The market is always in one of eight macro regimes. AlphaIntelLab scores three axes — inflation, risk, and the business cycle — resolves them into the regime that's actually running, and sends you the read with a position-sizing call before Monday's open. Same framework every week, so this Sunday is comparable to last Sunday.
One email, every Sunday. It's the same board that runs live on this site — captured at the weekly close, with the reasoning written out.
Five stages, in the same order every week. The machine collects and scores; a human decides and writes. That boundary is deliberate and it does not move.
Automated jobs rebuild a 125-row indicator warehouse every weekday evening — macro series from the Federal Reserve's FRED database, price and market data from Tiingo, and computed gauges derived from both. An intraday tape refreshes every thirty minutes during market hours.
Each series must carry enough history, be fresh, and contain no implausible daily moves. Anything failing those checks is skipped and logged rather than quietly averaged in. A cross-vendor price check flags divergence between sources. Calendar series — GDP nowcasts, CPI, payrolls — are dated to the publisher's release schedule, so a number isn't treated as stale simply because the calendar moved on.
A documented, fixed framework converts the indicator set into three scores: inflation, risk, and the business cycle. Alongside it, a weighted composite distills breadth, credit spreads, activity surveys, fiscal conditions and real yields into one risk read, and an earnings-revision measure acts as a leading indicator that tends to turn before the regime does.
The three scores map to a regime — from Goldilocks Launch through Nominal Growth Peak to Capital Preservation and Stagflation Stress — each with its own playbook of asset, factor and sector stances and its own exposure ceiling. The framework also produces the probability of moving to each adjacent regime, which is why the letter can tell you what would have to break for the call to change.
The model advises; the analyst decides. Every judgment field — the regime call, the confidence score, the stances, the written briefing — is authored by hand and published deliberately. Nothing in the commentary is machine-generated, and the automated pipeline is not permitted to overwrite a judgment field. What you read is what a person concluded, not what a model emitted.
Automation is very good at collecting a hundred series on schedule and applying the same rules without getting bored or scared. It is bad at knowing when the rules are about to mislead you. Splitting the work along that line is the whole design: the data layer is mechanical and auditable, and the judgment layer is accountable to a name.
It's a context instrument, not a trade signal. It answers "what kind of market am I in, and how much risk does that justify" — deliberately not "buy this on Monday."
The regime tells you which playbook applies. The confidence score tells you how hard to lean on it. A high-conviction regime call and a wobbling one are different instructions, even when the regime name is identical.
The exposure call — a percentage of the regime's maximum, with the throttle applied — is the highest-leverage number in the letter. Getting overall risk roughly right matters more over a cycle than getting individual positions exactly right.
The asset, factor and sector stances are directional leans relative to your own neutral allocation. They are not a model portfolio, they carry no position sizes, and they assume you already know what you own.
Firing warnings are the early-evidence layer. They frequently move weeks before the headline regime does — so a telltale that starts firing while confidence is still high is usually the most informative thing in that week's issue.
The framework is built to a weekly clock. Regimes persist for months, and the value comes from the sequence of readings rather than any single one. Reacting to it intraday reintroduces exactly the noise it was designed to strip out.
The public dashboard runs on the same data and refreshes nightly, so you can see the market numbers move mid-week. The judgment layer only changes on Sunday — by design.
The Sunday Reading is the first package. Two further tiers are in development; subscribers hear about them first.
The weekly regime board, exposure call, positioning and briefing. Free four-week trial, no card required.
The model portfolio built on the regime call, with individual holdings, weights and the reasoning behind each position.
The complete engine — full diagnostic axes, the eight-by-eight transition matrix, accuracy scorecard and data access.
If something isn't answered here, reply to any issue — the replies come to a person.
No. AlphaIntelLab publishes educational macro research. It explains a published framework and the stances that framework produces — it does not know your circumstances, your holdings, your tax position or your time horizon, and it will not tell you what to do with them. Personalized advice is declined by policy, not as boilerplate. Treat the reading as one input to your own process.
Once a week, on Sunday, timed to land before Monday's open. The underlying market data rebuilds every weekday night; the judgment layer is updated once a week. That's deliberate — regimes turn over months, and a daily letter would mostly be reporting noise.
Four full issues — not a sample or an excerpt. No card is required to start, so nothing charges automatically when the trial ends. You'll simply hear from us about continuing before the fourth issue runs out.
No. This one matters, so it's worth being precise. Automation does the collecting, the quality-gating and the arithmetic — 125 series, every weeknight, applying the same rules. Every judgment field is written by a person: the regime call, the confidence score, the stances and all commentary. The pipeline is explicitly barred from overwriting those fields. Models are used as a second opinion during review; they never author what gets published.
Macroeconomic series come from FRED, the Federal Reserve Bank of St. Louis database, which is public domain. Market and price data come from Tiingo, with a secondary vendor as a rescue path and a cross-vendor consistency check on top. Every series is quality-gated before it counts; failures are skipped and logged rather than silently interpolated.
No, but the reading assumes some fluency. It uses terms like credit spreads, breadth, real yields and volatility term structure without redefining them each week, and it assumes you already have a portfolio and a neutral allocation to tilt away from. If you're comfortable reading a market commentary, you'll be comfortable here.
Rarely, and never quietly. Fixed rules with fixed weights are what make one week comparable to the next — a framework rewritten after every surprise is just hindsight with extra steps. When the framework is genuinely revised, the change and its reasoning are stated in the issue.
Some of them will be — that's what the confidence score and the transition probabilities are for. The framework is designed to be wrong in a legible way: you can see which axis was misread and which telltale fired first. Corrections and regime changes are stated plainly in the following issue rather than being quietly dropped.
One click from any issue, effective immediately. No confirmation flow, no retention offer, no exit survey. Your email address is used to send the reading and for nothing else — it isn't sold, rented or shared.
Yes — the live dashboard shows the current regime call, the axes, the positioning grid and this week's briefing, free and without an account. The Sunday Reading is that board delivered to you, with the week's reasoning written out.
Four issues, free, no card. If it isn't earning its place in your Sunday, one click removes it.
Get the Sunday reading