Discovery — collecting for 2026-09-25

★ evening setup = peak AND dip both LAST NIGHT, score >60 on the evening candles alone. Two years (Sep 2024–Sep 2026, net after cost): it pays when LAST NIGHT'S thrust (a ≥5% push over a prior 5-min candle body, after the trough) is there — PM thrust + a small (5–10%) AM thrust 62% / +3.0% (n=48), PM thrust + a 10%+ AM thrust 55% / +0.4% (n=202), PM thrust only 53% / +1.8% (n=43). WITHOUT a PM thrust it lost: AM thrust only 50% / -0.5% (n=181), no thrust either session 37% / -0.8% (n=41). (The Jul–Aug 2026 story that the AM thrust was the key did not hold over two years.) Dip is measured against the 4 PM close. Watchlist so far: 3 names that crossed +30%. The ranked page (candlesticks + every column) builds automatically at 8:47 AM ET — in 6h 57m. This page refreshes itself every minute. As of Fri 01:49 ET.

Row notes: dark green = helped most over two years light green = helped; any green seen-note + green minutes = the buy rule amber = hurt grey = neither, information only. More green, better.

LIST BREADTH: MIDDLE — 13.0 names a day over the last 30 days (typical 12.5; quiet under 11, busy 13.5+; through 2026-09-24)

The number that matters is how BUSY the list has been, not how it did. Over two years, no steady read: the middle third was −2.2% in year one and +1.6% in year two. It is a mild tilt (correlation about 0.1), the only reading that held in both years; the old COLD/WARM/HOT mood did not.

For the record, not a forecast: the whole list over the last 10 days +3.2% (+2.2% after ~1 pt slippage) a name (148 names) · your PICKs alone +2.3% (+1.3% after ~1 pt slippage) (37 picks).

Last 5 days, PICK win rate: 62% (MIDDLE) on 24 picks — no lean either way; base rate 55%. A next-day tilt of about ten points, not a forecast, and nothing for the week ahead.

Ticker	Peak gap our chart, off the 4 PM close	Dip	Odds win% · avg/trade	PM thrust last night	AM thrust 7:00–8:45	Price	On the feed since first crossed +30%	Last seen
INLF PICK in the last 120 days ×2 last PICK Aug 5, won	+94% feed +99% · now +96%	7%	SKIP under 58 50% -1.7%/tr Sep 2024 – Sep 2026 · scored under 58 (n=1378) 52% +1.8%/tr last 20 days (n=88)	-	-	$5.86	Thu 16:32 chart: moving since last night (+94% evening peak)	Thu 19:58
IFBD never on our list ★ SETUP + PM THRUST gap +54%, dip 71% → score 61 · two years: 53% / +1.8% (n=43); a small AM thrust would add	+54% feed +53% · now +38%	71%	PICK 53% +1.8%/tr Sep 2024 – Sep 2026 · evening setup, PM thrust only (no AM) (n=43) 30% -0.1%/tr last 20 days (n=10) → 56% · +0.9% if the AM thrust comes	+10% at 19:49	-	$1.66	Thu 17:40 chart: moving since last night (+54% evening peak)	Thu 19:58
OFAL on the list in the last 120 days ×5 last Sep 4, won evening setup, no PM thrust gap +52%, dip 97% → score 63 · without last night’s thrust this ran 37% / -0.8% over two years (n=41)	+52% feed +59% · now +5%	97%	SKIP evening setup with no thrust last night (two years: 37% / −0.8%) 37% -0.6%/tr Sep 2024 – Sep 2026 · evening setup, no thrust either session (n=43) 18% -6.7%/tr last 20 days (n=11)	-	-	$0.738	Thu 16:04 chart: moving since last night (+52% evening peak)	Thu 16:29

The playbook — peak/dip timing × thrust

Each cell: the last two years (Sep 2024 – Sep 2026, 4,605 names on the rebuilt list) on the first line, then the same numbers over the rolling last 20 trading days (from Aug 27). Average return per trade, normal 9:00 entry, idealized fills — subtract about 1 point for real-world cost. The two-year line replaced the July-onward numbers on 9/21 (Mike): the summer flattered every cell. Refreshed nightly. Thrust = a ≥5% move over a prior 5-min candle body top. Small cells — treat the ordering as solid, exact dollars as soft.

Peak / dip timing	AM thrust (7:00–8:45)	No AM thrust
Peak + dip the night before ★	+0.2% · 54% win Sep 2024 – Sep 2026 n=452 +9.0% · 84% win last 20 days n=18	+0.6% · 45% win Sep 2024 – Sep 2026 n=86 −3.6% · 24% win last 20 days n=21
Peak at night, dip finished in the morning	+0.1% · 51% win Sep 2024 – Sep 2026 n=366 +10.4% · 83% win last 20 days n=6	+0.3% · 51% win Sep 2024 – Sep 2026 n=103 +4.0% · 67% win last 20 days n=12
Peak + dip in the morning	−0.3% · 52% win Sep 2024 – Sep 2026 n=2041 +2.4% · 60% win last 20 days n=62	+1.8% · 58% win Sep 2024 – Sep 2026 n=179 −1.0% · 50% win last 20 days n=26

Thrust timing, on the night-before setups only

PM thrust (last night) and AM thrust	+1.8% · 58% win Sep 2024 – Sep 2026 n=202 +11.4% · 86% win last 20 days n=14
AM thrust only	−0.5% · 50% win Sep 2024 – Sep 2026 n=181 +0.6% · 75% win last 20 days n=4
PM thrust only (no AM follow-through)	+1.3% · 51% win Sep 2024 – Sep 2026 n=41 −0.1% · 30% win last 20 days n=10
Neither	−0.8% · 37% win Sep 2024 – Sep 2026 n=41 −6.7% · 18% win last 20 days n=11

Read it as: when the peak and the dip both finished last night, an AM thrust on top is the green light. When the peak was this morning, a thrust after the dip means the price is already climbing back before 9:00, so the buy is above the dip — those did worse than names still quiet at 9:00. A thrust that came while the dip was still forming was the worst place to buy. The pill uses this split: a peak-and-dip from last night whose only thrust came last night (PM only) is quoted from the “PM thrust only” row and is a PICK; one with no thrust in either session reads the “Neither” row and waits for the AM thrust. Refreshed nightly.