I sell a governance product. Incomestead exists because I believe managing serious money is an operations problem, not a picking problem — and that belief is also my business model, which gives me a standing incentive to conclude you need what I'm building. The honest response isn't to hope you don't notice it; it's to publish the arguments against my own positions, in the strongest form I can write them.
For each of the eleven positions this blog is built on, I state what I believe, argue the best case against it, and tell you where that leaves me. Discount for my bias — but not by ignoring the argument; check whether each opposing case is represented fairly or quietly kneecapped. If the counter-arguments read as weak or strawmanned, I've failed the test that matters here, and you should trust the rest of the site less, not more.
How to read this page
Two kinds of claim run through everything I write. A fact is checkable and settled: under Regulation T you may borrow up to 50 percent of a marginable security's value, and a brokerage firm can sell your securities without contacting you if your equity falls too low. I cite those and move on. A position is a judgment a credentialed professional could dispute without being factually wrong — "margin is a dose problem" is a position, not a fact. This page is only about the positions.
Position 1: Margin is a dose problem, not a category problem
What I believe. Margin isn't categorically reckless — the danger scales with the dose, not the mere presence of a loan. Kept deep in the Clear zone and governed weekly it's a tool; run ungoverned near Forced, it's what ends you.
Why. The failure I watch for is a forced sale — a function of how close utilization sits to the line (Clear below 30%, Harvest 30% to under 35%, Freeze 35% to under 40%, Forced 40% and above), not whether a loan exists.
A sharp critic would put it to me this way: Leverage introduces a failure mode no discipline removes: a book with zero margin cannot be force-liquidated at any price, in any drawdown, ever — that is structural, not behavioural. Your zones govern the numerator, the loan; a crash moves the denominator, the collateral's value, which your discipline cannot touch. Your own SBLOC example proves it: a Clear-zone book at 28.7% went to 44.1% — Forced — on a market move alone, the borrower doing everything right. A discipline defeated by an event outside the operator's control is not a full answer to the risk.
Where that leaves me. I hold this with the most caution of the eleven. The steelman is right: governance manages the dose, not the denominator, and no rule stops a crash from repricing collateral. It's no proof — anyone who calls leverage safe rather than merely safer is selling.
Deep-dive: margin as a weekly management tool, governed to the Clear/Harvest/Freeze/Forced zones.
Position 2: Borrowing against assets can beat selling them
What I believe. For an asset you'd hold for decades, borrowing against it can beat selling it: you keep the compounding and don't lock in a taxable sale at a moment the market chose. Governed conservatively, it's a legitimate model, not a trick.
Why. A sale is irreversible and often badly timed; a modest, governed loan against the same asset lets it keep working while you spend.
A sharp critic would put it to me this way: You are paying interest indefinitely to defer a one-time tax, and betting the collateral never falls far enough to be called. Selling is simple, final, and cannot be margin-called. The advantage rests on a favourable spread between your borrowing cost and your asset's return — neither guaranteed nor under your control — and on a step-up in basis a legislature can revoke at will. And the people it worked for are a sample selected on survival: you hear from the families whose collateral held, not the ones who got called and sold at the bottom, and the survivors' outcome is not the expected one.
Where that leaves me. The steelman lands two clean hits: the spread isn't under my control, and the survivorship point is fair. So I hold it far narrower than the slogan — only when the loan is small, governed conservatively, and stress-tested against a call. Unconditional, it's reckless, and I say so.
Deep-dive: the borrow-don't-sell retirement model and its one catch.
Position 3: A governed DIY operator can beat an advisor
What I believe. A self-directed investor running a written governance process can match or beat a typical advisor on the thing that matters — not blowing himself up — while keeping the fee. I think most who fire an advisor were right to.
Why. Much of an advisor's value is behavioural and structural — stopping a dumb move, rebalancing without panic — and a written process supplies that without the conflict or the AUM fee.
