# QSR take-profit sweep — account `paper-main`

Generated 2026-08-26T22:26:26.489Z · 115 buy event(s) (70 anchor/runner duplicate row(s) collapsed), 41 symbol(s)

Current live setting: `QSR_V1.hardSellPct` = **5.5%** (flat, every ticker).

**Research only** — re-simulated from real daily bars, stop-before-target pessimism, no trailing-runner leg, no portfolio constraints. See this script's header for the full limitations. Never merge into the journal.

## All symbols combined

Read **%/day**, not %/trade. A wider target can only ever be reached later, never sooner, so per-trade expectancy mechanically favours the widest target on offer — but those extra points are paid for in days the capital sat unavailable for the next swing, which is the whole premise of QSR ("Quality Swing **Recycle**").

| TP % | Buy events | Win % | Exp %/trade | Avg days held | **Exp %/day** | Hit TP | Stopped | Timed out |
|---|---|---|---|---|---|---|---|---|
| 4% | 115 | 80.0% | +1.93% | 4.1 | **+0.474%** | 92 | 20 | 3 |
| 5% | 115 | 78.3% | +2.37% | 4.8 | **+0.490%** | 85 | 22 | 8 |
| 5.5% ← current | 115 | 78.3% | +2.66% | 5.1 | **+0.517%** | 81 | 22 | 12 |
| 6% | 115 | 75.7% | +2.71% | 5.6 | **+0.481%** | 75 | 22 | 18 |
| 7% | 115 | 72.2% | +2.82% | 6.7 | **+0.422%** | 64 | 22 | 29 |
| 8% | 115 | 70.4% | +2.88% | 7.6 | **+0.381%** | 54 | 23 | 38 |
| 9% | 115 | 68.7% | +2.62% | 8.5 | **+0.309%** | 44 | 25 | 46 |
| 10% | 115 | 61.7% | +1.94% | 9.4 | **+0.207%** | 36 | 29 | 50 |

- Best by **%/trade**: 8% (2.88%/trade, avg 7.6 days held)
- Best by **%/day** (capital efficiency): **5.5%** (0.517%/day, 2.66%/trade over 5.1 days)
- Current 5.5%: 0.517%/day (2.66%/trade over 5.1 days)

## Symbol-day clustered (the honest significance view)

115 buy event(s) collapse to **55 independent symbol-day(s)** (2.1 per cluster). Means below are per symbol-day; the CI is computed on those, not on the raw rows.

| TP % | Symbol-days | Mean %/trade | 95% CI | Differs from zero? |
|---|---|---|---|---|
| 4% | 55 | +1.11% | -0.17% … +2.39% | **no** |
| 5% | 55 | +1.18% | -0.27% … +2.63% | **no** |
| 5.5% ← current | 55 | +1.40% | -0.10% … +2.89% | **no** |
| 6% | 55 | +1.36% | -0.19% … +2.91% | **no** |
| 7% | 55 | +1.36% | -0.28% … +3.00% | **no** |
| 8% | 55 | +1.34% | -0.36% … +3.04% | **no** |
| 9% | 55 | +1.18% | -0.59% … +2.96% | **no** |
| 10% | 55 | +0.92% | -0.94% … +2.77% | **no** |

Best mean at symbol-day level: **5.5%** (+1.40%/trade) vs current 5.5% (+1.40%/trade). **Read the CIs before acting** — if they overlap heavily, the levels are not distinguishable on this sample regardless of which mean is highest.

## Per symbol

Sample sizes here are small (docs/METRICS.md wants ~20/bucket) — directional only. `supplied` is the screener's own `riseAvgPct` for that ticker where it was reachable; `best` is the highest-expectancy level in the sweep.

