Track record
Did the leans play out?
Every curated stock gets an overall lean from its offbeat signals: tailwind, headwind, mixed, or quiet. We log those leans and check what the stock actually did 2 weeks, 1 month, and 1 quarter later, on its own and against the S&P 500.
Backtest results Backtest · reconstructed
Loading…
| Window | Calls | Direction right | Vs S&P 500, as leaned | "Always guess up" | Avg vs S&P after tailwind | Avg vs S&P after headwind |
|---|
Per stock
1-month window. Each square is one monthly backtest checkpoint: lean beat (or lagged) the S&P 500 the way it leaned, it didn't, mixed or quiet (no call), window still open. Here green and red mean "played out" or "didn't", not "number up or down".
Live log Live
| Stock | Latest logged lean | Price when logged | Logged | Live results |
|---|
How it's scored
How the backtest avoids peeking
Signals left out of the backtest, and why
| Stock | Signal | Why it's not backtested |
|---|
Read this before trusting any number above
- Small samples. Per-stock results rest on a few dozen calls at most. A 60% hit rate on 30 calls can easily be luck.
- Overlapping windows. Checkpoints are monthly, so 1-quarter windows overlap and the real number of independent results is smaller than the count shown.
- Hindsight in the stock list. These stocks and their signals were chosen in 2026, knowing how the companies turned out. That flatters the "vs S&P" column for every lean, which is why tailwind and headwind leans can both show positive average moves.
- The backtest is not the live site. It uses only the signals with dated public history (listed above), conservative publication delays, and today's revised government data, not the numbers as first published.
- Correlation isn't causation. A lean that lined up with a price move doesn't mean the data caused it. Public data is often already priced in by the time it's published.
- Education only. Not financial advice. Nothing here tells anyone to buy, sell, or hold anything.