The CompoundWise Manual

Everything the app does, explained in plain language — what each screen is for, how it works, and how to get the most out of it, step by step.

The one promise behind everything here: CompoundWise never makes a number up. Every figure you see — a score, a fair value, a tax amount, a retirement projection — is calculated from real data with rules you can read, or it is clearly labelled as an estimate with its assumptions shown. When the app doesn't know something, it says so instead of guessing.

And the one disclaimer: CompoundWise is a research and education tool, not a licensed advisor. Labels like "Buy" or "Sell" describe what the model sees, not instructions. Always do your own thinking, and talk to a professional for personal advice.

Table of contents

1. What is CompoundWise?
2. Getting started — your first 15 minutes
3. Today — your daily brief
4. Overview — your money at a glance
5. True net worth — the full picture
6. Screener — finding great companies
7. Emerging Compounders — catching the inflection early
8. The stock page — deciding on one company
9. Position sizing — how much to buy
10. Compare — head to head
11. Portfolio — what you own
12. Am I beating the index? — the honest scorecard
13. Asset mix & your Investment Policy
14. The risk quiz — what mix suits you 14b. Your plan — what you can see
15. The Advisor — chat and AI reports
16. Sector Lab — mastering one industry at a time
17. Tax Centre — keeping more of what you make
18. Goals
19. Retirement income — will the money last?
20. Rebalance & stress test
21. Risk X-Ray — how concentrated you really are
22. Alerts, watchlist & thesis guards
23. Settings, household & security
24. The fine print — what the app deliberately does NOT do

1. What is CompoundWise?

CompoundWise is a personal investing co-pilot built around one philosophy: buy good businesses at sensible prices, hold them for years, and don't lose money to taxes and mistakes along the way.

It does four jobs for you:

2. Getting started — your first 15 minutes

Step 1 — Log in and look around

After logging in you land on the Overview. Nothing needs to be configured to explore — every screen works, it just gets personal once your holdings are in.

Step 2 — Get your holdings in (pick ONE of these)

• Connect Questrade (Portfolio page → Questrade button): a read-only connection. The app can see your accounts and trades; it can never place an order or move money.
• Another broker (Settings → Accounts & Data), if enabled on your account.
• Import a file (Portfolio → Import): download your transaction history from your broker as CSV/Excel and drop it in. The app reads dates, tickers, buys and sells.
• Type them in (Portfolio → Add transaction): fine for a handful of positions.

Why transactions and not just holdings? Because your transactions are what determine your ACB (adjusted cost base — what the CRA considers you paid). Get the transactions right and every downstream number — gains, taxes, performance — is right automatically.

Keep a copy. Settings → Accounts & data → Download my ledger (Android: Settings → Download my ledger) gives you every transaction, your holdings with their cost base, realized gains by year and the history of every edit as one CSV. Every change to a transaction — yours, or an exchange rate the system fills in later — is kept with its old and new value: press the clock icon on any row of Portfolio → Transaction history, or Change log for all of them, deletions included. Prices show as of when they were observed, live for a fresh quote in market hours, and stale in amber once a newer trading session has opened.

Step 3 — Tell the app who you are (2 minutes, big payoff)

Go to Settings → Profile and fill in: - Province and income band → the tax math uses the combined marginal rate for your province and income band, taken at the band's midpoint - Birth year and target retirement age → unlocks the Retirement planner - Annual contribution (roughly what you add per year) → makes goal math realistic

Step 4 — Take the tours

The first time you open any screen your plan includes, a short welcome box explains what it does — a couple of them (Rebalance, Risk X-Ray) with a small interactive demo you can drag. Read it, then press Got it to retire it for good, or Later (or the corner ×) to see it again in a week. The two demos have a Reset to put the slider back, and Open in a window if you want them on their own. Want an intro back any time? Open the ? menu at the top right and choose Show this screen's intro again.

Step 5 — Let Home keep suggesting the next thing

Home shows a Get started checklist for your first days and, once that is done, a Next steps panel that keeps suggesting the next thing you have not tried, in plan order. If it shows a locked tile, that is the cheapest step up from your plan — opening it tells you what that feature would say about your own book.

A note on the two currencies

The CAD/USD toggle at the top switches what currency amounts are displayed in. It never changes your stored records — your cost base stays in the currency you actually paid.

3. Today — your daily brief

What it's for: answering, in under ten seconds, the only two questions that matter daily: "Is my money okay?" and "Is there anything I should actually do?"

The health ring (top left)

Think of it as your portfolio's pulse: - Calm green, slow breathing — nothing needs you. Enjoy your day. - Amber, gentle pulse — a few things are worth reviewing. - The number inside is exactly how many decision cards are below. No hidden backlog: if more exist than fit, it says "+N more waiting."

The market pulse — the weather, made personal

Right under the health ring sits the day's market read, in two layers:

• The tape — small chips for the S&P 500, Nasdaq and TSX (green up, red down), the CAD/USD rate with an arrow if the loonie is moving, the VIX with a plain word for it ("calm" or "jumpy" — the VIX is the market's fear gauge), and the Bank of Canada rate.
• The story of the day — one sentence, with its source named, on why the market moved ("Stocks slipped after US inflation surprised higher — Reuters"). It describes; it never predicts. On genuinely wild days it refreshes within the hour instead of waiting for the daily cycle.
• "For your portfolio" — this is the part no generic news feed gives you. Up to three sentences that cross today's market with your actual holdings, each derived from a transparent rule and using your own numbers. Examples of what can appear:
• "76% of your money is in US dollars and the loonie is firming — every 1% the CAD rises trims about $1,125 of your CAD value. Over years this washes out; the USD tilt is a choice, not a flaw."
• "Markets are jumpy (VIX 28). Your portfolio moves about 0.9× the market, so a 1% market day swings roughly ±$1,347 of your value. That's noise, not information — unless a business itself changed."
• "Technology is facing a macro headwind right now, and it's 38% of your portfolio — it's the day to re-read the thesis on your biggest tech names."
• If a stock you own is in today's news, you'll see it flagged by name with a one-line why and a link to its page.
• A Big picture line above "Coming up" tells you where the market cycle sits (e.g. "Late Cycle / Caution") and which way conditions are trending — the long-lens context for everything above.

