A portfolio dashboard reports; a governance tool gates. That is the whole difference, and it is not a difference of quality. A dashboard shows what is true right now — balances, allocation, margin loan. An instrument takes the same reading, tests it against a written policy, checks whether the breach was also there last week, then either issues an instruction or says nothing. One describes. The other decides whether this is a week in which something has to happen.

I run a leveraged, income-oriented book myself, and I spent years with excellent dashboards. They were never wrong. They were also never responsible for anything. The number that eventually mattered — a leverage reading that sat in the same zone two Fridays running — was visible on every screen I owned and acted on by none of them. The book worked through below is constructed rather than lifted from my statements; the arithmetic and the rules are the real ones.

TL;DR — the short version

  • A dashboard reports; an instrument gates. A different job, not a better version of the same job.
  • Same data, two responses. One book reads 37.1%, then 37.8%. The dashboard displays both identically. The instrument treats the second completely differently.
  • The gate is the product. A breach must persist 2 consecutive weekly snapshots before it drives an action. Week one records; week two fires.
  • An instrument ranks. Leverage and allocation both out of line? Margin comes first — it is the existential one.
  • Run The Dashboard Test below on your own account this week. Six questions, ten minutes.

What does a portfolio dashboard actually do?

A portfolio dashboard aggregates positions, prices and balances into a current-state view: net liquidation value, allocation, margin loan, buying power, performance since a date you chose. Broker consoles and aggregators do it, mostly better than anything I could build.

So here is the distinction: a dashboard is accountable for accuracy; an instrument is accountable for a decision. If a dashboard shows a true number and you do nothing, it did its job. If an instrument shows a true number and issues no instruction when your policy required one, it failed.

A dashboard is accountable for accuracy. An instrument is accountable for a decision.

Same reading, two responses: week one

One leveraged book, Friday 10 July. Nothing ambiguous in the inputs:

Margin utilization = margin loan ÷ gross securities value

Week 1  ·  $1,020,000 ÷ $2,750,000 = 37.1%

Week 2  ·  $1,020,000 ÷ $2,695,000 = 37.8%

That is our house definition, and it should be stated as such. The phrase is not ours alone — some brokers do publish a figure called margin utilisation — but where they do it usually means margin reserved over collateral, a different denominator that reads far higher than loan over gross securities value. Same name, different ratio, which is precisely why the definition has to travel beside the thresholds. Loan over gross securities value is simply the denominator we chose, because it is the one both inputs of which appear unambiguously on a statement. Another perfectly reasonable denominator gives a wildly different answer on this same book: the same loan over net liquidation value reads 59.0%. Our 37.1% and that 59.0% sit 21.9 points apart. Same portfolio, same Friday, two numbers. A band is meaningless unless you know which denominator produced it.

The dashboard. Net liquidation value $1,730,000, margin loan $1,020,000, an allocation ring, and — if it is a good one — 37.1% tinted amber. All true. Then it waits for you.

The instrument. It reads 37.1% against a written band table — the four zones are the ones set out in my note on using margin safely; the stricter convention this piece runs on, the gross-market-value denominator and the inclusive/exclusive boundaries, is stated here — places it in Freeze, writes an immutable snapshot, records a first breach — then deliberately tells you to do nothing. Not because the reading is acceptable, but because one reading is not evidence. This is the two-week persistence rule: a breach must persist across 2 consecutive weekly snapshots before it drives an action. Week one is allowed to be noise.

One thing about that rule has to be said out loud, because it is the cost side of the trade. The rule applies with one deliberate exception — Forced, at 40% and above, acts on the first reading — and below that line, waiting is not free. If a breach is rising fast, a week of deliberate silence is a week of real exposure, and the rule will occasionally look foolish in hindsight. That is accepted deliberately, traded against the larger risk of acting on noise every time a Friday close moves against you. It is a judgement, not a proof, and it is mine.

Zone Utilization This book What the instrument does
Clear below 30% — Records the week. Says nothing.
Harvest 30% up to but not including 35% — Flags. New leverage stops being routine.
Freeze 35% up to but not including 40% 37.1%  →  37.8%
Week 1  ·  Week 2		 Week 1 records a breach. Week 2 fires the gate.
Forced 40% and above — Reduction is the only recommendation on the page — and unlike Freeze, it does not wait for a second week.

Every boundary is stated once and only once, so no reading is ambiguous: a book at exactly 35.00% is in Freeze, and a book at exactly 40.00% is in Forced, not Freeze.

These are my own operating numbers — not a regulatory standard, not an academic one, and not a recommendation for your book. The instrument enforces the band; it has no opinion on whether the band is wise, and a badly chosen threshold applied perfectly is still badly chosen. Choosing the number is the part you cannot outsource.

Note what has diverged on identical data. The dashboard produced a colour. The instrument produced a record — a dated reading next week can be compared against. A display that overwrites itself cannot hold one — and reconstructing why you sold eighteen months later is a question that now has its own post.

Week two: does the gate fire?

Friday 17 July. The loan has not moved. Holdings are down 2% to $2,695,000. Utilization is $1,020,000 ÷ $2,695,000 = 37.8%. Still Freeze.

