Fortress Script Reference

🏰

Fortress Optimizer

fortress_v26.py — ~5100 lines

Three-archetype optimizer (PIONEER / DIAMOND / VAULT). Selects LC+SP+HP+SC strikes for minimum capital, max leverage.

HTML + JSON IBKR or CSV

Common Usage

Live IBKR — single ticker

python3 fortress_v26.py --live --tickers NVDA --contracts 10

Live — multiple tickers

python3 fortress_v26.py --live --tickers MSTR COIN BMNR --contracts MSTR:5 COIN:10 BMNR:20

From CSV (no IBKR needed)

python3 fortress_v26.py ~/fortress/runs/CSVs/NVDA_20260401.csv --contracts 10

Short-term mode + risk cap

python3 fortress_v26.py --live --tickers XLE --short-term --risk XLE:50k

Sigma analysis + parallel

python3 fortress_v26.py --live --tickers QQQ SPY --sigma --parallel 4

Key Flags

Flag	Description	Default
csv_files	One or more fortress export CSVs. Required unless --live is used. If a ticker appears in multiple files, first occurrence is used.
--contracts	Contracts per ticker. Single number (e.g. --contracts 5) or per-ticker pairs (e.g. --contracts XLP:10 XLU:5 MSTR:10). Mix: --contracts 5 XLP:10 (5 default, override XLP to 10).
--risk	Max loss budget per ticker in dollars. Per-ticker pairs (e.g. --risk MSTR:80000 COIN:50000). Supports k/K suffix (e.g. MSTR:80k = 80000) and m/M (1m = 1000000). Without --risk, ML ceiling defaults to notional (stock × shares). Removes two-pass RM bootstrapping. Single-pass ML-capped search.
--fight-otm	FIGHT mode CC OTM% from stressed stock (default: 0.01 = 1%). Range 0.00 (ATM) to 0.15 (15% OTM). Lower = more premium, more management. Higher = less premium, less management.
--run-label	Custom label for output folder. Routes outputs to runs/{label}/ and generates a combined portfolio survival+FIGHT dashboard. Example: --run-label RESTRUCTURE
--dd-level	Max drawdown level for survival curve (default: 40, max: 90). 5% steps to 40%, then 10% steps beyond. E.g. --dd-level 60 produces [5,10,15,20,25,30,35,40,50,60]. Use 60 for high-beta, 90 for crypto. Resilience thresholds (ARMORED=25%, RESILIENT=15%) unchanged.	40
--drawdown-pct	Drawdown % for stress testing (default: IV-scaled)
--min-income-dte	Minimum DTE for income calls (default: 7)	7
--delta	SC delta cap for income calls (e.g. --delta 0.20). Without this flag, default is 0.35.
--sigma	Show CC sigma analysis in dashboard (e.g. --sigma 1.0). Displays alternative CC at the given sigma floor. Does not affect CC selection, only dashboard display.
--tickers	Filter to specific tickers (e.g. --tickers TSLA NVDA)
--no-dashboard	Skip HTML dashboard generation
--no-browser	Do not open the dashboard in the browser after completion
--no-json	Skip JSON results file
--all	When --contracts has per-ticker overrides, include ALL tickers from CSV (not just the mentioned ones). Unmentioned tickers get the default contract count.
--landscape	Run landscape correlation analysis per ticker (verifies metric independence assumptions)
--short-term	Short-term income mode: uses all LEAPS chains (not just far), defaults delta to 0.50 and min-income-dte to 28. Adds friction model (assignment probability, rebuild cost, net-after-friction) and exit profit to output and dashboard. Designed for aggressive income strategies that accept assignment.
--live	Live mode: connect to IBKR TWS, fetch chains in real-time, run optimizer on live data. Requires ib_insync and running TWS. Use --contracts TICKER:N to specify tickers. Saves fetched data to CSV for reuse.
--host	IBKR TWS host (default: 127.0.0.1)	127.0.0.1
--port	IBKR port (default: auto-detect. Tries 4001 Gateway Live, 4002 Gateway Paper, 7496 TWS Live, 7497 TWS Paper)	0
--client-id	IBKR client ID (default: 99)	99
--account	IBKR account for the --margin whatIf (TIMS maintenance margin varies by account). Default: first managed account.
--parallel	Number of parallel IBKR connections for live mode (v27 default: 1 = mega-batch, one connection, one cascade across all tickers so tier sleeps amortize; CEILING: 4). Values > 4 silently deadlock IBKR Gateway on default-tier accounts (see PARALLEL_FETCH_SPEC.md Lessons Learned #1). Each worker gets its own connection at clientId=base+i.	1
--income-dte	DTE range for income calls in live mode (default: 5 46). Example: --income-dte 7 46
--leaps-dte	DTE range for LEAPS in live mode (default: 180 9999, short-term default: 180 450). Limits which LEAPS chains are fetched. Example: --leaps-dte 180 450
--publish	Publish dashboard to Cloudflare Pages via wrangler. Copies dashboard as index.html and deploys via wrangler. Example: --publish ~/Downloads/fortress-dashboard
--notify	Send push notification via ntfy.sh when dashboard is ready. Install ntfy app on phone, subscribe to your topic. Example: --notify fortress-abhi
--telemetry	v20.10: Always on (kept for backward compat). Scan telemetry exported in every JSON output.	True
--no-audit	Skip auto-generating audit HTML after scan.	False
--margin	Compute REAL IBKR portfolio (TIMS) maintenance margin per archetype via a whatIf simulation (never a live order) and add a Maint. Margin row to the dashboard. Needs a write-enabled TWS/Gateway during US RTH. When it cannot price (off-RTH, no API write access, qualify fail) the row shows NA rather than disappearing. Works in --live and CSV modes (port auto-detects).	False

