Methodology

We built a transparency product, so here is the whole method — the same arithmetic the app shows on every card, written out once.

Data & sources

Prices are daily split/dividend-adjusted history (12 years) from a market-data provider (Tiingo by default). Macro series — CPI, core PCE, the federal funds rate, the 10-year yield, and unemployment — come from FRED, the St. Louis Fed's database. When a live feed is unavailable the app falls back to a clearly-labelled synthetic series — it never presents demo data as real.

1 · Technical signals

We compute standard indicators directly from the price history — moving averages (50- and 200-day), RSI, MACD, Bollinger bands, ATR, and recent support/resistance. These are plain arithmetic, not a model. Each signal becomes a named line item with a point value and a one-line reason.

2 · The macro regime base rate

We classify today's economy from the macro series — is inflation rising or falling, is the Fed hiking, holding, or cutting — then find every past month whose regime matched today's and measure how this stock actually behaved over the next 90 days. You get a hit rate, a median move, and the sample size. Confidence is labelled high(12+ analogs), medium (6+), or low — and we never dress a thin sample up as certainty.

3 · The score

The 0–100 score starts at 50, then adds the green and subtracts the red from the signals above. Nothing is hidden: the weights live in configuration, not a vault, and the full ledger is shown on every analysis. Labels map exactly to the number — Bullish at 70+, Cautiously Bullish 57–70, Neutral 43–57, and the mirror on the downside.

4 · The analyst read

A language model writes the plain-English “read” — bull case, bear case, synthesis, and a concrete “what would change my mind” level. It only narrates the numbers computed above; it does not invent signals or levels.

5 · The Scorecard

Every analysis is logged with a timestamp, horizon, and entry price, then resolved against the real outcome — right or wrong — and posted, including the misses. We also publish a calibration curve: when the read implies a 70% chance, does it land near 70%? See the live record on the Scorecard.

This is a historical and technical read for information and education only — not financial advice. See the full disclaimer.