A sharp critic would put it to me this way: An advisor is a second human being who acts when you are paralysed, ill, grieving, or simply wrong — a process you run alone still has exactly one point of failure: you. Governance doesn't remove the single-point-of-failure problem; it documents it. A page of rules cannot phone you in a crash, override you when you're panicking, or run the book the year you're too sick to open the laptop. An advisor also supplies what no document can: someone to argue with before you act, and a fiduciary if they steer you wrong. The honest comparison isn't fee-versus-no-fee; it's circuit-breaker-versus-none.
Where that leaves me. This is the one that worries me most — governance as the layer above strategy, my product's own metaphor, is what the steelman turns against me. A process is not a second person; it surfaces the weak-link problem, it doesn't dissolve it. A man who knows he'll freeze in a crisis should hire the circuit-breaker, even though saying so costs me the sale.
Deep-dive: portfolio governance — the layer above strategy that solo investors skip.
Position 4: Income should be counted only when realized
What I believe. Options income should be booked only when realized — on expiry, buyback, or assignment, on a cash basis, never marked to market. If the premium isn't yet yours to keep, it isn't income.
Why. Realized-only is auditable and can't be gamed: it never counts premium you haven't captured, and it reconstructs years later.
A sharp critic would put it to me this way: Realized-only systematically understates a book that is working. A quiet week with no expiries reads near-zero while premium is genuinely accruing across every open position — lumpiness that reads as underperformance to anyone but the person who chose the convention. Mark-to-market has a real argument: it tells you what the position is worth now, the number that matters if you have to close early — the exact scenario a risk system exists to handle. Realized-only buys auditability at the cost of timeliness — a genuine trade-off you keep presenting as free.
Where that leaves me. I hold this near full strength, but the steelman names the cost correctly: realized-only understates a working book in any given week, and it's the wrong lens for "what could I unwind this for today." My defence is scope, not superiority. Calling my convention the honest one is fair; calling the other one dishonest would not be.
Deep-dive: why counting premium you haven't booked lies to you.
Position 5: The 4 percent rule is the wrong frame
What I believe. The 4% rule quietly frames retirement as draw-the-pile-to-zero, then smuggles in its own biggest risk — sequence-of-returns — as if selling were the only option. Treating it as the only shape retirement can take is the error, not the number.
Why. Its buried assumption is that you fund retirement by selling. Change the mechanism — borrow against the portfolio, or live off realized income — and you've changed the question the rule was built to answer.
A sharp critic would put it to me this way: The 4% rule is the most stress-tested withdrawal heuristic in the retirement literature. William Bengen built it in 1994 on actual historical U.S. return sequences, including the worst on record, and its critics rarely bring comparable empirical backing. It is deliberately simple because simple rules survive contact with a tired, eighty-year-old version of yourself — exactly who has to run the plan. "Never draw down the corpus" demands more capital and more competence for decades; it may be better for those who can afford it and worse for everyone else. You are comparing your best case to its median case.
Where that leaves me. The steelman is right on both counts: Bengen's rule has empirical backing my frame doesn't, and "never deplete" assumes wealth and competence that aren't universal. So I've narrowed the claim — the 4% rule is an excellent default for drawing a portfolio down. My quarrel is only with treating a well-tested default as the only shape retirement can take.
Deep-dive: an alternative to the 4% rule, and its own downsides.
Position 6: A weekly cadence is necessary
What I believe. Governing a leveraged, income-oriented book needs a weekly cadence. It has to run on a schedule, not a mood, because the weeks you skip are the dangerous ones — drift doesn't wait for you to feel like checking.
Why. Margin and allocation drift are continuous; a breach that must persist two weekly readings before it drives an action can't be caught by a quarterly glance.
A sharp critic would put it to me this way: By your own account, most weeks the correct action is none. A quarterly review may capture nearly the same signal at a quarter of the cost, because the marginal value of the fifty quiet weeks is close to zero. There is also a real cost to looking often: frequent observation invites intervention, and intervention is where most self-directed investors destroy their own returns. A cadence that manufactures fifty-two decision points a year may be creating the very behavioural risk it claims to govern.