On a genuinely quiet day it says so plainly: "Nothing in today's tape changes your plan." And if market data is ever unavailable, the pulse simply shows less — it never guesses.

Decision cards

Each card is one decision, fully self-contained: - A coloured verb — SELL (red), TRIM (amber), ADD (green), RELOCATE (cyan for tax moves) - The ticker and account it applies to, and the source it came from (e.g. "Rule engine" or "AI Tax & Structure Audit") - One plain-language sentence of why — e.g. "Debt/Equity of 3.5x combined with Deep Score 32 indicates balance sheet stress." - An Act button that jumps you straight into the right next step — a pre-filled trade form, the Tax Centre, or the Rebalance tool. You never hunt for where to do the thing.

Where do these cards come from? Five sources feed one list: the 27-rule signal engine (checks every holding against fundamentals and policy — the only source of buy, sell, trim and add cards), the AI Tax & Structure Audit (structural moves only — migrate, consolidate, harvest — labelled as the audit's), your goals (if one falls behind), your concentration limit (if one stock passes the single-stock limit you declared, else the 25% house limit), and your Investment Policy (if your mix drifts outside its band). Duplicates are merged, and if one rule fires on five holdings at once (say, "too much USD"), you get one card naming all five — decisions, not an echo.

Three things to know about the cards

1. Dismiss = snooze, not mute. The little × hides a card for 30 days. If the problem still exists after that, the card comes back. You can't accidentally silence a real issue forever.
2. Cards clear themselves. Make the suggested trade (or any matching trade) and the card checks itself off — the app watches your transactions.
3. Nothing is ever executed for you. Every "Act" lands on a review screen where you confirm.

Coming up & Working well

Below the decisions: real dates for your holdings (earnings reports, dividend ex-dates for the next 14 days — click one to jump to that stock's page), plus money nudges (idle cash, unused TFSA/RRSP room). And on a good day, Working well tells you what's right — a calm day never looks like a broken screen.

Tip: if a holding reports earnings tomorrow, the app also sends you a Telegram heads-up the evening before (if you've connected Telegram in Settings → Alerts).

4. Overview — your money at a glance

What it's for: the numbers dashboard. Today answers "what should I do?"; Overview answers "where do I stand?"

Top to bottom: - Invested assets — your holdings plus account cash, with total gain. It's deliberately not called "net worth" — that word is earned one section down. - Holdings vs cash strip — tap it to see value split by account. - Your top-3 decisions — a compact version of Today, so the most important actions greet you at login. See all → opens the full brief. - KPI tiles — number of holdings, idle cash, portfolio beta (how bumpy your ride is vs the market: 1.0 = same as market, lower = steadier), USD exposure, best & worst performer. - Sector allocation & top positions — where your money is concentrated, at a glance. - Goal progress and quick links.

5. True net worth — the full picture

What it's for: your investments aren't your whole financial life. Your house, your mortgage, savings at another bank, a pension — they all belong in one honest number.

How to use it

1. On the Overview, find the strip under the big number ("See your true net worth…").
2. Tap it and add items: choose a type (Real estate, Mortgage, Cash elsewhere, GIC, Pension, Loan…), give it a name and a value in CAD.
3. That's it. The strip now shows the honest equation: Invested + Other assets − What you owe = Net worth.

Three design choices that protect you

• You never enter minus signs. You pick the type — "Mortgage" automatically subtracts. A typo can't flip your net worth.
• These entries never touch your investing analytics. Your stock scores, signals and performance are unaffected — with two helpful exceptions: cash and GIC entries do count in your asset mix (section 12), because they're genuinely investable money.
• It's private to you — household members don't see each other's manual entries.

6. Screener — finding great companies

What it's for: hunting. Roughly 2,900 US and Canadian stocks, each given a Deep Score out of 100, refreshed four times each weekday.

What the Deep Score means (in plain words)

The score measures business quality, deliberately ignoring the share price: - Profitability (30 pts) — does the company make good money on the money it uses? - Growth (25 pts) — are sales and profits actually growing, year after year? - Financial strength (20 pts) — could it survive a bad recession without begging for cash? - Valuation (15 pts) — is the price at least sensible relative to earnings and cash? - Moat (10 pts) — signs of a durable advantage competitors can't easily copy.

85+ is exceptional, 70+ is strong, 55–69 is decent, below 40 the app says avoid.

How to use the screener well

1. Start with the map (the big chart): each bubble is a company, plotted cheap↔expensive against weak↔strong. The magic corner is top-left: high quality at a low price — the app calls these compounders and lists the best six under the chart.
2. Filter down: by market (US/Canada), sector, or with the Advanced Filters (e.g. "ROE above 15% AND debt below 0.5").
3. Keep "Hide multi-year decliners" ON (it's on by default). A stock can score well on fundamentals while its price has fallen for three straight years — that pattern is called a value trap, and this switch keeps them out of sight.
4. Customize your columns (⚙ Columns) — 40+ metrics available; your layout is remembered.
5. Click any row to open the full stock page.

The ETFs tab works the same way for ~50 curated funds, scored on cost, long-term returns, size and income — and the ranking adapts to whether your profile says growth or income.

7. Emerging Compounders — catching the inflection early

What it's for: a forward-looking companion to the Deep Score. The Screener rewards businesses that are already great; this pane hunts for the ones just starting to accelerate — before the trailing numbers catch up.

Two different questions

The Deep Score grades proven quality — by design a rear-view mirror: a company only earns a high score after years of good results. The Inflection Score looks at the rate of change instead. Deep Score asks "is this a great business today?"; the Inflection Score asks "is this business inflecting into a great one?" A name can be low on one and high on the other — the table shows both side by side.

