← Back to alphaedgetest.com
RUN IT YOURSELF, ON YOUR DEVICE
Method v1.2 · same input, same verdict

Does your forecast or budget beat a fair benchmark? Bring two columns (what you forecast or budgeted, and what actually happened) and get a straight verdict in about a minute. Even when the answer is no.

New here? Tap that to watch a real café forecast earn a PASS. To check your own numbers, use Step 1 below. It’s two columns from any spreadsheet.

Everything runs in your browser. Your data is never uploaded, stored, or transmitted. There is no server, no account, and no payment. After it has loaded once, it runs offline. Install it from your browser menu to use it as an app.
📲 Add it to your home screen: how (free)
iPhone / iPad, Safari: tap Share (the □ with an ↑ arrow, at the bottom) → Add to Home Screen → Add.
Android, Chrome: tap 📲 Install this app above, or the ⋮ menu → Add to Home screen.
Computer, Chrome / Edge: click 📲 Install this app, or the install icon at the right of the address bar.
It opens full-screen, works offline, and your data stays on your device.

Step 1 · Your data

or drag & drop it here. The test reads the .csv file every spreadsheet already exports
(Excel, Google Sheets, Numbers, your sales or accounting software: File → Download / Save As → CSV)

Two columns is all it takes: what you predicted and what actually happened. Not sure it'll work? Try the sample first. It's one click.

Two built-in examples, one of each verdict, because both are the product:

The PASS: six weeks of a café's daily sales. The forecast knows the weekly rhythm and the upcoming promo days, so it beats even “same day last week” (the seasonal benchmark v1.2 grades against). That extra signal is a real, provable edge. The FAIL: a monthly demand model that looks busy but loses to the baseline, the verdict that saves you money.

Need a template? Download a blank CSV. Required columns: actual and forecast. Optional: date (for ordering), and sigma or lo/hi (to also score calibration & CRPS).

Step 2 · Confirm columns

The Alpha Edge Test · Result

What to do with this verdict
PASS: it beat the toughest naive; keep using it and re-test as data grows (a floor cleared, not a ceiling). The verdict says it adds value, not how much to do. That decision is still yours.
FAIL: it loses to the simple baseline; use that baseline instead (free, and more accurate here) or fix what the model is missing.
INCONCLUSIVE: too little signal yet; gather more history and re-run before relying on it.

🧾 One honest input: a verdict only means something if the history you scored is representative of how the model really runs, not a cherry-picked good stretch. Score your full, normal record.

This is the real verdict. You ran it on your own data, on your own device. Want to take it with you? Pro exports a shareable proof PDF, runs batches, and adds deeper analysis. Unlock Pro →
Scope & disclaimer. Self-scored, method-only statistical analysis of data you provide, computed entirely on your device for your internal use. Standard published scoring rules (Forecast Value Added, calibration/PIT, CRPS, error-vs-naive) applied out-of-sample against a demanding naive benchmark. This reports whether an edge is real; it does not provide, promise, or imply one. It is not an audit, assurance, actuarial or expert opinion, not financial, investment, or trading advice, and not intended for third-party reliance. McLaughlin Capital Investments LLC is not a registered investment adviser, broker-dealer, or CTA. © McLaughlin Capital Investments LLC · alphaedgetest.com
A FAIL is a valid, valuable outcome, often the most valuable one.