Where that leaves me. It changed how I state the position: most weeks are correctly no-ops, and frequency can manufacture intervention. My answer is that it's a monitoring cadence, not a trading one — the two-week persistence rule means a single week's reading can't trigger an action. Weekly is necessary for the reading, not the acting.
Deep-dive: why governance is a cadence, not a mood.
Position 7: The family bank can live in a brokerage
What I believe. The useful mechanics of a whole-life "family bank" — a compounding asset you borrow against instead of selling, enforced saving, something to hand on — can live in an ordinary brokerage account, at a fraction of the cost and with market returns instead of an insurer's dividend scale. What does not come with it is the contract's discipline. You have to supply the enforcement yourself.
Why. Strip the duelling return forecasts out of the whole-life comparison and what remains is structural: the premium is enforced saving, the policy loan is borrowing against your own accumulated value. A brokerage replicates both — cheaper, more liquid, more transparent — for anyone who will actually govern it.
A sharp critic would put it to me this way: The contract's promises are contractual; yours are aspirational. A policy loan is a right the insurer cannot deny — at worst defer for a few months — drawn against a cash value that does not reprice in a crash. Your version is a lender's discretionary credit against collateral that can fall 30 percent in the week you need it — and the broker can sell you out without calling first. Worse, the feature you dismiss as cost — the punishing premium, the near-worthless early surrender value — is the product: it is what makes an ordinary person actually keep the asset for thirty years. Your replication works precisely for the disciplined minority who least need a product to make them behave. You are selling the difficult version to the people the easy version was invented for.
Where that leaves me. The steelman's last line is the one that stings, because it's my own argument turned around: I believe discipline is the scarce ingredient — that's why I built a governance product — so I can't also claim the contract's enforcement is worthless. My answer is that the replication wins on cost and expected value but is genuinely worse on guarantees: it imports the two risks the contract removes, collateral that reprices and a lender with discretion. I take that trade for my own book, governed to the Clear zone — and I'd add that the contract has its own quiet failure mode, a policy loan compounding toward lapse. But for someone who knows they won't run the discipline, the expensive contract that runs it for them may honestly be the better buy — and I'd rather tell you that here than have you discover it after building the bank without the banker.
Deep-dive: the brokerage family bank, and the honest whole-life comparison.
Position 8: The Governance Score is a fair single number
What I believe. How well a book is governed can be compressed into one number without lying about it — Margin 35, Allocation 30, Cadence 20, Policy 15, my weighting and not a measured one — because the four things it weighs are the four things that actually end operations, and a single reading is the only form a person will look at every week.
Why. A dashboard of twelve metrics is a dashboard nobody reads. The Score exists to make one question answerable on a Saturday morning: is this better or worse than last week? Every component is itemised underneath, so the number is a door into the detail, not a substitute for it.
A sharp critic would put it to me this way: A single number is the thing you warn against everywhere else on this site. You tell readers not to act on one reading and to distrust precision a model cannot support — then you compress four components into one score using weights you chose. The weights are the argument, and no study says leverage is worth 35 and a written policy 15. Worse, a score invites the behaviour you police hardest: a reader who lifts their Score by trimming the position the Score deducts for has not governed anything, they have moved a number you wrote and sell.
And two facts you know but do not put here: the parameter file is stamped DRAFT_v0, so by your own admission the absolute level is uncalibrated — nobody has shown that 82 means "Governed" — and the free tool cannot measure Cadence at all, so it drops that component and renormalises to Margin 43.75, Allocation 37.5, Policy 18.75. Every reader who meets your Score in public therefore meets a three-component number on different weights from the four you just published. Sharper still: Policy, 15 points of the private number and 18.75 of the public one, awards full marks for having written a Charter — an artifact you supply. Your Score pays readers for adopting your framework, on the very page where you promised to disclose your incentive. And on the free tool the reader types their own utilisation, so the largest component of the public number is self-graded; capping a stated sub-score at 85 blunts that, it does not remove it. You are defending an instrument most of your readers have never seen, and grading the ones who have on whether they bought in.