The eleven signals (Inflection Score out of 100)

Each is measured as a trend — recent versus earlier — not as a level. Since 2026-09-28 (engine 3.0.0) six carry weight, each scored as the company's percentile against its sector peers and weighted by what the historical test showed; five are shown as facts with no weight: - Share dilution (28) — whether growth is funded by issuing new stock. The strongest signal in the historical test: buybacks and flat share counts rank high, dilution ranks low, and heavy sustained dilution caps the label. - Margin expansion (23) — gross and net margins widening as the business scales, ranked against sector peers. - Operating leverage (14) — profits climbing faster than sales. - Improving returns (13) — return on invested capital trending up, not just already high. - Earnings surprise (12) — the latest quarter's earnings against the same quarter a year ago, scaled by how much that change usually varies, read from the company's own quarterly filings. - Cash-flow inflection (10) — the free-cash-flow margin trend: real cash backing the growth, not just accounting profit. - Growth acceleration (shown, no weight) — growing faster than a year or two ago. On the historical test it did not separate winners from losers, so it is displayed but not counted. - Early price breakout (shown, no weight) — the market beginning to notice. No point-in-time history exists to test it. - Insider buying (shown, no weight) — on the historical test, companies whose insiders were net buyers went on to lose; buying is a hint to investigate, not a point. - Forward growth (shown, no weight) — next fiscal year's analyst-consensus growth; untestable point in time, so shown but not counted. - Estimate revisions (shown, no weight) — which way analysts are moving their forecasts; untestable point in time, so shown but not counted.

Tested and left out (29 Sep 2026). Eight "stability" signals from the accounting research — steady quarterly earnings, steady sales growth, profitability, cash-flow profitability, cash flow above reported profit, and spending on research, equipment and advertising — were added to the score on the same historical test, with the pass mark fixed in advance. They made the score look stronger on average but much less consistent from one period to the next, so it failed the certification rule; and most of what they added is what the Winners list already measures. Emerging Compounders stays the "what is changing" tool and Winners the "what is proven" tool.

Insider buying, given its best chance (29 Sep 2026). Researchers have shown that most insider trades are routine — the same person, the same month, every year — and only the unusual ones carry information. We applied their filter to every officer and director filing. It helped, but companies whose insiders were net buyers still did worse over three years than companies whose insiders were net sellers. Insider buying stays a fact you can see and look into, never a point in the score.

Bands: Inflecting 80+ · Emerging 65–79 · Building 50–64 · Early/mixed 35–49 · No inflection below 35. Because every factor is a trend, a mature giant with high-but-flat numbers scores near zero here even when its Deep Score is excellent — the two lenses are meant to disagree.

How to use it

1. Read it next to the Deep Score, never instead of it. High inflection + a modest Deep Score is a young business worth researching; high on both is rarer and stronger.
2. Filter by market, sector, minimum score and confidence.
3. Watch the confidence chip — companies with less financial history score fewer of the eleven factors.
4. Open any name to see its factor radar and revenue-acceleration curve on the stock page.

Early means uncertain. An inflection is a possibility, not a prediction. Most fast-accelerating companies never become the next great compounder, and early-stage names swing harder. Treat it as a research starting point and let position sizes reflect the risk. Every decision is yours.

8. The stock page — deciding on one company

What it's for: everything you need to decide about ONE company — organized so a beginner isn't buried, and an expert misses nothing.

Essentials view (what you see first)

The page opens in Essentials — four questions, answered top to bottom: 1. What is it? Name, sector, price, and the verdict badge (Strong Buy / Buy / Hold / Avoid — the model's read, not an instruction). 2. Is it good? The Deep Score with its breakdown, and a quality radar. 3. Is it cheap? The fair value (what 14 different valuation models collectively estimate the business is worth per share) and the price-vs-fair-value thermometer. If the fair value isn't trustworthy for this company, the app says "no reliable estimate" instead of showing a shaky number. It says why in the same place, and the fair value is re-worked every time the price refreshes, so it always matches the price you are looking at — which also means that, for a few companies, it can appear or disappear during the day as the price moves. 4. What do I do? Decision tools — including the suggested position size (next section) and a quality checklist. A company the framework cannot measure — a pre-revenue biotech, a pre-production miner, a fresh listing — shows no Deep Score and no checklist, rather than a middling score built on the half of it we cannot see.

Full analysis (one click away)

Press Full analysis (top right) or the "Show the full analysis" band and seven more sections appear: company profile & news, complete valuation workbench (interactive DCF you can play with), growth and analyst forecasts, ten years of financial history, balance-sheet health, dividends, and insider/institutional ownership. Your choice sticks — pick Full once and the page always opens that way for you.

Reading the price banner

A yellow banner saying "Price −40% over 3 yrs, still near its 3-year low" is the value-trap warning: the fundamentals may look fine, but the market has been disagreeing for years. Confirm the story before buying. A green banner means the price trend agrees with the quality.

9. Position sizing — how much to buy

What it's for: the question every tool ignores. Finding a good stock is half the job — the other half is how much of your portfolio it deserves.

The rule, in plain words (no AI, no black box)

1. Quality sets the tier. Deep Score 85+ → high conviction (suggested 8–12% of your portfolio). 70–84 → solid (4–8%). 55–69 → moderate (2–4%). 40–54 → starter (0–2%). Below 40 → avoid (0%).
2. A genuine bargain earns one step up — but only one. If the stock trades 25%+ below a fair value the models are confident about, the tier rises one notch. If the fair value is shaky, it moves nothing — an untrusted number should never size a position.
3. Paying above fair value steps you down one.
4. A weak business never gets promoted for being cheap. A 60% "discount" on a bad company is the value-trap signature, not an opportunity.

Every step of the reasoning is printed next to the suggestion. Nothing caps above 12% — well inside the 25% house single-stock limit. (Declare your own limit in your profile and that is the one the app polices instead.)

