Research & planning

Every markdown doc in the canslim-sketch repo. Auto-listed at build time; ordered newest-first within each section.

Top-level

13 docs

Session Summary — 2026-05-12 afternoon + post-compact

Sprint 7 SHIPPED end-to-end. Sprint 8 **COMPLETE** — all five primary O'Neil chart patterns now in the library, with a composite orchestrator and a general selector that plugs them into any variant.

SESSION_SUMMARY_2026-05-12.md · 2026-05-12 · 9.1KB
Post-Compact TODO — 2026-05-12 afternoon

He'll be back in ~30-90 min. Massive ingest has ~6h left.

POST_COMPACT_TODO.md · 2026-05-12 · 5.4KB
Sprint 7 Handoff — for post-compact continuation

plus the original (Sprint 6.5b's cb1ac8c performance-stance section).

SPRINT_7_HANDOFF.md · 2026-05-12 · 15.0KB
SPRINT_6_5B_CRITERIA.md — pre-registered bar for the smarter pattern test

---

SPRINT_6_5B_CRITERIA.md · 2026-05-12 · 4.6KB
SUCCESS_CRITERIA.md — pre-registered acceptance bar for Phase 6

---

SUCCESS_CRITERIA.md · 2026-05-12 · 7.8KB
canslim-sketch

A multi-variant CANSLIM-style stock screener + Position Advisor, built on yfinance and Massive PIT fundamentals with a FastAPI/HTMX UI.

README.md · 2026-05-12 · 7.8KB
Phase 6 — Position-Management Simulator + Comprehensive System

---

PLAN_PHASE6_POSITION_SIMULATOR.md · 2026-05-12 · 24.0KB
Framework Notes — what to steal, what to skip

---

FRAMEWORK_NOTES.md · 2026-05-12 · 8.6KB
canslim-sketch — Overnight Report 2 — 2026-05-12

Mr. Matthew —

OVERNIGHT_REPORT_2_FILLED.md · 2026-05-12 · 14.0KB
canslim-sketch — Overnight Report 2 — 2026-05-12

Mr. Matthew —

OVERNIGHT_REPORT_2.md · 2026-05-12 · 10.8KB
Tech Debt + Deferred Work

Anything we defer ends up here. Touch it up rather than let it rot.

TODO.md · 2026-05-12 · 10.5KB
canslim-sketch — Overnight Report — 2026-05-11 → 2026-05-12

Mr. Matthew —

OVERNIGHT_REPORT.md · 2026-05-12 · 22.4KB
Architecture

How the pieces fit together. Optimized for cloning and adding on to.

ARCHITECTURE.md · 2026-05-12 · 10.4KB

Phase 6 outputs

12 docs

Sprint 5 v2 — Variant baseline through SelectionLayerProtocol

Sprint 12 Bonferroni accounting will treat these as captured trials.

data/phase6_outputs/sprint5_v2_selectors.md · 2026-05-12 · 2.8KB
Sprint 7 — MinerviniSEPASelector smoke test

| Window | Excess CAGR | CAGR | Trades | Time (s) | |---|---|---|---|---| | 2023 recovery | -20.31pp | +6.64% | 8 | 38.9 |

data/phase6_outputs/sprint7_minervini_selector_smoke.md · 2026-05-12 · 604B
Sprint 6.5b — Smarter pattern test (Trend Template + Base Quality Score)

| Design | Window | Baseline excess | Design excess | Δ excess | Avg passes/day | |---|---|---|---|---|---| | A_trend_template | 2023 recovery | -18.91% | -8.48% | +10.43pp | 78 |

data/phase6_outputs/sprint6_5b_smart_pattern.md · 2026-05-12 · 2.4KB
Sprint 5 — PIT SP500 universe walk-forward

Sprint 4 used a 17-name evergreen subset — names that traded in all three windows. This produced a +95% baseline in 2022 because those 17 names are *exactly* the 2022 survivors (no Carvana, no SiVB, no Bed Bath & Beyond, no FCEL, no PTON crash, etc.).

data/phase6_outputs/sprint5_pit_universe.md · 2026-05-12 · 4.7KB
Sprint 6.5 — Cup-with-handle pattern feasibility spike

Per plan v4 §3 Sprint 6.5: throwaway 2-day exercise. Minimum-viable cup-with-handle detector wired in as a hard filter on the existing factor-only candidate list. NO parameter tuning permitted — parameters frozen from O'Neil's textbook BEFORE seeing

data/phase6_outputs/sprint6_5_pattern_spike.md · 2026-05-12 · 2.2KB
Sprint 6 — Operational foundations + v0 advisor

| Sub-step | Deliverable | Status | |---|---|---| | 6.1 | `scripts/snapshot_db.py` + `CANSLIM_DB_PATH` env override | ✓ |

data/phase6_outputs/sprint6_operational_foundations.md · 2026-05-12 · 3.6KB
Sprint 6.4 — Benchmark panel + Fama-French 6-factor regression

This sprint demonstrates the benchmark + factor-regression infrastructure working end-to-end on real ETF data, BEFORE we run our own variants through it (Sprint 12). SPY's near-zero alpha is the validation check — by construction

data/phase6_outputs/sprint6_4_benchmarks.md · 2026-05-12 · 3.8KB
Sprint 4 — Cross-regime walk-forward + Deflated Sharpe

with N = 18 (variants × windows). **E[max Sharpe over 18 zero-skill trials] = 1.657** — any observed Sharpe below that is presumptively noise.

data/phase6_outputs/sprint4_walk_forward.md · 2026-05-12 · 6.6KB
Sprint 3b — All 5 named variants on 2023

- `canslim/sim/rules.py`: Intent objects (StopUpdate, ProfitTake, PyramidOrder), PositionRules - `canslim/sim/variants/{oneil,minervini,weinstein,livermore,qmom}.py`: per-variant rule packs - Engine pyramid handler with new-21-day-high veto and per-lot tracking

data/phase6_outputs/sprint3b_all_variants.md · 2026-05-12 · 4.7KB
Sprint 3a — Trail + profit-take applied to factor-only

Sprint 1 showed the static 8% stop costs 5.3 pp of CAGR in 2023's recovery. Sprint 3a asks: does adding a **trail** (only fires on real reversals from a high) and/or a **profit-take target** (locks in winners) recover any of that lost alpha?

data/phase6_outputs/sprint3a_trail_profit.md · 2026-05-12 · 4.2KB
Sprint 2 — Which mechanics matter for factor-only over 2023?

---

data/phase6_outputs/sprint2_results.md · 2026-05-12 · 5.5KB
Sprint 1 — Does an 8% hard stop change factor-only's answer over 2023?

---

data/phase6_outputs/sprint1_stops_vs_no_stops.md · 2026-05-12 · 2.8KB

docs/

3 docs

O'Neil Model Review — root cause of 0-trade windows

have something wrong or too tight. Maybe it needs full fundamental data? We want to capture the spirit of the O'Neil system… not convert it into hard gates that no stock can end up passing."

docs/ONEIL_MODEL_REVIEW.md · 2026-05-12 · 8.9KB
markets.msimoes.dev — Planning Document

card-style homepage so Mr. Matthew can read, navigate, and hold the whole system in his head — and start spotting gaps from outside the codebase.

docs/markets.msimoes.dev-PLAN.md · 2026-05-12 · 10.9KB
canslim/ Codemap

_Generated: 2026-05-12T14:54:39_

docs/CODEMAP.md · 2026-05-12 · 18.4KB