AI Can Challenge The Evidence. It Still Can't Change The Model.
The Morning Dividend AI Income System turns source-receipted market and fundamental evidence into a transparent 100-point Income Score, a 100% target map beside an actual cash-and-transaction ledger, and an on-site queue when rules detect change.
Founding price: $199 a year. Checkout shows the current billing mode before you enter any details.
Dear Friend,
A dividend investor does not need another black box blinking BUY.
You need to know where the check comes from, what can stop it, whether the balance sheet can carry it, and what price makes the whole bargain worth owning.
AI can help challenge source-receipted market and fundamental records, find a number that moved, and catch two facts telling different stories. That is useful work.
But a machine does not know your worry when a dividend gets cut. It does not get to hide a stale source behind a handsome score. And it certainly does not get to invent confidence because the market had a good week.
So I built two working doors and one visible publication gate.
First, the evidence must be current and source-receipted. Second, the company must clear seven visible income tests. The current inception artifact records those rules and the AI challenge. It does not yet record a named human reviewer.
Human review is the required gate before a future editorial action change can be described as approved. Until that field is real, I will show it as “not recorded”—not hide the gap behind the byline.
That gives you more than a list of tickers. It gives you a model you can inspect, research you can challenge, and an on-site queue tied to the weekday snapshot.
I want AI on the research desk. I just refuse to put it in the editor's chair.
— Arthur Whitmore, The Morning Dividend research desk
Arthur Whitmore is the house pen name for The Morning Dividend editorial and research desk—not a fabricated individual resume. The desk owns the publication policy, corrections, and decision-log standard. The inception snapshot has no named human approval record. Questions and correction requests go to hello@morningdividend.com.
Seven Lenses. One Score. Every Point Has A Receipt.
Nothing earns a number until the source, date, calculation, failure condition, and rules result are visible. Human review is reported separately.
Rules gate the evidence
A stale price, missing source receipt, mismatched accounting measure, or unsupported dividend claim stops the case. No evidence, no score.
AI challenges supplied facts
The challenge pass tests supplied facts, compares periods, and raises contradictions. It proposes research questions—not unreviewed recommendations.
Human review is a separate gate
Policy requires a named reviewer for a future editorial action change. The inception snapshot does not contain that signoff and is not presented as human-approved.
| Income lens | Weight | What earns the points | What can break the score |
|---|---|---|---|
| Payout quality | 22 | Points from the supplied payout ratio: 22 at 60% or less, 18 at 75% or less, 12 at 90% or less, otherwise 5. | The supplied ratio is missing, implausible, or above a published veto threshold. This is not free-cash-flow coverage. |
| Cash durability | 18 | Bounded arithmetic blend of supplied annual free-cash-flow growth and revenue growth. | Inputs are missing or implausible. This lens does not establish absolute cash generation or future durability. |
| Balance sheet | 14 | Points from supplied net debt to EBITDA: 14 at 1.5× or less, then 11, 7, or 3 by published bands. | The ratio is missing, implausible, or too high. Maturities, liquidity, and covenants are outside this lens. |
| Dividend growth | 14 | Arithmetic blend of supplied five-year dividend CAGR (with fallback) and consecutive increase years. | Vendor history is incomplete or the supplied growth record weakens; this is not issuer verification. |
| Valuation | 12 | Inverse blend of supplied P/E and price-to-free-cash-flow ratios; richer multiples receive fewer points. | A multiple is missing, implausible, or rich under the published scale; cycles and sector accounting are not normalized. |
| Business resilience | 12 | Points from supplied beta only, with lower market sensitivity receiving more points. | Beta is missing, implausible, or high. This is not a moat, customer, regulatory, or event-risk assessment. |
| Entry setup | 8 | Current quote position inside the supplied 52-week range; lower range position receives more points. | The range is invalid or price misses the entry zone. This is not valuation proof, momentum, or a forecast. |
| Total | 100 | A comparable research score—not a promise of return, safety, or dividend continuity. | |
“A score without its sources is a sales trick. A source without a decision is a filing cabinet.
— Arthur WhitmoreThe Model Shows The Whole Decision—Not Just The Ticker
Eight researched targets. Exact intended weights. An actual cash ledger. Separate new-money actions and unapproved machine holder signals.
- Positions
- 8
- Inception allocation
- 100%
- New money
- Build · Wait
- Machine holder signal
- Hold · Trim · Exit
The member desk uses the generated August 1, 2026 snapshot. The target map totals 100%, but the live model ledger holds $10,000 cash, zero securities, and zero transactions.
Find durable income
Use the seven-lens Income Score to compare unlike dividend businesses without pretending one yield tells the whole story.
Reject weak cases
A rules gate stops incomplete, stale, or contradictory research before a stock can earn a model position.
Size the model
See an eight-name target map beside actual cash, holdings, and transactions, with unapproved machine signals kept separate from published editorial guidance.
Monitor the facts
Follow the on-site queue and decision logs that name what changed, why it matters, and what would reverse the rules-derived status.
A model action cannot change because a chatbot grew enthusiastic. It changes when a named evidence gate, price gate, or risk limit changes—and the decision log shows you which one.
New-money starting point
There Are No Build Positions Today. Do Not Backfill The Holds.
