qsr · Paper 02

Paper02qsr · paper · Trading mode: live · performance →

Quality-Swing-Recycle: consumes the yahoo-screener QSR signal rather than recomputing it, then scales in as an anchor leg (fixed take-profit) plus a runner that converts to a trailing stop once well ahead. No calendar time-stop — it holds the swing.

Conviction sizing off (the account defaults) Favourites, mega-caps & proven stocks →

Risk sizing overrides

Values here override the matching global constant in src/risk/rules.ts for THIS account only.

Field	Applies to	What it means	On a $10,000 account
Risk per trade %	ONE stock, one trade	How much $ you're OK losing if THAT trade's stop-loss gets hit. The system works backward from this to pick how many shares to buy.	0.5% → sized so a stop-out on any one stock loses about $50.
Max notional %	ONE stock, one trade (same one — not the whole account)	A separate hard ceiling on that SAME stock's total position value (shares × price), computed once before any tranche-splitting (e.g. qsr's anchor+runner legs just divide this one total, they don't each get their own cap). Exists because a very tight stop can otherwise math its way into buying far more shares than makes sense.	8% → that one stock's position can never be worth more than $800 total, no matter what Risk per trade % alone would size it to.
Daily kill switch %	The WHOLE account, not one stock	Once today's total P&L across everything drops to this % of account value, NEW trades stop for the rest of the day. Positions already open are untouched.	-2% → new entries pause for the day once you're down about $200 today.

All three are % of account equity (cash + market value of whatever's currently open) — deliberately not free cash, since cash alone shrinks whenever money's tied up in open trades and would understate real risk mid-session. Every % field below shows a live $ estimate against this account's current equity ($9,501.06) — edit the number to see it update.

▾ qsr

✓ Can buy — nothing currently blocking new entries in this pool.

Require manual approval before every buy.

When checked, this pool's trading script builds the order plan as usual but stops short of submitting it, queuing it on /approvals instead. Nothing gets bought for this strategy without you clicking Approve.

Confirmation entry — two independent parts, each on or off per account

1 · Waiting pen on

Wait for a confirmation rally before buying.

Rather than buying the signal immediately, record it and buy only once price rallies the trigger % above it — don't catch the falling knife, wait for it to stop falling. This account so far: of 148 signals that expired without rallying, buying straight away would have hit 68 stop-outs (dodged) vs 33 take-profits (missed), 47 not scored yet. A missed target costs 0%, a stop costs 6–8%. Entry mechanics are unchanged — same bracket, same runner, same trailing conversion. Waiting signals are listed on Confirmations.

The original 1 Sep research (historical, in-sample — the live figures are in the hints)

Trigger % — 2% is the only level with a real sample behind it (142 signal clusters). Other levels were never swept.

Expiry — 5 days was a fixed constant in the backtest, not a tested optimum. It is the window every figure here was measured over, so changing it means those figures no longer describe what you are running.

The two filters — tested, but only in-sample, on one account's pool, with thresholds chosen after seeing the data:

variant	entered	resolved	win	avg	dodged	missed
trigger only	66	36	64%	−0.14%	18	7
+ MA50 slope > −1.5%	46	20	70%	+0.70%	21	19
+ relVol ≥ 0.1	22	15	80%	+1.60%	21	30
+ both	16	10	90%	+2.80%	23	33

Each filter buys a higher win rate by entering far less: both together took 66 entries down to 16, and missed 33 winners to dodge 23 stops. At n=10 resolved that 90% is a hypothesis, not a setting — which is why both ship blank. The metrics are recorded on every trigger whether or not they gate, so these can be re-scored on forward data instead of another in-sample sweep. The per-signal charts behind every one of these numbers are on Watchlist — one chart per simulated signal, trigger/target/stop drawn in, grouped by how each comparison came out.

Trigger anchored to fixed = evidenced loss-avoider; trailing backtested 63% win / +0.04% vs fixed’s 55% / −1.13% — both crossings are logged either way Trigger % the level measured: 15 stops dodged, 6 winners missed Expiry (days) the window the evidence was gathered under, not a tested optimum

2 · Buy checks on

Blank = off. Checked on every QSR buy, whether or not the waiting pen is on — when a waiting signal triggers, or straight away when the pen is off.

Min daily move % typical daily range (ATR %). This account: under 3% 20% win* (n=5) · at/above 79% win* (n=14). Backtest (9 Oct): off −1.05%/buy (n=42) → 3% −1.19%/buy (n=25). Min rel. volume live = today's volume so far, so mostly time of day. This account: 0.05–0.1 0%* (n=1) · 0.1–0.25 69%* (n=13) · 0.25–0.5 100%* (n=1) Min MA50 slope % backtest pending — refreshed weekly by the qsr-config-sweep job Min MA200 slope % how much the 200-day average rose over the last 20 sessions; 0 = must be rising. Backtest (Sep, current exit, 3% move on): real buys +3.29% → +6.04%/buy, losses cut by two-thirds, better in both halves of the month