The dashboard is unchanged. Same layout, same amber tint, a slightly different number. It has no view about this being the second, because it never kept the first.

The instrument changes completely. Two consecutive snapshots in breach satisfies the two-week persistence rule, and the gate fires. It solves for the reduction that brings utilization back below 35%, into Harvest — the arithmetic is worked in full in the margin-zones guide — giving an exact solve of $118,076.92 (118,076.923…). It then rounds up to the next $1,000, away from the band edge, and verifies that the resulting reading still rounds strictly below the edge at display precision.

Both halves of that rule earn their keep. Round down to a tidy $118,000 and the reading is 35.0019% — still Freeze, still breached. Round merely to the nearest hundred and you clear the exact solve by $23.08, yet the book still displays as 35.00% at the two decimals this card uses: out on the arithmetic, sitting on the edge on the page. A threshold you can land exactly on is not a threshold. So the recommendation is $119,000, at which the book reads 34.98% — $923.08 of headroom over the exact solve, rounded away from the edge by enough that ordinary transaction costs do not push the reading back inside. Then the instrument does what a display structurally cannot. It names the options.

Recommendation: reduce the margin loan by $119,000. Top three options, ranked by margin impact.

Position Value Proceeds Margin impact
Long-Horizon Growth · holding A (1,050 sh) $164,850 $164,850 37.85% → 33.80%
Income Engine · holding B (2,400 sh) $141,600 $141,600 37.85% → 34.40%
Short-Horizon Growth · holding C (3,150 sh) $121,275 $121,275 37.85% → 34.92%

Proceeds shown are gross. Tax, commissions and spreads are not modelled — tax estimation is a deliberate exclusion, not an oversight — and all three will reduce what actually reaches the loan.

Two honest caveats about that ranking. Under the formula above, margin impact is monotone in proceeds, so "ranked by margin impact" is, mathematically, just "largest position first" — the ranking earns its keep by attaching the impact figure to each line, not by reordering anything. And the relief is modelled dollar-for-dollar: a dollar of proceeds repays a dollar of loan. Your broker does not work that way. Real relief depends on each position's maintenance requirement, so selling a concentrated or low-marginability holding frees materially more room than selling a broad ETF of the same value. Treat these as governance arithmetic against your own policy, then check the actual effect in your broker's own numbers before you act.

Three options, each achieving the goal, each with the four facts you need to choose between them — and no fourth, because a list long enough to browse defers the decision back to you while pretending to help. Which you take is yours. The instrument does not trade.

Now change one figure and it inverts. Had holdings risen to $3,020,000, utilization would be 33.8% — Harvest. The dashboard shows a greener tile. The instrument shows more: the breach did not persist, the gate does not fire, and no reduction is recommended. The Harvest flag still stands — 33.8% is in Harvest, so new leverage stops being routine — but last week's alarm is now noise, on the record.

Half the value of a gate is the weeks it stays shut. A tool that cannot tell you to do nothing will eventually talk you into doing something.

None of which makes leverage safe. The downside travels with the upside: margin amplifies losses as fast as gains, and your broker's house requirement is a separate ruler from your policy's. A maintenance requirement is the minimum equity you must keep against a margined position: FINRA's floor is 25% of market value and house requirements are typically higher. That is the only threshold in this piece that can forcibly liquidate you — mine cannot. FINRA notes that firms "may increase the house requirements at any time and aren't required to provide you with advanced written notice." A gate at 35% exists so it bites while you still have three options. Be clear about what this instrument is, though: a weekly reading is a governance cadence, not a margin-call defence. A Friday-to-Friday instrument will never see the Tuesday move that liquidates you — that is your broker's console and your own attention, and nothing here replaces either.

Which comes first when two things are wrong at once?

Suppose that in the same week two, one strategy Layer — a Layer being one strategy in the Incomestead Stack, the fixed set of strategies a book is divided into — has drifted 4.2 points above its upper band just as the margin gate fires. A dashboard renders two red tiles of equal size, because priority is a policy question and a display holds no policy.

An instrument holds one, and the ordering is not taste: margin first, then allocation. Being wrong about allocation costs return; being wrong about leverage can cost you the choice of when you sell. So the margin instruction ranks above the drift instruction — which still appears, with per-position buy and sell targets to the midpoint between the target weight and the band edge that was breached — overweight, halfway back from the upper band to target; underweight, halfway up from the lower band to target. Not the middle of the band, and not all the way to target: enough to clear the edge with room, without churning the book to hit a number exactly. Both are ranked but independent; sequencing is a decision, and decisions are yours.

Our own scoring reflects that ordering rather than proving it. In the Governance Score, margin does not merely contribute — it caps. A book in Freeze cannot score above 65 however immaculate everything else is, and a book in Forced is capped at 34, the bottom band, Exposed / At the Edge. The component weights, the full band ladder and the method's limits live on that page.

The Dashboard Test

Run this against the tool you actually use, on your own account, this week. Ten minutes. It is not a grading scale for dashboards — it is a detector for governance features, and most of those features are ones a dashboard was never built to have. The first question is one a good dashboard passes outright and a governance tool can fail. The point is to find out precisely which jobs your tool is doing and which nobody is doing.

