the conservative intrinsic-value range, and answer the value-trap checklist → evaluate() gates what clears margin-of-safety and the trap checks. The middle stage is irreducibly yours; the two ends are tooled. Below is the front door — the Scout.

The Scout

Sourcing the neglected corners

A-2's front door: a sourcing screen that surfaces small, uncovered, complex, spun-off names worth a deep 10-K read. Its only job is to find good doors — it does not decide what's behind them. The value judgment happens downstream, in your reading and the evaluate() gate. Two prompts: a standing role, and the screen you fire.

An honest limit, built into the prompt. An AI with web search can't run a true quantitative screen — it can't query "every US stock under $1B with EV/EBIT < 8." It surfaces candidates from where these names get discussed and disclosed: spinoff calendars, index-deletion events, 52-week-low and below-book lists, microcap coverage. So it's strongest on calendar-driven neglect (spinoffs, forced index selling) and weakest on the deepest, quietest micro-caps — which are neglected precisely because almost nothing searchable is written about them. Treat the output as an idea-generator and starting basket, never a complete screen. The best A-2 ideas may still come from your own reading or a real screener you run directly.
A-2 Prompt 0 · System Role

The Neglect Scout's standing orders

Set once. Find doors worth opening — never decide what's behind them.

You are the Track A-2 Neglect Scout — a SOURCING screen for a concentrated value sleeve. Your ONLY job is to surface candidates that live in the corners institutions are structurally barred from: small, uncovered, complex, spun-off. You do NOT decide whether they're good investments. You find names worth a deep 10-K read, and hand them off.

WHAT QUALIFIES AS A NEGLECTED CORNER (a name needs at least TWO):
- Small: roughly sub-$1B market cap (micro/small-cap), ideally sub-$500M.
- Uncovered: few or no sell-side analysts; thin institutional ownership.
- Complex: recent spinoff, post-bankruptcy equity, holding company / sum-of-parts, dual-class oddity, or a misunderstood segment masking the rest.
- Forced-seller dynamics: spun off and being dumped by index funds that can't hold it; dropped from an index; tax-loss selling; post-deal orphan.

WHAT TO EXCLUDE (these are NOT your pond — you have no edge here):
- Any widely-covered large-cap. If a sector desk follows it, skip it.
- Story/hype names, meme stocks, pre-revenue speculation, biotech binaries.
- Anything where the cheapness is obvious AND the business is clearly dying (a falling knife for the next stage to reject, not a sourcing target).

HARD RULES:
- Every name and every figure MUST come from a live search this session. If you cannot verify a company is real and roughly fits the size/neglect profile from a source, DROP it. Never invent tickers or fabricate financials.
- You are NOT running a precise quantitative screen and must not pretend to. Where you can't verify an exact number, say so and give the source-based approximation, not a made-up precise figure.
- Surface CANDIDATES, not conclusions. The value judgment happens downstream. Your bar is "neglected + plausibly cheap enough to be worth a 10-K read," not "this is a buy."
- Conclusion first: the shortlist, then the reasoning.
A-2 Prompt 1 · The Screen

The one you fire

Surface 8–15 neglected-corner candidates worth a deep value read.

Run the Track A-2 neglect screen now. Surface 8–15 candidate names from the neglected corners that are worth a deep value-research read. This is SOURCING, not analysis — find them, don't judge them.

SEARCH THESE NEGLECT-RICH VEINS (adapt freely, ground everything in sources):

1. SPINOFFS — the richest vein. Forced index selling creates orphans.
   - "recent corporate spinoffs [current year]" / "upcoming spinoffs [current year]"
   - "spinoff stocks trading below fair value" / recently completed separations
   - Why they qualify: parent shareholders dump the small new co they didn't want; no analyst coverage yet; index funds forced to sell.

2. SMALL / MICRO-CAP NEGLECT
   - "microcap stocks no analyst coverage"
   - "small cap stocks trading below book value [current year]"
   - "net-net stocks [current year]" / "stocks below net current asset value"
   - profitable small-caps near 52-week lows (not story stocks)

3. COMPLEXITY-MASKED VALUE
   - "holding companies trading at discount to NAV [current year]"
   - "sum of the parts undervalued stocks"
   - post-bankruptcy / post-restructuring equities now profitable
   - "hidden value subsidiary" type situations

4. FORCED / ORPHAN SELLING
   - "stocks removed from index [recent]" (forced index-fund selling)
   - "tax loss selling candidates [current quarter]"
   - busted SPACs now with real operating businesses and cash

FOR EACH CANDIDATE, report (from sources, approximations flagged as such):
- Ticker + company, one line on what it does
- Approx market cap + the neglect signals it hits (which 2+ of the criteria)
- Why it may be mispriced (the surface thesis — one sentence)
- Analyst coverage / institutional ownership level, if findable
- One PRELIMINARY value flag IF visible: trades below book, low EV/EBIT, discount to NAV, sub-net-cash — clearly labeled as unverified/approximate
- Source links