▾ Per-strategy config

Strategy pot % $1,612.17/$8,465.44 ≈ $8,465.44 = this strategy’s pot this strategy’s share of the account AFTER the cash reserve — the pot every % below is measured against Max open positions — ALL stocks 5/55 how many buys can be OPEN at once, every stock combined Max buys per day — ALL stocks 3/12 total buy orders per DAY across every stock combined — nothing to do with any one ticker Daily kill switch % +0.11% today ≈ -$169.31 $ pause THIS strategy’s new buys for the day once its own positions are down this % of its pot Hold back % of profit % of a WINNING trade’s realized P&L to hold back from reinvestment, tracked as a running balance (see journal/profitReserve.ts). Blank = reinvest everything. A losing trade never reserves anything. Destination pie id (T212) T212 Pie id this pool’s held-back profit is destined for — display/cross-check only, shows the pie’s live cash next to the reserved amount up top so you can reconcile before clearing. T212 accounts only; find the id in the pie’s URL or via pnpm t212-connection.

▾ Per-buy config

Governs ONE stock’s buy at a time — how big it is, and (for QSR) how often the same ticker can be re-bought. Every stock gets its own independent counters. Nothing here limits how many DIFFERENT stocks are bought; that’s the per-strategy config above (pot, open positions, buys per day).

Example: GLW at its live price of $156.60 — this config buys:

with a 5% stop → 2 shares ($313.21 buy value, risking $15.66) — limited by max 2 shares.

Updates as you edit the pot %, risk %, value cap % and max shares fields — the number shown is what a save would produce.

Max BUY VALUE per buy % — ONE stock ≈ $423.27 max value of one buy → buys ≈ $313.21 at a 5% stop (GLW) THE MAIN BUY-SIZE LEVER — the most one buy can be, as % of this strategy’s POT (units × price). Your actual buy is the SMALLER of this and the max-loss size below, so at a normal stop this usually IS the buy. The example above reacts to every field here. Max LOSS per buy % — ONE stock ≈ $507.93 max loss on one buy if its stop hits The risk guardrail, NOT a size — the most you’d LOSE (as % of POT) if one buy stops out. With the stop distance this implies a buy size (tighter stop ⇒ bigger buy); the actual buy is the smaller of that and the value cap above. So editing this moves the buy only when loss is the binding limit (usually at wide stops). Tier-3 max LOSS per buy % — ONE stock ≈ $10.16 max loss on one buy if its stop hits same, but for Tier-3 (speculative) names Max shares per buy — ONE stock a single buy decision can never take more than this many shares of that stock — even when split into anchor + runner orders Max value per buy ($) — ONE buy a fixed money cap on ONE buy’s total value, in this account’s own currency — on top of the % caps; whichever is stricter wins. Blank = no cap

▾ Exit — target & stop

Where a QSR buy sells. Applies to NEW buys only — each open position keeps the target and stop it was bought with. Default: +8% target, stop 3×ATR capped at 10%. Evidence for alternatives: pnpm qsr-bplus-exits --buy-tier.

Target % (take profit) blank = 8%. The anchor sells here; a runner (if trailing is on) trails instead Fixed stop % (instead of ATR) a flat stop this far below the buy price. Blank = use the ATR stop below ATR stop multiple blank = 3× ATR(14). Ignored when a fixed stop is set Max stop % blank = 10%. Caps both the ATR and the fixed stop

▾ B+ buying

B+ names are the screener's near-misses. By default they're only watched (to collect data, pnpm qsr-bplus-report). Backtest: B+ ≈ +1.3%/buy vs ≈ +4.3% for full buy signals (same exit) — keep these small.

On top of the usual QSR buy checks, a B+ name also needs:

nothing extra — set a B+-only check or B+ max below

Buy B+ names on this account B+ max value per buy ($) the most ONE B+ buy can be; the stricter of this and the normal caps wins. Blank = normal caps only B+ only: min MA50 slope % 0 = the 50-day average must be rising. B+ buys only — full buy signals are unaffected (it made them worse in the backtest). Backtest (5 weeks of triggers): B+ +1.44% → +2.88%/trade, 72% → 83% wins with this on.

▾ Per-ticker & leverage caps

Max buys, 7 days — ONE stock COHR 2/4 buy ORDERS of ONE stock in a rolling 7 days — every stock has its own separate counter (anchor+runner = 1 buy). Does NOT limit buys of other stocks Max buys, 30 days — ONE stock COHR 4/6 same per-stock order counter, rolling 30 days Max shares, 24h — ONE stock total SHARES of ONE stock in a rolling 24 hours — tighter sibling of the 7-day cap below Max shares, 7 days — ONE stock total SHARES of ONE stock across all its buys in a rolling 7 days — per-stock counter, other stocks unaffected Max shares per listing — ONE stock’s signal total shares of ONE stock while its current list appearance is active; if it re-lists later that’s a fresh allowance Max value, 7 days — ONE stock (% of pot) total $ VALUE of ONE stock’s buys in a rolling 7 days, as % of this pot; cheap stock allows more shares, expensive fewer Leverage every Nth buy counts ALL stocks’ buys in one running sequence; exactly one buy in every N may draw on real margin; blank = never Leverage multiplier sizing multiplier on that ONE leveraged buy