Where that leaves me. The weights are a judgment and the page should say so before the steelman does, not after — which is why the number above now carries that label inline. What I defend is the ordering rather than the arithmetic: leverage can end the operation in a week and a missing policy cannot, so margin has to outweigh policy, and the precise ratio matters far less than that.
On the gaming charge I have to be exact, because my first answer to it was wrong: the margin gate caps the composite, but that is a margin mechanism and it does nothing about a reader trimming a concentrated position to recover an allocation deduction. The honest answer is that this particular "gaming" is the system working — a book above the concentration threshold really is more fragile, so trimming it really does improve governance, and metric and goal coincide. The exploit I cannot dismiss is the calibration one: on a DRAFT_v0 pack, the week-over-week delta is the load-bearing signal and the absolute level is a working estimate. That is a real limit on the word "fair", and the renormalised public number deserves to be stated wherever the Score is, not just here.
Deep-dive: what the Governance Score measures, component by component.
Position 9: Recommend-only beats autonomous execution
What I believe. The engine should measure faithfully, recommend clearly, and leave the decision — and the accountability — with the person whose money it is. I do not want software trading my book.
Why. A person acting slowly can catch themselves mid-mistake. Software acting autonomously executes the whole mistake first and reports it afterwards: the failure and the discovery arrive in the wrong order.
A sharp critic would put it to me this way: Your objection is that software executes the whole mistake before reporting it. Granted — and it is the wrong comparison. The comparison is not software against perfection, it is software against you, at the worst possible moment, and you concede the decisive point yourself: a page of rules cannot phone you in a crash. A system that acts cannot be talked out of acting by the frightened person it was written to protect, which is the entire reason to write rules down in advance.
You also present a false choice. Nobody has to hand you the keys to get pre-commitment — a conditional order sitting at the broker enforces a rule with no vendor in the loop at all, which makes "recommend-only versus autonomous" a distinction about your product rather than about the reader's options. And the moral distance you are claiming is thinner than it sounds: your engine already ranks the three liquidations it thinks best. You have built everything up to the button and then declined to fit the button, which protects you more clearly than it protects them — and the thing it protects you from has a name: software that decides and trades on a client's behalf invites exactly the licensing and discretionary-adviser exposure that recommending does not.
Where that leaves me. This is closer than I would like, and the steelman has the better of the general case: for an investor who reliably abandons their own plan, a rule that executes itself probably beats a rule that asks — and the broker-side conditional order is a genuinely good answer that costs nothing and does not involve me, so I should say so rather than let the choice look binary.
The last blow is the one that lands hardest, because it is true: ranked liquidation options are everything up to the button, and declining to fit the button does protect me. I will own the asymmetry rather than dress it up. My position is narrower than "recommend-only is safer" — it is that autonomous execution moves the accountability to me while the consequences stay with the reader, and its mistakes are the kind you discover afterwards, at scale, in the market that caused them. I would rather ship the design whose failure mode is a reader ignoring good advice than the one whose failure mode is my code selling a book at the bottom. That is a choice about which error I am willing to be responsible for, not a claim to the higher-performing design.
Deep-dive: why I want an engine that recommends rather than trades.
Position 10: A governance tool beats a spreadsheet
What I believe. Past a certain complexity — real leverage, several strategy Layers, a target you are supposed to measure against — a purpose-built instrument beats a spreadsheet, because the spreadsheet cannot stop you editing the target in the same cell you measure against it.
Why. Governance needs a record that does not move: frozen targets, a version-stamped history, arithmetic that reproduces byte-for-byte. Those are properties of an instrument, and they are what a hand-maintained sheet quietly gives up.
A sharp critic would put it to me this way: You are the party selling the tool, so start there. The spreadsheet's weakness is that it is manual — and that is the whole of its strength. Typing the numbers is what makes you look at them. A dashboard that refreshes itself is one you stop reading by month three, because nothing in the ritual needs you any more. Every failure you describe on this site is an attention failure, and software is superb at removing exactly the friction that was holding the attention in place. A spreadsheet a reader opens on a Saturday beats an instrument they subscribed to and stopped visiting, and you cannot tell those two readers apart from the outside.