On a holding you already own, the same card compares the band to your actual weight: within the band / room to add / above the band.

10. Compare — head to head

What it's for: you've narrowed it to two or three candidates. Put them side by side.

Type up to three tickers and press Compare: quality, valuation, growth and safety metrics line up column by column, the better number highlighted in each row, with a ranked verdict and each stock's bear/base/bull price range. Click through to any stock page from here.

11. Portfolio — what you own

What it's for: the source of truth. Every other screen builds on what's recorded here.

The pieces

• Account tabs — All, TFSA, RRSP, Margin, etc. Each shows its own value, gain, and stats.
• The holdings table — every position with live price, value, gain, weight, Deep Score, and a signal badge (ADD/TRIM/SELL/HOLD from the rule engine). Click any row for that holding's own page — price chart, quality radar, sizing verdict, and your conviction log.
• The growth chart — your actual value over time, rebuilt from your transactions and historical prices. Toggle between total value and profit.
• Transactions — the ledger. Every buy, sell and dividend, editable. This is what drives your tax math.

Recording a trade properly

Press Add transaction, and the form asks for ticker, buy/sell, shares, price, account, date, and optional commission. Two things happen automatically: - Your ACB (cost base) updates the way the CRA requires — including the tricky cases like selling part of a position, or the superficial loss rule (sell at a loss and rebuy within 30 days → the CRA denies the loss; the app flags it so you're not surprised at tax time). - Any Today card or AI recommendation that suggested this trade checks itself off.

If your numbers drift from your broker's

Portfolio → Questrade → Reconcile does a live, read-only comparison against your actual brokerage account and shows any differences for you to approve.

12. Am I beating the index? — the honest scorecard

What it's for: the question most tools hide from, answered three ways on the Portfolio page.

The three numbers, in plain words

• Your return — what your actual dollars earned, counting when you invested them. (The technical name is money-weighted return. It's the number that matches how your account feels.)
• The twin — the killer feature. The app replays every buy and sell you ever made, dollar for dollar, on the same dates — into a boring index fund instead (your own policy mix when you have set one, or XEQT or SPY via the toggle). The twin had your exact timing and zero stock-picking. The gap — your holdings today, plus your dividends counted as if reinvested in the index the day you got them, minus the twin today — measures how far ahead or behind the index you are, not your skill: over a short history, most of a gap is luck. "You're ahead by $12,400" or "behind by $18,046" — either way, now you know.
• The skill curve — the chart strips out deposit timing entirely and plots what your holdings did against the index, both starting at 100 — a difference, not a measure of skill.
• Returns by period — a table of each calendar year and the last 1, 3, 5 and 10 years, you against the benchmark on the same basis. A part-year is shown for the period, never as a yearly rate.
• Your account returns — statement basis — choose an account (or the whole book) and a period end (today, or 31 Dec to match your broker's annual report). This is the number to compare with your broker's annual performance report: it includes cash, counts deposits and transfers at market value, and is after commissions, account fees and interest. If an account says not measured, the cash the app rebuilt from your transactions does not match the balance on record, so it shows no number rather than a wrong one.

Why you can trust it

The small print on the panel isn't decoration: holdings only, total-return on both sides — dividends reinvested the same way for both, so the comparison stays fair — and every formula is locked by automated tests against hand-computed answers. If the app can't compute it honestly (say, you have under a month of history), it says "not enough history" instead of guessing. Under a year, nothing is annualised: you see the cumulative return for the period. A benchmark that launched after your book began (XEQT, August 2019) is not compared at all, and SPY is drawn in Canadian dollars, like your book.

How to actually use this: check it quarterly, not daily. If the twin beats you for two or three straight years, that's real information — the honest response might be indexing more and picking less. The app will never hide that from you; that's the point.

13. Asset mix & your Investment Policy

Also on the Portfolio page — What your funds cost a year. An estimate, in dollars, of the management fees inside the funds you hold: each fund's value today times its MER, per fund and per account, and as a share of your whole book (shares and cash count at 0%). A fund whose MER is not on file is named and any total it belongs to shows "—". The estimate leaves out the trading expense ratio (TER), so your dealer's annual cost report — which from 2026 must show your total fund expenses in dollars — will be somewhat higher. The same fee is deducted from your retirement and goal projections.

What it's for: the decision above stock picking: how much of your investable money is in stocks vs bonds/GICs vs cash — and what you want it to be.

How it works

• The app classifies your investable money — every holding (funds are sorted into equity vs fixed income by their category), your idle account cash, plus any cash/GIC entries from Net Worth. Your house and pension stay out — you can't rebalance a house.
• You set a target mix (say 80% equity / 15% fixed income / 5% cash) and a drift band (±5 points by default). This pair is your Investment Policy — the same discipline professional advisors write down for clients.
• Your cash reserve is not part of the mix. If you've entered a cash reserve in your profile (Liquidity & Safety), that amount is taken out of cash before the percentages are worked out, and the panel says how much was left out.
• Each class shows your actual percentage as a bar, your target as a tick, and the band as a shaded zone. Drift outside the band → a red DRIFT tag, and a single card appears on Today.

Why the band matters (the discipline part)

Without a band, you'd be nudged to rebalance every time the market breathes. With a band, the app is silent until drift is meaningfully outside policy — then it speaks once. Sitting exactly on the edge counts as inside. Rebalance discipline without rebalance noise.

Honest detail: fund category data is sometimes junk. Anything unrecognizable is counted as equity and disclosed on the panel ("$X in funds assumed equity") — never silently classified.

14. The risk quiz — what mix suits you

What it's for: most people don't know whether they're "80/15/5" material. Five questions produce a starting point.

Find it on the Portfolio page under the Asset mix panel: "Not sure? Take the 5-question quiz."

The five questions

When you'll need the money · what you'd actually do in a 30% crash · how stable your income is · whether you've lived through a crash · what this money's job is.

