a rule firing, not a verdict · How the app turns all of this into a grade
Because “buy” is an instruction and this is only a rule firing. The app’s own historical test of this timing rule shows it has NOT beaten simply owning the same names over the tested window (a −0.72% edge on 1,058 signals), and the live record is still accruing. The words say what the model checked, and the Signal Ledger shows how those checks have actually turned out — read them together.
It all comes from the one thing the app’s own backtest said mattered: never buy support in a downtrend.
The technicals can pass on a setup the rule was never written for. A name down 10% or more in ten sessions sits on its 40-day low by definition — that is the knife still falling, not a pullback to a floor — so the rule waits until the drop settles. And a name under a signed cash acquisition is pinned to the deal price: every technical state is true and none of them means anything, so it reads “Wait for entry” with the deal named, until the deal closes. Neither guard touches the grade.
“Entry condition met” is a technical statement. When the model’s own fair value sits at or below the price, the chip carries a “no modeled upside” qualifier: the rule fired, and the valuation says there is nothing to be paid for it. Read the reward:risk before treating that as a setup.
below 200-DMA, or support broke → Trend rule broken at support · uptrend · not hot → Entry condition met hot (RSI ≥ 66) or extended → Wait for entry otherwise: composite ≥ 74 → entry met, else wait down ≥ 10% in ten sessions, or a pending cash deal → Wait for entry (guard) (no technicals available → Data too thin)See it run on live stocks →Weekly Monitor computes this on ~1,000 names — free, in your browser, every formula shown. Nothing here is advice.