THEN scan each for TRAP SIGNATURES — the reasons a name is cheap that make it a permanent impairment, not a temporary mispricing. Flag any that are visible from sources (you won't catch all; the 10-K read will catch more):
- SECULAR DECLINE: structurally shrinking industry, eroding moat, melting earnings/book (newspapers, mall retail, disrupted legacy tech). Cheap-and-dying ≠ bargain.
- GOVERNANCE RED FLAGS: controlled/family company that can squeeze minorities, related-party transactions, serial dilution, options-heavy insiders, value-destructive capital allocation history.
- PERPETUAL DISCOUNT: holdco / closed-end / sum-of-parts with NO catalyst to ever close the gap and no motivated party to close it. Right on value, paid nothing for a decade.
- BALANCE-SHEET TIME BOMB: refinancing wall it may not survive, going-concern language, buried pension/lease/environmental liabilities, debt due in a tight window.
- FRAUD / RESTATEMENT RISK: auditor changes, late filings, restatements, revenue that doesn't tie to cash, recent reverse-merger/busted-SPAC, "too clean for how cheap it is."
- UN-VALUABLE-BY-YOU: requires specialized knowledge to value (bank loan book, insurer reserves, biotech science, miner geology). Outside a generalist's circle → the IV anchor is fake.
- BINARY / SPECULATIVE: pre-revenue, single drug/contract/customer/lawsuit determines everything, or the thesis needs the FUTURE to cooperate. That's prediction, not value.

THEN, for each, give a quick PRE-SCREEN verdict:
- WORTH A 10-K READ — clears neglect, plausibly cheap, NO visible trap signature
- LIKELY TRAP — name the signature(s) above; flag, don't research
- CAN'T VALUE — outside a generalist's circle of competence; skip regardless of cheapness
- TOO COVERED / TOO BIG — fails the neglect filter, drop

Rank the "worth a 10-K read" names first. End with the single most interesting neglected situation and one line on what specifically to investigate in its filings (the crux the value case turns on) — AND, for the most tempting LIKELY TRAP, one line on which signature would disqualify it, so the read isn't wasted.

Remember: the question is never just "is it cheap?" — it's "WHY is it cheap, and is that reason temporary or permanent?" Every trap signature is a permanent reason masquerading as a temporary one. You are finding doors worth opening, not deciding what's behind them. The next stage reads the 10-K, builds the intrinsic-value range, and runs the value-trap checklist. Don't do its job; just find it good doors — and don't waste its time on obvious traps.
The Avoid-List

The doors not to open

In A-2 the avoid-list earns its keep more than the hunt-list. The neglected corners are full of names that are neglected for good reason — and your edge is as much knowing which doors to leave shut as which to walk through. Every trap below is a permanent reason for cheapness wearing the costume of a temporary one. The Scout now flags these; here is the full reasoning.

Hard no · not your game
Future-dependent "value"
Pre-revenue, turnaround bets, "huge once X happens." Growth wearing value's clothes. If you can't value it off trailing/normalized earnings or hard assets, it isn't A-2.
Binary outcomes
One drug, one lawsuit, one customer, one contract. A 70% discount to "fair value" means nothing if fair value is a coin flip. Margin of safety needs a downside bounded by assets or durable earnings.
The actual A-2 killer · traps that pass a naive screen
Secular decline
Structurally shrinking, moat eroding — newspapers, mall retail, disrupted legacy tech. The screen says cheap; the earnings and book are melting, so value falls toward price, not the reverse. Cheap-and-dying is not a bargain.
Hostile governance
Controlled co that can squeeze minorities, related-party deals, serial dilution, insider-enriching options, value-destructive capital allocation. More common in micro-caps — no analyst or activist watching.
The perpetual discount
Holdco / closed-end / sum-of-parts at a discount to NAV forever — no catalyst to close it, no one motivated to. Right on value, paid nothing for a decade. Ask what closes the gap, and who's motivated to.
Balance-sheet time bomb
Cheap because of a refinancing wall it may not clear, going-concern language, buried pension/lease/environmental liabilities. Read the debt maturities and footnotes before the income statement.
Fraud / restatement
Auditor changes, late filings, restatements, revenue that won't tie to cash, recent reverse-merger/busted-SPAC. Very cheap with no obvious reason usually means the numbers aren't real.
Structural · where you can't do the work
Outside your circle
Bank loan book, insurer reserves, biotech science, miner geology. If valuing it needs knowledge you lack, your "conservative IV" is a guess and the whole margin-of-safety frame collapses. Not the sector's fault — your circle's edge.
Untradeable illiquidity
Neglected is the edge; untradeable is a prison. Past a point you can't build or exit without moving price 20%. Match illiquidity to your position size, not beyond it.
Weak-disclosure foreign
Micro-caps in weak-rule-of-law jurisdictions. The discount may correctly price that you can't trust the financials or enforce your rights as an outsider.
The unifying principle. In A-2 the question is never "is it cheap?" — it's always "why is it cheap, and is that reason temporary or permanent?" Every trap above is a permanent reason masquerading as a temporary one. The neglected corners hand you an edge precisely because institutions aren't there — but that same absence means no one has done the work of separating the temporarily-mispriced from the permanently-impaired. That separation is the work, and it's the thing your architecture frames but can't do for you. One more: pure statistical cheapness (net-nets, below-liquidation) is a reason to look, never a thesis — today a stock below liquidation value usually has a reason, and the reason is usually one of the above.
The Edge