The two safety rules built in (this is the important part)

1. Your timeline overrules your bravery. Answer every question like a daredevil, but if you need the money within 3 years, the suggestion caps at Conservative (45/50/5). Money needed soon can't ride out a long bear market — no appetite changes that arithmetic.
2. The crash answer overrules everything above Balanced. If you'd sell everything in a panic, the suggestion caps at Balanced (60/35/5) — because the mix you can actually hold through a crash beats the mix that looked best on paper. Selling at the bottom does more damage than any allocation can repair.

Every applied cap is named in the result ("Capped at conservative — money needed within 3 years…"). And nothing changes until you press Apply to my policy — the quiz suggests, you decide. Retake it whenever life changes.

14b. Your plan — what you can see

CompoundWise comes in four plans. What appears in your sidebar depends on which one you are on.

Three pairs are easy to confuse, so they are worth naming:

• The Tax Centre (the screen, with your contribution room, harvest candidates and withholding drag) is in Basic. The Tax Audit — the written AI report — starts at Standard.
• Winner Odds is a calculated ranking, no AI involved. The Winner Thesis is the written bull/bear case, and that is an AI run.
• Reading a Sector Lab report is Pro. Generating one is Premium, because each report costs real money to produce.

If you hit something outside your plan, the screen tells you which plan includes it and, where it can, what that feature would say about your own book today — how many asset classes sit outside your policy band, or how far your book fell through a past crisis at today's weights. If your plan lapsed, it shows your own last report instead. For the AI tools it shows an example, clearly labelled as not your data, with the rough cost of a run. When it cannot say anything honestly — no targets set, a holding with no price yet — it says that and where to fix it, never a zero. It then offers a Request access button. That request goes straight to the account owner, who can turn that one thing on for you without changing anyone else's plan.

15. The Advisor — chat and AI reports

What it's for: one place for everything the AI does — instead of five scattered buttons.

The layout

• One budget meter (top): every AI action costs real money (cents); this shows exactly what's been spent this month. No silent billing, ever.
• The advisor's desk (three cards): the latest Strategy review (a full portfolio health report with grade), the latest Tax & Structure audit (grade + date), and Sector theses. Each card shows its status and links to the full report.
• The chat (centre): ask anything about your own portfolio in plain language.

What makes this chat different from ChatGPT

It is grounded: the AI can only read numbers the app has already computed — your holdings, scores, fair values, tax figures, goals. It cannot invent a price, browse the web, or see anyone else's data. Ask "am I too concentrated?" and the answer cites your actual weights.

Good starter questions: - "What's my biggest concentration risk right now?" - "Where should I hold my US dividend stocks for tax efficiency?" - "Which holdings look expensive versus their fair value?"

Trade drafts

If the chat suggests a trade, it appears as a proposal card — nothing is recorded until you press Review & Confirm and approve it in the standard trade form, same as a hand-entered trade.

The Strategy review (from the desk)

A generated report (~20¢, you confirm the cost first) that grades your whole portfolio A–F, reads the macro environment, ranks your priority actions with timing ("act now" vs "next quarter"), and simulates what your portfolio looks like after the top three moves. It's saved forever — regenerate only when things change.

If you've set a gain budget, a trim that breaks it says so. On your investor profile you can state the capital gain you're willing to realize in a single tax year. When a recommended sale would take the year past that figure, the advice says so and suggests splitting it across two tax years — a gain is reported in the calendar year of the disposition (CRA, T4037), so a December sale and a January sale land in different years and are genuinely different trades. Leave the field blank and nothing is assumed: an absent budget is unknown, never zero and never unlimited. A stated 0 is treated as a real instruction — realize nothing this year. The budget reaches every surface that shows you signals, so the report, the app lists and the AI chat cannot disagree about the same trade.

It tells you where your next contribution goes. If you're still saving, what you're about to add usually matters more than rearranging what you hold. With an annual contribution and an investment policy on file, the report splits that contribution across stocks, bonds and cash in dollars, tells you how many years of contributions reach your policy without selling anything, and says what it does to a sector you're heavy in. When one year is enough, the rebalancing action says so beside the trade — so you see the route that sells and the route that doesn't. It is arithmetic on what you hold today: growth is ignored and no return is assumed. It needs a policy to work against, and it doesn't pick the account for you.

A breach you've already seen asks about your policy: If your book was outside your investment policy in an earlier Strategy report too, in the same direction, and you haven't changed the policy since, the report stops simply repeating the trade. It asks whether the trade is owed or the policy no longer describes what you want, and says to revise it deliberately, not to make the alert go away. It also asks you to review a policy you last saved more than a year ago; saving it unchanged counts as a review.

Your own concentration limits: Under Profile → Risk Comfort you can set Max comfort in one sector (%) beside the existing Max comfort in one stock (%). Blank keeps the house limits: 45% of your holdings in one sector (new buying stops at 40%) and 25% in one stock (new buying stops at about 22.2%). A limit you set gets the same headroom — ADD suggestions stop at 40/45 of it — and Today's concentration card fires at it. Every size check uses the whole position: the same stock summed across all your accounts. A lower number makes the report trim sooner. A higher one is honoured: the trims stop, and the Strategy report says plainly, for each sector or stock concerned, that you are above the house limit by your own choice, with the weight, the house limit and yours. Before this change a looser single-stock limit was applied without saying so.

Cash you keep aside is left out. Enter a cash reserve in dollars under Profile → Liquidity & Safety — an emergency fund, or savings you do not want invested. It is removed from your asset mix before any percentage is worked out, so a rebalance never proposes investing it and the Strategy report never counts it as spare cash. If the reserve you declared is larger than the cash on file, the shortfall is shown and your next contribution rebuilds it before anything is invested. Leave it blank and nothing changes — but where a rebalance would draw on savings you entered, the report tells you how much of the cash that is and asks whether any of it is a reserve. It never guesses one from an account's name.

