# Paper03QSR — Day 1 review: watching QSR instead of buying it

**Date:** 2026-09-02 (first live session of the confirmation-entry module)
**Account:** Paper03QSR — $10,000 paper, QSR-only, 5% cash reserve
**Module config:** confirmation entry ON · **trailing** mode · +2% trigger · 5-day expiry

## What this account does differently

Every other QSR account buys the moment yahoo-screener flags a name. Paper03QSR
doesn't. It records the signal to a watch list and **buys only once the price
rallies +2% off its lowest point since the signal** — the "it has stopped
falling and started going up" confirmation you asked for. In **trailing** mode
that +2% is measured from the running *trough*, not the signal price, so the
trigger ratchets down as a name keeps dropping and fires on the first real
bounce off the bottom.

## What happened

| | |
|---|---|
| QSR names **watched** (not bought on signal) | 29 |
| **Triggered** (fired the +2% bounce) | 10 |
| **Expired** unbought | 0 *(5-day window still open)* |
| Triggers that became **real positions** | 7 |
| Triggers that fired but **didn't buy** (cash/caps) | 3 — CAT, SNDK, FIX |
| Capital deployed | ~$3,519 of the $9,500 investable pool |
| End-of-day equity | **$9,991.98** (unrealized **−$8.01**, −0.08%) |

All seven positions are still open, most within ±1.5% of entry. Nothing has
closed, so this is a report on **how the entry mechanism behaved**, not on
whether it made money.

## The headline finding: trailing caught 10, fixed would have caught 2

Because both trigger constructions are logged on every resolution, we can see
exactly what a "+2% off the signal price" (fixed) rule would have done instead:

| Entered under **trailing** (+2% off the trough) | Would ALSO fire under **fixed** (+2% off signal) |
|---|---|
| TER, LRCX, ETN, ADI, ASX, ROK, CAT, SNDK, FIX, PANW — **all 10** | **only ADI and ROK** |

**Eight of the ten entries exist only because we measured the bounce from the
low, not from the signal.** Those eight names kept falling *after* the signal,
bottomed, then bounced ~2% off that bottom — a level fixed mode would have sat
waiting for and, on the day, never seen (their `fixedCrossedAt` is empty). This
is precisely the "at the bottom → going up" pattern the module was built to
catch, and trailing vs fixed is by far its biggest lever: same watch list, 5×
the entries.

## What we can't conclude yet (and shouldn't pretend to)

- **No closed trades.** The whole claim — that waiting dodges more stop-outs
  than it misses winners — needs round-trips. Day 1 has none. The current
  scoreboard (2 green: TER +1.1%, PANW +1.6%; 5 red: ETN, LRCX, ASX, ADI, ROK;
  net ≈ flat) is noise, not evidence.
- **"Waiting" was minutes, not days.** Most triggers fired within 0–1h of the
  signal (the latest, PANW, took 5h). On day 1 this behaved as *intraday
  reversal capture*, not multi-day patience — worth remembering when reading
  the 5-day expiry window: nothing came close to using it.
- **The names were deep-downtrend.** MA50 slope at signal was negative on 9 of
  10 (only PANW positive, +2.4%). Trailing's job is to catch a bounce inside a
  downtrend, so this is expected — but it's also the setup most prone to a
  dead-cat bounce that rolls back over. relVol-at-signal reads tiny (0.02–0.36)
  purely because of the IEX feed's fractional volumes; ignore the absolute
  numbers.

## Early read and what to watch

- **Trailing is very permissive** — 10 of 10 watched-and-moved names fired.
  That's the intended sensitivity, but it also means it will enter bounces that
  are just noise. Whether that's a feature or a leak is exactly what the closed
  trades will tell us.
- **The counterfactual logging works.** Every resolution records where both the
  fixed and trailing triggers would have fired, so "would +2% off the signal
  have been better?" stays answerable from forward data — no re-run needed.
- **Bar to clear before trusting it:** ~40+ *resolved* confirmation entries with
  a dodge-to-miss ratio of at least 2:1. We have 10 entries, 0 resolved. Roughly
  a few more weeks of sessions.

**Verdict for day 1:** the mechanism did its job — it watched instead of
buying, and trailing mode caught eight bounces that a signal-anchored rule would
have missed entirely. Whether those eight were worth catching is a question only
the exits can answer, and none have happened yet.