Then the parts you never price: the sheet is free and permanent, yours is a subscription from a single operator, pre-revenue and invite-only. Your headline benefit is a version-stamped history that cannot be overwritten — and it evaporates completely the week you lose interest or run out of money. You are asking a reader to move the system of record for serious capital onto a one-person dependency. And they cannot audit the guarantee you are selling: the engine is closed, so "reproduces byte-for-byte" is a promise, not something anyone outside can check. And the fee is not the real exposure: to use it at all a reader hands full brokerage position exports to a one-person operation, which for money this size is a custody and data-security question you never raise while carefully pricing the subscription. The sheet also lets them model their book their way; yours imposes your taxonomy on it.
Where that leaves me. The friction argument is the one I have no clean answer to, and I should stop treating it as a straw man: maintaining the sheet is doing real behavioural work, and automating it away is a cost I price nowhere. My honest position is narrower than the one I am tempted to argue — a spreadsheet you open every week beats an instrument you don't, and if that is your situation you should keep the spreadsheet. What the instrument buys is not attention, which it cannot manufacture; it is a target you cannot silently edit and a history you cannot accidentally overwrite.
On auditability I have to concede rather than substitute: you cannot audit my implementation, only my method. Publishing the arithmetic and the thresholds lets you check my reasoning against your own numbers; it does not let you verify that the code does what I say, and no amount of published method closes that gap. The dependency point is fair and mostly unanswered — an export that leaves as a dated file you keep is the least I owe anyone whose record of decisions lives in my product, and "the vendor might vanish" is a reason to hold your own copy rather than a reason I can argue away.
Deep-dive: the honest spreadsheet-versus-software trade-offs, and why a dashboard is not a governance tool.
Position 11: A written concentration ceiling is right
What I believe. You should decide in advance how much of your net worth one position may become, write it down, read it weekly, and act when a breach persists. Not because there is a correct number — there isn't, and it is a house rule either way — but because the decision has to be made before you are attached to a winner.
Why. A position allowed to grow without limit stops behaving like an investment and starts behaving like a single point of failure. The ceiling is that decision, made while you can still think clearly about it.
A sharp critic would put it to me this way: Your own Position 2 spends a section arguing that selling is the expensive move — that you should borrow instead, because a sale locks in a tax bill at a moment the market picked. Then this position tells me to trim on a calendar. Enforcing a ceiling by trimming pays a certain, immediate capital-gains bill to buy an uncertain, probabilistic reduction in risk, and on a long-held winner that tax is often the largest number in the whole decision. You cannot hold both positions without saying which one gives way.
Second, you say nobody can name a universal number and that anyone quoting one is selling — while the product you sell does exactly that: the Governance Score deducts 15 points if any single position tops 20% of the book, and 35 if it tops 30%, applied to everyone identically. So you do publish a universal concentration number; you just decline to defend it here. Third, almost every large private fortune was built by not diversifying, and a ceiling fires on the position that is working, at the moment it is working hardest, on a schedule that knows nothing about the business underneath it.
Where that leaves me. The tax blow is the strongest thing on this page against me, and the two positions really do collide if the only way to honour a ceiling is to sell. My answer is that trimming is one instrument and not the required one: a persistent breach can be met by borrowing against the book rather than selling into it, by hedging the position, by donating appreciated shares, or — the cheapest and slowest — by directing every new dollar elsewhere until the weight comes down on the denominator instead of the numerator.
The critic gets the last word on this, though, and I have not got a clean answer: borrowing against the book to avoid realising raises margin utilisation — against the very position that is the single point of failure — and my own Score gates hardest on exactly that, at 35% weight. "Reduce without realising" can therefore trade a tax bill for a leverage problem I score more harshly than the concentration I was fixing. Directing new money is the only one of the four that escapes the trade, and it is the slowest. Position 2 governs how you reduce; this position governs whether you decided in advance. Where they genuinely conflict, the tax bill usually wins and the honest instruction is to reduce without realising.