A suggested fund is one the engine has looked inside. Funds come from the scanned fund list, not from memory, and each states how much of the fund sits in the sector you're heavy in and what your sector weight becomes if new money equal to a tenth of your portfolio goes into it. A fund holding more than 40% in any one sector isn't offered as a diversifier — the same line the report uses to call your own portfolio concentrated. If one sector is above 40% of your portfolio and no such fund was suggested, the report places one first, with no position size.

Every suggested company says why it counts as diversification — and admits when it can't be checked. The screen picks names outside the sectors you're heavy in. That's a real reason, and a weak one: two companies in different sectors can still rise and fall together, which is why the Risk X-Ray counts effective bets rather than holdings. Where there's enough price history, the report measures whether adding the name actually raises your effective bets and tells you the number. Where there isn't — most of the time, today — it says not measured instead of leaving you to assume. And if a name measurably doesn't diversify, it says that too, beside the suggestion.

If the screen keeps losing, it switches itself off. Once picks have been graded against the market and their own sector, a screen that persistently underperforms stops suggesting individual companies at all and the report falls back to allocation advice. Nothing has been graded yet, so the rule reports that honestly rather than implying it's been watching.

Individual companies are research, not sized advice. When the report suggests a single stock to diversify into, it arrives labelled Research with no position size, and it tells you how many picks from that screen have a graded forward result — today, none, along with the date the first one matures. The reason is plain: a size is what turns a name into a recommendation, and this screen's picks have never been tested against what happened next — the Deep Score it ranks them by was tested only on its fundamentals, which ranked later returns modestly, not on its valuation inputs or labels. Funds are different — an ETF still carries a suggested weight, because that is an allocation decision resting on the policy you set and on tax rules, not on picking a winner.

Haven't set a policy? Your questionnaire already proposed one. If you've answered the risk questionnaire, the Investment Policy panel shows the mix it proposed, why it proposed that, and a Review and accept button that fills the editor in for you. You still press Save — a proposal only becomes your policy when you accept it, and nothing is measured against it until you do. Change any number before saving if you disagree.

If you've set an investment policy, a breach of it leads the plan. Your policy is the target split between stocks, bonds and cash, with a drift band — and it is the first thing checked, ahead of any single-stock idea, because asset allocation is the largest decision in the portfolio. When a class falls outside the band the report gives you the whole trade, not half of it:

• Every class outside the band, not just the biggest. The one you're over in and the one you're under in are the same trade — being told "reduce cash toward 5%" without being told the money belongs in bonds is half an instruction.
• In dollars, and it balances. Every class moves to the target you declared, so what goes in equals what comes out. A class inside its band still appears, because it's where money comes from or goes.
• What funds it, because that's what it costs. Moving idle cash into bonds is not a sale and realises no tax. Selling stock to do it realises a capital gain or loss in a non-registered account — so new contributions are offered first, since they shift the mix without a sale.
• The tax itself is not estimated. Which shares, in which account, at which cost base decides it, and the report isn't given that — so it tells you how much requires selling and stops, rather than quoting a number it can't stand behind.
• Money it can't classify is named, not absorbed. A holding the fund look-through can't split into an asset class is excluded from the trade and disclosed.

Two things the report discloses about its own reasoning, unprompted. Both are written by the engine after the AI has finished, so the AI cannot leave them out, and neither changes a recommendation, a ranking or a suggested size.

• How much a pick leans on past growth. The Deep Score's Growth pillar is 25 of its 100 points, and both of its inputs are historical growth. Where a recommended stock owes that pillar more than an evenly-scoring name would — above its own 25% share — the recommendation says exactly what share, and cites the research: long-term earnings growth has shown no persistence beyond chance (Chan, Karceski & Lakonishok, 2003), and the fastest-growing names have historically underperformed over five years (Lakonishok, Shleifer & Vishny, 1994). The note calls the rank a screen, not a forecast — nothing here has been validated against future returns.
• How much of your portfolio is individual stocks. Measured by value, not by count, and shown whenever it exceeds the 20% default budget (you can set your own). Counting misleads: thirty holdings at 3% each look spread out and are not. The figure comes with its basis — only 42.6% of roughly 26,000 US stocks since 1926 beat Treasury bills over their lifetime, with the best 4% of firms creating all the net wealth (Bessembinder, 2018), and removing 90% of diversifiable risk takes 40–50 names, not 15–20 (Statman 1987; Domian, Louton & Racine 2007). It is stated, not enforced: nothing is blocked or resized.

16. Sector Lab — mastering one industry at a time

What it's for: getting genuinely good at ONE industry — because a software company and a pipeline play by completely different rules.

Where to find it: Sector Lab lives under the Advisor — open it from the Advisor's Sector theses desk card, or press ⌘K and type "sector". (It keeps its own room because it's the one AI surface allowed to research the live web — clearly separated from the grounded portfolio chat.)

Pick a sector and you get: a scorecard (how many companies tracked, the sector's median valuation and quality numbers, its current macro tailwind/headwind), the top quality names in that sector, and — if you generate one — a full sector thesis: a cited, structured briefing that fuses live web research with your own holdings (the only AI surface allowed to read the web; every outside claim carries its source).

The Conviction Journal is the quiet gem: write down why you believe in a sector and — more importantly — what would prove you wrong (your "kill criteria"). Future-you, mid-panic or mid-euphoria, gets to read what calm-you actually thought.

17. Tax Centre — keeping more of what you make

What it's for: Canadian investors lose more to tax mistakes than to bad stock picks. This screen exists so you don't.