| Symbol | Buy events | Supplied | %/day @4% | %/day @supplied | Best level (%/day) | %/day @best | Avg days @best |
|---|---|---|---|---|---|---|---|
| AEM | 2 | – | +0.550% | – | 10% | +1.000% | 10.0 |
| AEP | 1 | – | -0.313% | – | 4% | -0.313% | 13.0 |
| AMAT | 1 | – | -1.333% | – | 4% | +4.000% | 1.0 |
| AMZN | 1 | – | +5.500% | – | 7% | +7.000% | 1.0 |
| APH | 2 | – | +0.878% | – | 5% | +0.937% | 4.5 |
| APP | 3 | – | +1.833% | – | 9% | +2.250% | 4.0 |
| ARGX | 3 | – | +0.688% | – | 10% | +1.250% | 8.0 |
| AZN | 3 | – | +0.193% | – | 4% | +0.300% | 13.3 |
| CARR | 6 | 6% | +2.063% | +2.000% | 9% | +3.000% | 3.0 |
| CIEN | 3 | – | +0.600% | – | 4% | +3.000% | 1.3 |
| COHR | 2 | – | -8.000% | – | 4% | -8.000% | 1.0 |
| COIN | 4 | – | +0.647% | – | 4% | +1.600% | 2.5 |
| DDOG | 3 | – | +5.500% | – | 8% | +8.000% | 1.0 |
| EBAY | 4 | – | -0.160% | – | 4% | +0.344% | 3.8 |
| ENB | 3 | – | -0.345% | – | 4% | -0.345% | 8.3 |
| HDB | 1 | – | -0.219% | – | 4% | -0.219% | 19.0 |
| HON | 1 | – | -0.889% | – | 4% | -0.889% | 9.0 |
| IBKR | 3 | – | +3.300% | – | 5% | +5.000% | 1.0 |
| ISRG | 1 | – | +5.500% | – | 5.5% | +5.500% | 1.0 |
| ITUB | 1 | – | -0.765% | – | 4% | -0.765% | 7.0 |
| LHX | 3 | – | +1.179% | – | 4% | +2.400% | 1.7 |
| LLY | 3 | – | +2.750% | – | 4% | +4.000% | 1.0 |
| META | 5 | 8% | +1.833% | +2.667% | 8% | +2.667% | 3.0 |
| MO | 3 | – | -1.318% | – | 4% | -1.318% | 4.3 |
| NFLX | 2 | – | +1.375% | – | 6% | +1.500% | 4.0 |
| NKE | 2 | 5% | -0.577% | -0.577% | 4% | -0.577% | 10.0 |
| NOK | 3 | – | +4.125% | – | 5.5% | +4.125% | 1.3 |
| NVO | 5 | – | +0.529% | – | 4% | +1.250% | 3.2 |
| NXPI | 4 | – | -3.200% | – | 4% | -3.200% | 2.5 |
| PLTR | 2 | – | +5.500% | – | 10% | +10.000% | 1.0 |
| RKLB | 2 | – | -4.000% | – | 4% | -4.000% | 2.0 |
| SNY | 3 | – | +0.611% | – | 7% | +0.778% | 9.0 |
| SPOT | 3 | – | +2.750% | – | 5.5% | +2.750% | 2.0 |
| SU | 2 | – | +1.375% | – | 7% | +1.750% | 4.0 |
| TDG | 1 | – | +1.833% | – | 6% | +2.000% | 3.0 |
| TRGP | 2 | – | +0.423% | – | 10% | +0.769% | 13.0 |
| TT | 6 | – | +1.375% | – | 6% | +1.500% | 4.0 |
| TXN | 1 | – | -0.364% | – | 4% | -0.364% | 22.0 |
| UMC | 2 | 9% | +5.500% | +6.000% | 8% | +8.000% | 1.0 |
| VST | 10 | – | +1.288% | – | 5% | +1.423% | 2.6 |
| WMB | 3 | – | +0.688% | – | 6% | +0.692% | 8.7 |

## Per-ticker targets vs the best flat target

Scored over the same 15 buy event(s). "Per-ticker" = give every symbol its own supplied `riseAvgPct`; the flat rows apply one number to everything. Per-ticker has to beat the best FLAT row to be worth the extra machinery — beating today's 4% isn't the bar, since a flat 6% already does that.

| Policy | Win % | Exp %/trade | Avg days | **Exp %/day** |
|---|---|---|---|---|
| **Per-ticker (supplied riseAvgPct)** | 86.7% | +5.50% | 3.7 | **+1.472%** |
| Flat 4% | 86.7% | +2.70% | 3.0 | +0.899% |
| Flat 5% | 86.7% | +3.56% | 3.5 | +1.009% |
| Flat 5.5% (current) | 86.7% | +4.00% | 3.5 | +1.131% |
| Flat 6% | 86.7% | +4.43% | 3.7 | +1.208% |
| Flat 7% | 86.7% | +5.30% | 3.7 | +1.445% |
| Flat 8% | 86.7% | +6.16% | 3.7 | +1.681% |
| Flat 9% | 86.7% | +6.60% | 3.7 | +1.768% |
| Flat 10% | 60.0% | +3.96% | 8.9 | +0.447% |

**Verdict:** per-ticker does NOT beat the best flat target (flat 9%) by -0.295%/day. On this sample the simpler flat number wins — per-ticker targeting isn't earning its complexity.

## Cautious ladder for tickers with a supplied target

dan's ask: test the supplied target and one/two points BELOW it, rather than assuming the average up-leg is reachable every time (targeting the mean means missing roughly half the swings by construction).

Ranked by %/day, same reasoning as the combined table above.

| Symbol | Supplied | %/day @supplied | %/day @−1 | %/day @−2 | Best of the three | vs current 4% |
|---|---|---|---|---|---|---|
| CARR | 6% | +2.000% | +1.875% | +2.400% | **4%** | better (+0.337%/day) |
| META | 8% | +2.667% | +2.333% | +2.000% | **8%** | better (+0.833%/day) |
| NKE | 5% | -0.577% | -0.577% | – | **5%** | worse (0.000%/day) |
| UMC | 9% | +6.000% | +8.000% | +7.000% | **8%** | better (+2.500%/day) |