THE DASHBOARD TEST — SIX QUESTIONS FOR THE TOOL YOU ALREADY USE

  1. The completeness test. Does every account, holding and cash balance you own appear in one place, and did today's total tie to your broker without you correcting it? A good dashboard passes this cleanly, and a governance tool fed one broker export can easily fail it. Coverage is the job dashboards do best.
  2. The instruction test. Open it now and write down, in one sentence, what it told you to do this week. If you can only write down what you hold, it reported.
  3. The threshold test. Point to where the number separating "fine" from "act" is stored. If it is not written outside the display, you do not have a threshold — you have a feeling.
  4. The last-week test. Do not ask it for a ratio: a console may well show one called margin utilisation, computed on a different denominator than the one used here. Ask instead for the two raw inputs, as they stood exactly seven days ago: your margin loan balance, and your gross securities value. If the tool cannot hand you last Friday's pair, it cannot measure persistence, because it never kept the evidence.
  5. The precedence test. If leverage and allocation were both out of line this morning, which does it tell you to address first? Two red tiles is not an answer.
  6. The silence test. In a week where nothing is out of band, does it say "nothing to do this week" in as many words? A tool that cannot say clear is not gating. It is decorating.

What a dashboard is still better at

Breadth, speed, intraday, every account a governance engine does not read, and exploration — poking at a holding, sorting by something you invented thirty seconds ago. Keep yours. Governance replaces nothing; it is a second, slower instrument that runs once a week. The weekly cadence is itself load-bearing — a gate that only fires when you remember to look is not a gate.

Nor does an instrument decide better than you. As said beside the band table, it enforces your thresholds without any view on whether they are the right ones. It does not forecast and it does not trade. You can run the discipline by hand, too — a well-built spreadsheet can do this arithmetic, where the costs are convention, record and succession rather than capability. So this is not an argument that you need software, but that reporting and gating are two different jobs and most people own a tool only for the first. The wider case sits in the operating layer family offices run and solo investors skip.

What happens if we disappear?

The vendor question, which I would rather raise than have you find later. Incomestead is early. Software companies get acquired, change direction, or fold, and a governance record locked inside a vendor is a record you do not really own — the discipline exists to remove concentration risk, not to introduce a new one at the filing cabinet.

So two commitments matter: portability and publication. Your snapshots and your written policy should be yours, exportable in a format you can open in ten years without us. And the method should be public — which is why the weights, bands, zone boundaries and persistence rule are on this site for free rather than behind a login. The test is portability of the record: if this company were gone on Monday, the file of weekly snapshots would still open on your own machine, in your own format, and would still be a usable governance history for whoever inherits it.

If the difference has landed, the next step is not to buy anything. Run The Dashboard Test before Friday and see which of the six your tool passes. For years I could not have answered the third one — I had no threshold written down anywhere. That is a ten-minute problem to fix, with or without us.

Do this next

Run the six questions first. Then, if what you found was that nobody is doing the gating, the Dispatch is where that job gets done in writing each week — the Weekly Governance Dispatch, plus first access to the software when it opens to early users, and founding-cohort pricing.

Join the founding cohort →

Frequently asked questions

What's the difference between a portfolio dashboard and a governance tool?
A dashboard reports current state; a governance tool tests that same reading against a written policy, checks whether the breach existed in the previous weekly snapshot, then either issues a ranked instruction or says there is nothing to do. The dashboard is accountable for accuracy; the instrument is accountable for a decision.

Is a portfolio dashboard enough if I run a leveraged portfolio?
It is enough to see your leverage; not enough to govern it. Governing needs three things a display does not hold: a threshold written outside the tool, a record of last week's reading so persistence can be measured, and a rule about which problem ranks first. Only a record knows 37.8% is the second reading, not the first.

Does a governance tool tell me what to buy and sell?
It recommends and ranks; it never executes. When a margin gate fires it states one high-level reduction — "reduce the loan by $119,000" — plus exactly three liquidation options ranked by margin impact, each with position, value, proceeds and impact. Reductions are rounded up to the next $1,000, away from the band edge, with enough headroom that ordinary transaction costs do not push the reading back inside the zone it was solving to escape. When allocation is out of band it gives per-position buy and sell targets to the midpoint between the target weight and the breached band edge. What you do is yours.

Why does a breach have to last two weeks before anything happens?
Because a single reading is noise. Prices move, cash settles, a dividend lands, and any threshold gets crossed and uncrossed by ordinary drift. Requiring the breach to persist across two consecutive weekly snapshots filters that out — measurable only if last week's reading was recorded on the same basis.

Written by Stefano Starkel, founder of Incomestead, from lived experience running a leveraged, income-oriented portfolio — as a practitioner, not a licensed adviser. This is education, not personalized advice; figures are illustrative. Margin amplifies losses as well as gains, a broker can liquidate without contacting you, and house requirements can be raised without notice. Consider your own situation and, where appropriate, a professional who knows it.

Incomestead recommends. You decide.