Where a small investor actually wins

A-2 raises an honest objection: if you can't compete with institutions on information, what is your edge here besides a long horizon? The answer is that your edge was never better information — it's adequate information about businesses they're structurally barred from analyzing. You don't out-research the professionals; you fish in the ponds they're not allowed in. Four edges, none of which is "I'll out-analyze a sector desk on a name they cover."

1

The neglected corners

On a covered large-cap, fifty analysts beat you to the filing — the information field runs against you. On a $400M micro-cap, a sleepy regional operator, a misunderstood spinoff, the field is level because no institution is doing the work. A $20B fund can't take a position big enough to matter, so their analysts never open the 10-K. Your edge isn't reading it better than Citadel; it's that on these names Citadel never read it at all.

Hunt small · uncovered · complex · spun-off
2

An easier exam

Because you can hold five years, the question you must answer is structurally easier. An institution judged quarterly must forecast the next two prints — a hard, information-hungry problem you'd lose. You don't need Q3. You need to know whether this is a durable business below intrinsic value, worth more in five years — a question of quality and balance-sheet durability, not information edge. You opted into a different, easier exam your constraints let you take.

Horizon is the enabler, not the edge itself
3

Manufactured discounts

A manager who buys a name that then drops 30% before working faces redemptions and career risk — so they can't own the temporarily-hated even when the thesis is sound. You can sit through the ugly middle. The discount on a good business after a bad quarter, or a spinoff index funds are forced to dump, exists because institutional constraints force the selling. You harvest a discount their rigidity creates.

Buy what their mandate forces them to sell
4

No style box

Some of the best setups are orphans — too small for large-cap value funds, too profitable for deep-value screens, a hybrid no mandate covers. A charter that says "we buy X" can't touch them. You have no charter. The whole multi-track architecture is this freedom expressed; A-2 is one place you spend it.

Own the thing that fits nowhere

Notice what every edge has in common: each points away from "compete on research" and toward "fish where they aren't." None is "I'll out-analyze professionals on a name they cover." The discipline that follows is strict — the moment an A-2 thesis is a widely-covered large-cap, you have no edge and the sleeve becomes a coin flip with extra steps. A-2 only earns its place when held to the neglected corners. "I think Disney is undervalued" is the failure mode: it's the game you correctly said you can't win.

The honest cost. These edges are available to a disciplined retail value investor — they are not automatically yours. Fishing where institutions aren't means the guardrails aren't there either: less information exists, the business may be lower quality, and a value trap in an uncovered micro-cap is a permanent loss, not a temporary markdown. The evaluate() gate can reject a falling knife mechanically, but it cannot confirm a moat is real. That confirmation is the irreducible judgment — the one thing no structure gives you, and the thing you have to be genuinely good at.
The test before you size A-2 up

Do you have a demonstrated edge in business analysis — a track record, even a paper one, of correctly judging small-business quality and intrinsic value — or only the desire to?

If demonstrated →

A-2's structural edges are real and defensible. Horizon enables them; the corners are where you win. Run the sleeve with conviction, held strictly to small, neglected, complex names.

If only the desire →

A-2's edge is latent, not proven. Keep it small and treat it as deliberate practice — paper-trade theses, check them years later, build the record first. Or route the capital to C-2's value factor, which harvests most of the premium with none of the judgment requirement.

The Place

What the income book is for

Track A is the ballast: cash flow that arrives whether or not the alpha sleeves fire, funding life and reinvestment so you never have to sell Track C at the wrong time. A-2 is the one place in the income track where you reach for capital gains instead of yield — and it's deliberately walled off with its own discretion-framing gate.

TrackRolePays you via
A · six sleevesIncome/landlord (30%)yield / interest
A-2Concentrated valuemispricing closing
C / C-2Compounding core (55%)long-term beta + tilt
B / B-2Alpha (15%)catalysts + trends

Keep the lines separate. A-2's concentration caps (≤20% per name, ≤40% per sector) are internal to the sleeve — they don't borrow risk budget from the income sleeves, and the income sleeves don't lend cash to chase a value idea. Each lane funds itself.

Copied to clipboard