What it watches, in plain words

• Tax-loss harvesting, checked against the real rule. A loss is only harvestable if the CRA will actually let you claim it. The check asks how many shares you rebought within the ±30-day window, whether you still hold them at the end of it, and whether a spouse bought the same security — affiliation under s.251.1, counted only when you have both declared each other. A $0 stock-split row is not a repurchase, and settlement dates decide the window, not trade dates. If the check cannot be completed it says “wait” rather than “harvest now”, because the wrong answer there invites a trade whose loss is denied. The headline, Tax you could save by harvesting now, counts only the losses a sale today would keep; the ones that have to wait out the 30-day window are on a second line.
• Capital gains — your realized gains this year and which losing positions could be sold to offset them (tax-loss harvesting), with the superficial-loss 30-day trap flagged.
• If every account is registered. When nothing is held in a taxable account and there are no sales from one on record, the capital-gains, tax-loss-harvest and T1135 panels are replaced by one sentence saying why: they only apply to non-registered accounts, and they come back as soon as you hold something in one.
• Asset location — which account a holding sits in matters. Example: US dividend stocks in a TFSA silently lose 15% of every dividend to US withholding tax; the same stock in an RRSP loses nothing (a Canada–US treaty exempts RRSPs). Funds are judged by how they are built: a Canadian-listed fund of US stocks loses the 15% inside the fund even in an RRSP, and a US-listed fund of overseas stocks loses the overseas countries' tax in every account — the Tax Centre shows, in dollars, what each holding loses and what an RRSP would actually keep. Every holding gets a placement score out of 100 with a plain note like "better in your RRSP — saves the withholding." Where the fix would mean moving an existing holding rather than directing new money, the note is priced first: an in-kind transfer into a TFSA or RRSP is a deemed disposition at market value, so the app states the gain that move would realise today — and tells you not to move a position sitting on a loss, because a loss transferred into a registered plan is denied outright (ITA 40(2)(g)(iv)).
• T1135 — own more than $100,000 (cost) of foreign investments in taxable accounts and the CRA requires a form, with real penalties for forgetting. The app tracks your total and warns you before you cross.
• Contribution room — your TFSA/RRSP/FHSA room (from Settings), with nudges when room sits unused.

The Tax & Structure Audit

The Generate audit button (~5¢, confirmed first) has the AI review your whole structure and produce a graded report with a to-do list — move this, harvest that, consolidate these. Each item is a checkbox that checks itself off automatically when the app detects you made the matching trade. The report's cover shows the risk tolerance and time horizon you declared, or "Not on file"; a harvest saving is priced only on a loss the superficial-loss rule would let a sale today keep (a blocked loss is listed as blocked); and the T1135 status names its tax year. Only the audit's structural moves — migrate, consolidate, harvest — reach Today, labelled as the audit's.

18. Goals

What it's for: a number without a purpose is just a number.

Add a goal — a name ("$1M by 2031", "House down-payment"), a target amount, a year. The app shows your progress bar, years remaining, and the return you'd need to get there. If you've set your annual contribution in your profile, it does the honest version of that math: contributions do most of the work, so the required return on top is usually much lower than the scary raw number. On-track goals show green; a goal that has fallen behind quietly becomes a card on Today.

19. Retirement income — will the money last?

What it's for: the biggest question in personal finance, answered with real Canadian rules. Lives at the bottom of the Goals page.

Setting it up (5 minutes)

1. Settings → Profile: birth year and target retirement age (the planner won't guess your age).
2. On the panel, press Set up your plan and enter: - Your annual spending target in retirement (today's dollars — what would you actually live on per year?) - Your CPP and OAS estimates at 65 — get the real numbers from your My Service Canada Account in two minutes. The app deliberately refuses to assume "maximum benefits" because almost nobody gets the maximum. - When you plan to start CPP (60–70) and OAS (65–70) - A real return assumption before fees (default 3%/year above inflation — conservative on purpose). The plan deducts your funds' fees (their MER, weighted over everything the plan draws on) and any advisory fee you enter in your profile, and shows both figures. If a fund's fee is not on file, nothing is deducted and the panel says the figures are gross.

What comes back, card by card

• Sustainable spend — the level, inflation-adjusted amount your investable money can pay you every year until your planning horizon, landing at exactly zero if returns match your assumed real return every single year. That makes it the middle case, not a safe figure: when returns vary, spending at that level runs out early in roughly half of the simulated lifetimes. This is your number from your balances — not the generic "4% rule." (Money for kids' education — RESP — is excluded; that money already has a job.)
• CPP / OAS at your chosen ages — with the official adjustments: starting CPP at 60 cuts it 36%; waiting to 70 raises it 42% (OAS: up to +36% at 70). The breakeven age tells you when waiting pays off in total dollars — e.g. defer CPP to 70 and the bigger cheques overtake around age 82. Live past that, deferring won.
• Total vs your target — surplus or shortfall, in plain dollars. Shown only once both your CPP and OAS estimates are entered. An estimate you have not entered is unknown, not zero, so until then this card shows what your portfolio supports on its own beside your target, and names the estimate it is waiting for — it does not tell you that you are short on the assumption you will receive no government benefits.
• Planning to age … — your horizon is the age at which you have a 1-in-4 chance of still being alive (with a partner's birth year on file: at least one of you), from FP Canada's survival table. Enter your sex in your profile to use your own column; left blank, the plan uses the longer one. The card also shows what you could spend if the horizon were five years shorter or longer.
• Odds your plan lasts to your horizon — instead of assuming markets return the same thing every year, the app replays your retirement through 1,000 simulated market lifetimes, starting from the pot projected at your retirement (the same one the sustainable spend uses), and reports how often the money survives. 90%+ is robust; below ~70% the plan is leaning on luck. With a spending target set, the panel adds the sustainable spend's own odds and the spend that lasts in 90% of simulations. It also shows your ending balance in tough (p10), median, and kind (p90) markets, and — because the same inputs always produce the same odds — the number won't flicker every time you refresh.
• Bridge years — retire at 60 with CPP at 70 and your portfolio carries everything for the gap. Those are the years a market crash hurts most (the technical name is sequence-of-returns risk), and the panel calls them out explicitly.
• Two Canadian traps, checked automatically:
• OAS clawback — above ~$95,000 of retirement income (2026), the government takes back 15¢ of OAS per extra dollar. The panel shows if your plan crosses the line.
• Forced RRIF income — the year after you turn 71, the CRA forces you to withdraw a rising percentage of your RRIF (starting at 5.28%), fully taxed, whether you need it or not. If your RRSP is big enough that this forced income would trigger the clawback, the panel warns you decades in advance — while you can still fix it by drawing the RRSP down earlier in low-tax years.
• Drawdown order — which account to spend from first and why: taxable first (only the gain is taxed), RRSP steadily through the middle (fill the low tax brackets, shrink the future forced withdrawals), TFSA last (tax-free growth, and TFSA withdrawals never count toward the clawback — your pressure valve).

