Results and Key Performance Indicators

How to read this. Entry price is the report's stated price that day (bf = report stored no price, so the market close is used). Peak = highest price traded since that entry, Low = lowest, Now = latest close — each measured from its own entry date to today (hindsight extremes; no one buys the low or sells the exact peak). The What if… exit at +9% or a Hold/Sell report column simulates putting $10,000 in at that entry and selling at whichever came first: the price touching +9% (a $900 gain), or the next report downgrading to Hold/Sell (selling at that report's price). The Outcome is then a WIN or LOSS on that exit; positions that never hit +9% and were never downgraded are still open and show the current BUY with its latest report date and price. Rows are ordered by report date and split into signals older than two weeks (Previous) and those from the last two weeks (Recent). "Short-term" is the stock report's 1–3 month call — not the 6–12 month medium or 3–5 year long-term rating. Losses are shown as loudly as wins.

Entry method — conviction-based sizing. How I'd actually take each signal depends on the report's entry conviction (the badge on each row). Full conviction (two-plus entry paths agree): take the full position at the signal and target +9%. Half conviction (one path — usually cheap on fundamentals but the timing isn't confirmed): scale in — 50% at the signal, then two 25% adds sitting as resting limit orders at −1× and −2× ATR below it (volatility-scaled, not a flat %, anchored to support where the report names one). The row's entry price is the blended average, and the ✓/✗ show which adds the dip actually reached. Because the plan is to average the cost basis down, every scale-in (half-conviction) position targets +12%. (Illustrative: a real pullback ladder is time-boxed to ~4 weeks, and scaling in averages down into losers too — it shapes risk, it isn't free return.)

What's excluded, and why. A few early signals are left out because they were issued before I added the valuation-ceiling, earnings-quality and short-term technical-confirmation checks to the analysis — under the current process they'd have been rated Hold, not a short-term Buy (Eli Lilly, Nvidia and Amazon; Microsoft's mid-June "accumulate on weakness" call into a bearish tape; and several precious-metals miners). I exclude them rather than count them as losses of a process that has since changed, and the reports and calibrations all stay on the site — the exclusion is a line of code you can read, not a deleted file.

Two corrections to that, found in the audit on 9 September. First, those three reports had in fact been deleted back in July rather than merely skipped; they are restored, and every excluded signal is now a line in the generator you can check against the commit that removed them — nothing is dropped implicitly. Nvidia's second mid-June Buy is listed too; it was excluded on the same basis and had been slipping through as a re-entry at a worse price. Second, an excluded signal now ends the position rather than being skipped over: skipping it quietly re-anchored the same position on its next re-affirmation at a better price, which booked Microsoft at +$344 where the honest figure was +$171. Microsoft's early-July Buy came after the fixes and is counted on its own merits.

What they would have cost, so you can check the reading rather than take it on trust: counting every excluded signal gives 30 signals, 18 wins and 8 losses (69%) and +$10,742 — a +$1,403 difference. I think the exclusions are right, and that is exactly why the number they cost belongs on the page.