The model began on August 1, 2026 with $10,000 cash, no securities, and no transactions. Reference prices establish a research baseline—not cost basis—and do not turn existing-holder guidance into a new-money Build.
- Model inception
- August 1, 2026
- Reference capital
- $10,000 Arithmetic example, not a required minimum
- Intended holding period
- 3–5 years Editorial expectation; a named evidence gate can change it sooner
- New money today
- Remain uncommitted Wait for a Build record with named human review
New money may move toward the published target weight only after the rules produce Build and the required human-review field is recorded.
New money remains at 0% for that name; undeployed model capital stays in cash until a later canonical snapshot changes it.
Proposed machine signal for an existing owner: maintain rather than add. It is not published editorial guidance while approval is absent.
Proposed machine signal for an existing owner: reduce because a material gate is failing. It is not published guidance or a model transaction until reviewed and recorded.
Proposed machine signal for an existing owner: leave because an exit-severity gate is failing. The evidence remains visible, but no human approval is recorded.
At the $10,000 reference scale, one percentage point equals $100. Fractional shares can support a smaller implementation; whole-share constraints, taxes, and trading costs can require more capital. A future Build record must name the target weight, transition step, and human-review status. Until then, new members should not copy the inception ledger as if every position were open for purchase.
A Research Rhythm You Can Actually Follow
The implemented workflow refreshes on weekday mornings and derives changes for the member portal. It does not watch continuously or send paid alert email today.
Weekday mornings
FMP snapshot refresh
The workflow refreshes source-receipted market and fundamental inputs, recalculates the rules, and timestamps the new snapshot.
After each refresh
Rules-derived portal queue
Detected status and metric changes appear inside the member alert queue. No external paid alert email is configured today.
Twice-monthly policy
Review checkpoint
The portal schedules a full model checkpoint. The inception snapshot does not yet contain a named human-review record.
First-business-day target
Member issue
The intended launch cadence is a dated portal issue after editorial review; this is a delivery target, not a claim of publication history.
Everything In The Founding Membership
One desk for the portfolio, the evidence behind it, and every change after publication.
- Member dashboardThe current model, alert queue, latest issue, and research status at a glance.
- Eight-name target map plus model ledgerIntended weights total 100%; actual cash, holdings, and transactions remain separately accounted, with Build/Wait actions and inspectable Hold/Trim/Exit machine signals awaiting editorial approval.
- Research dossiersScore breakdowns, source receipts, risks, valuation, and the reversal condition.
- Decision ledgerA dated record of what changed and why the rules responded; the reviewer field remains “not recorded” until a real human signoff exists.
- Monthly member issueThe first-business-day research letter plus the strongest opportunity on Arthur's desk.
- On-site alert queueRules-derived dividend, thesis, risk, and entry changes from the weekday snapshot. External paid alert email is not configured today.
- Methodology and archiveThe complete scoring rules and every issue in one searchable member library.
Thirty Days To Read The Research, Not Guess At The Promise
Once paid access begins, you may request a full subscription refund during your first 30 days. That policy covers satisfaction with the membership. It does not promise an investment result, protect a portfolio from loss, or make any dividend safe.
There are no hidden partner memberships, preselected offers, or brokerage permissions in the order.
Questions I Would Ask Before Paying
If a research service cannot answer these plainly, keep your wallet in your pocket.
Does AI pick stocks by itself?
No. AI challenges the source-receipted market and fundamental facts supplied to it, looks for contradictions, and names missing evidence. Fixed rules compute the current score and status. The inception snapshot does not record a named human reviewer; policy requires that separate gate before a future editorial action change is described as approved.
Is the model portfolio personalized for me?
No. It is one educational model for all members. It cannot know your taxes, other holdings, income needs, time horizon, or capacity for loss. You decide whether any idea belongs in your own plan, preferably with qualified advice when the decision is consequential.
Will the system trade in my brokerage account?
No. The Morning Dividend never takes custody, connects to your brokerage, or places a trade. The member desk shows model weights and research actions; execution remains entirely with you.
How often will the portfolio change?
It is reviewed on a schedule twice each month and whenever material evidence changes. That does not mean it will trade twice a month. A review can end with a written hold decision, and every change must name its trigger.
Are returns or dividends guaranteed?
No. Share prices can fall, dividends can be reduced or suspended, and a highly scored company can disappoint. The system is built to expose assumptions and react to evidence—not to erase investment risk or predict a certain result.
What does the founding membership cost?
The founding price is $199 per year and renews annually at $199 unless you cancel before renewal. Paid members may request a subscription refund during the first 30 days. When secure billing is configured, Stripe handles payment; otherwise the checkout route clearly opens as a no-charge preview.
The Evidence Order
The August 1, 2026 inception ledger is the published baseline built from Financial Modeling Prep market and fundamental endpoints, with the source URL and fetch time preserved for every metric. AI receives supplied facts only and cannot set or override an action. The intended editorial policy requires issuer or regulator verification and named human review before a future action change is described as approved. That approval is not recorded in the inception artifact.
- Financial Modeling Prep market and fundamental data
- SEC company filings
- Investor.gov education and risk guidance
- U.S. Treasury interest-rate data
- Company investor-relations releases and original dividend records for human verification
Put AI on the research desk.
Keep The Evidence—and Every Model Status—Where You Can See Them.
Founding membership is $199 per year, backed by a 30-day subscription refund policy once paid access begins.