Everything is in today's dollars, pre-tax, and every assumption is printed on the panel. It's a planning instrument, not a promise.

20. Rebalance & stress test

Rebalance (its own page): drag holdings' weights around and watch quality, concentration and upside recalculate live; or set targets and let it propose the exact trades — each with its tax cost shown, preferring moves that trigger the least tax. Nothing executes; it's a planning table.

Stress test (button on the Portfolio page): "what if the market fell 10/20/30/50%?" — what your invested holdings would lose (beta-weighted, so defensive holdings cushion; cash is not shocked), and what it does to your goals. A holding with no beta on file is left out and named with its value — never given one — and the portfolio's value after the fall is then not stated; a CIBC CDR (e.g. NVDA.TO) uses its US company's beta. With no betas at all, nothing can be modelled, so the table states neither a loss nor a value after, and says so. Run it before crashes, so the numbers are familiar during one.

21. Risk X-Ray — how concentrated you really are

What it's for: a sector pie tells you what you own; the Risk X-Ray tells you how much of it is really the same bet. It reads the daily price history of your actual holdings to reveal the risk a weight list hides.

The headline. You might hold fourteen names but — because they rise and fall together — carry the risk of only ~4 independent bets. That gap is what a single shock exploits.

What it shows

• Effective bets — your name count versus how many genuinely independent positions they amount to, from the correlation of their daily returns in Canadian dollars.
• A map of where your money clumps — your holdings drawn as bubbles, grouped by sector; big same-colour clumps show your money leaning the same way (the detailed correlation grid is one click away).
• Factor tilts — whether your book leans to momentum, value, size or quality; a shared style wins and loses together.
• Realized volatility & Sharpe — how much your book actually swings per year in Canadian dollars (a US holding's currency move is part of the swing), and the return you earn for that ride — real statistics, not just beta.
• Maximum drawdown — the worst peak-to-trough fall in your holdings' prices, with deposits, withdrawals and trades taken out — so moving money to cash isn't a drop and buying through a fall doesn't hide one — and whether it has recovered.

Not just a read-out — simulate it

Drag a slider to trim a holding and your diversification ratio and volatility recompute instantly. Send the freed weight to cash (lowers volatility, but your remaining bets stay just as linked) or pro-rata into your other names (what actually improves diversification). The X-Ray also flags your biggest lever — the single trim that would help most. When you like the result, Take this to Rebalance hands the target to the Rebalance tool to build the tax-aware trade list and execute it.

A mirror, not a verdict. The diagnostics describe the risk you already hold — they don't tell you to buy or sell, and the what-if never trades on its own. Trimming winners is usually taxable, which is exactly why the action routes through Rebalance (where the capital-gains bill and tax-neutral options appear) rather than a one-click "de-risk" button. Names with too little price history are excluded and the coverage is disclosed; a US holding with no exchange-rate history is excluded too, and named, rather than priced as if it were Canadian. Find it at Portfolio → Risk X-Ray.

22. Alerts, watchlist & thesis guards

Found under Tools (wrench menu) and Settings → Alerts. All delivered via Telegram (connect once in Settings → Alerts).

• Watchlist price alerts — "tell me if AAPL hits $150." Patience, automated.
• Thesis guards — the underrated one: set fundamental tripwires per stock ("alert me if the score drops below 60 or debt/equity rises above 1.0"). Breached on two consecutive scans → alert. This catches a business deteriorating before the price fully reflects it.
• Earnings heads-up — automatic: the evening before any holding reports, you get a note. Expect volatility; review your thesis, not the ticker tape.
• Fair-value crossings & discovery digest — when prices cross fair-value bands, and when new names enter the quality universe.

23. Settings, household & security

• Profile — province, income band, birth year, retirement age, contributions, contribution room. Feeds tax and retirement math.
• Accounts & Data — broker connections (read-only, revocable anytime), file imports, account management.
• Preferences — CAD/USD display, theme, notifications.
• Household — invite your spouse by consent (they accept from their own login). You each keep your own login and private data; a Household / You / Partner toggle shows combined or individual views. Tax, Goals, Strategy and Retirement always stay individual — as they should.
• Security — change your password, and sign out of every device. When you set a new password it must be at least 10 characters, not one of the common/easily-guessed passwords the app screens against, and different from your current one. These checks apply only when you set a new password — your existing login keeps working unchanged. Changing your password (or an admin resetting it) signs you out everywhere else, so you'll sign back in with the new one.

24. The fine print — what the app deliberately does NOT do

Knowing a tool's edges is part of trusting it:

• It never trades for you. Every broker connection is read-only. Every suggested trade lands on a confirmation screen.
• The AI never invents numbers. It reads values the app computed; if data is missing, you see "no reliable estimate," not a guess.
• Suggested ≠ instructed. Scores, badges, sizing bands, drawdown orders — all describe what transparent rules see. You decide.
• The retirement plan is pre-tax and in today's dollars, with its assumptions printed on the panel. It's a compass, not a GPS.
• Manual net-worth entries are CAD-only and don't affect investing analytics (except cash/GICs joining your asset mix, which is disclosed).
• Some data can be imperfect (fund categories, calendar dates come from public sources). Where the app can't verify, it says "assumed" out loud.
• It is not a licensed financial, tax, or investment advisor. For decisions with serious consequences — retirement, big tax moves — use this to arrive prepared at a professional's desk.

CompoundWise — invest like an owner, not a trader.