On the universal number: that is a fair catch, and I should tier it rather than wave at it — the 20 and 30 thresholds are the published Score spec, while the 15- and 35-point deductions behind them are still v0 settings of mine awaiting calibration, so they are a house rule wearing a spec's clothes. They are also a different instrument from a personal ceiling: the Score's thresholds are a hazard screen that grades everyone the same way so the number is comparable, while a ceiling is a personal limit you set. I should not use "anyone quoting a universal one is selling" as a rhetorical move while shipping one; the Score's threshold is defensible as a common yardstick, and I would rather defend it than imply I have no number.
On concentration building fortunes: true, and different jobs — fortunes are made concentrated and preserved diversified, so pointing a preservation rule at someone still building is a category error I would rather name than defend. What I will not claim is that the ceiling protects returns. It protects against one holding being able to end the operation — a rule about surviving being wrong, not about being right.
Deep-dive: how much in one stock is too much, and whether you are diversified enough.
What would change my mind
None of these is held on faith. Here is the specific, checkable thing that would move me off each.
- Margin. A book governed to the Clear zone force-liquidated anyway on utilization that never left Clear. That breaks the dose discipline.
- Borrow-don't-sell. The edge disappearing once you weight it by the families whose collateral did get called. Then it's a survivorship illusion.
- DIY vs advisor. Governed DIY operators, as a group, blowing up at the same or a higher rate than clients of typical advisors. Then "a process replaces the person" is false.
- Realized-only income. A realized-only book routinely mispricing its liquidation risk badly enough to cause worse decisions than mark-to-market would. Then auditability costs more than it saves.
- The 4% rule. Dynamic-withdrawal and never-deplete frames underperforming a plain 4% across the sequences Bengen tested. Then "wrong frame" is just a preference.
- Weekly cadence. Quarterly-reviewed governed books surviving as well as weekly-reviewed ones. Then the cadence is ceremony.
- Family bank. Brokerage replicators abandoning or raiding the account at rates that erase the cost advantage over the contract — or lenders freezing the borrow-leg in a broad crisis. Then the premium’s coercion is the product, and I’m wrong about who can run it.
- Governance Score. Books that score well going on to fail at the same rate as books that score badly. Then the weights measure nothing and the number is decoration.
- Recommend-only. Readers who receive a clear recommendation ignoring it at rates that make the recommendation worthless — while rule-executing systems demonstrably protect the same people. Then I am selling the comfortable design, not the effective one.
- Tool over spreadsheet. Spreadsheet users sustaining a weekly cadence longer than tool users. Then the friction was the feature and I automated away the thing that was working.
- Concentration ceiling. Ceiling-enforced books ending up worse off, after tax and after the trimmed winners, than books that let the position run. Then the rule costs more than the failure it prevents.
What this page is not
This page is not advice, and not a recommendation set. It is deliberately not a hedge, either — I'm not walking back the pillar posts, I'm showing you the load they carry. It tells you where the honest disagreements sit on eleven questions credentialed people genuinely dispute. Deciding well means knowing the best argument on the other side — the entire job of this page.
Do this next
Read the counter-arguments above before you read anything else on this site — that is the actual next step. If you later want the weekly governance report this whole argument is about, the one that reads your margin zone and allocation drift from your own numbers, it arrives as the Weekly Governance Dispatch, with first access to the software and founding-cohort pricing. No urgency: it will still be here once the argument has settled in your own head.
This is education, not personalized financial, legal, or tax advice. I write as a practitioner running my own leveraged, multi-strategy book — not as a licensed adviser — and nothing here is a recommendation for your situation. Margin amplifies losses as well as gains, a broker can liquidate without contacting you, and leverage that's fine in a calm market can end you in a fast one. Tax and estate questions belong with a CPA and an attorney.
Incomestead recommends. You decide.
Last reviewed: 5 August 2026. This is a living document — the positions, counter-arguments, and what would change my mind are revised as the evidence and the pillars change.