Output Location


runs/fortress/{TICKER}/latest.html
runs/fortress/{TICKER}/{TIMESTAMP}.html
runs/fortress/{TICKER}/audit_{TIMESTAMP}.html
runs/CSVs/{TICKER}_{TIMESTAMP}.csv  ← live mode only

deploy/fortress/MSFT.html

↗ open full

🏗

LEAPS Builder

leap_builder.py — ~2120 lines

Builds optimal LC+SP+HP fortress structures for new tickers. Bucket A (deep ITM) vs Bucket B (OTM leverage), configurable split.

HTML + JSON Audit IBKR

Common Usage

Single ticker, default split (50/50 ITM/OTM)

python3 leap_builder.py --tickers NVDA:250k

Multiple tickers, explicit mode

python3 leap_builder.py --tickers MSTR:100k:itm COIN:80k:otm

Custom budget split (70% ITM, 30% OTM)

python3 leap_builder.py --tickers GOOG:200k:both --budget-split 70/30

From CSV (skip IBKR)

python3 leap_builder.py --tickers XLE:150k --csv ~/fortress/runs/CSVs/XLE_chain.csv

Ticker Format


--tickers TICKER:BUDGET[:MODE]

MODE options:
  itm   → Bucket A only (deep ITM, income-focused)
  otm   → Bucket B only (OTM leverage play)
  both  → Split between A and B (default)

Key Flags

Flag	Description	Default
--tickers	One or more positions. SIZE = 30k/100k/1m (budget) or 20c (contracts). MODE (optional, NORMAL only) = itm or otm to narrow; omit for both. BREAKOUT mode forces ITM internally (no MODE needed). See epilog.
--csv	Use a saved chain CSV (offline mode, skip IBKR). CSV must include the requested tickers.
--cc-overlay	Scan CC overlay (short-dated calls above safe strike) for the Bucket A winner. Off by default.
--parallel	Parallel workers (live mode only, hard cap 4; default 4). Each worker uses its own IBKR connection. --parallel 1 forces serial mode. Above 4 silently deadlocks IBKR Gateway, do not raise.	4
--min-dte	Minimum DTE for LEAPS chain discovery (default: 270)	270
--max-chains	Number of furthest-out chains to scan (default: 3)	3
--html	Override HTML output path (default: routed via OutputManager)
--audit	Override audit JSON output path (default: routed via OutputManager)
--host	IBKR host (default: 127.0.0.1)	127.0.0.1
--port	IBKR port (default: 0 = auto-detect TWS Live 7496 / Gateway Live 4001 / Gateway Paper 4002 / TWS Paper 7497)	0
--client-id	IBKR client ID (default: 95)	95
--account	IBKR account for the --margin whatIf (TIMS maintenance margin varies by account). Default: first managed account.
--breakout	Breakout mode: score by weighted P&L across sigma-multiples of the stock's expected move over HORIZON months (default: 6). Targets auto-scale by stock IV (UAMY high-vol -> bigger % targets, MU low-vol -> smaller % targets). Forces Bucket A (deep ITM), lowers --min-dte to 60 and scans EVERY chain >= 60 DTE (no max-chains cap). Skips CC overlay. E.g. --breakout (6mo), --breakout 3 (aggressive).
--breakout-target	Override sigma-scaled scoring with a SINGLE fixed-pct target (e.g. --breakout-target 0.50 = +50%). Use when you want a specific scenario, not the IV-adaptive default. Only meaningful with --breakout.
--stop-loss-pct	Score by reward / MTM-loss-at-stop instead of reward / structural-ML. E.g. --stop-loss-pct 0.15 evaluates each structure as if you cut losses when stock drops 15% from entry. Adds a Stop Profile card and re-ranks. Currently only meaningful with --breakout; accepted with --yolo but not yet wired (silent no-op there).
--yolo	YOLO mode: widened OTM ranges (LC 95-300% of stock, SP 30-130%) and tail-weighted scoring (peak at +200% / +300% = 3-4x stock). Bucket A still runs as a context comparison; Bucket B renders FIRST. Use only on high-conviction directional bets (e.g. MARA in a crypto bull cycle). Mutually exclusive with --breakout.
--yolo-cap	Override LC max strike multiplier in YOLO mode (default: 3.0 = 300% of stock). Lower it for low-vol megacaps (e.g. --yolo-cap 1.5 for NVDA so LC max = 150% stock); raise it for hypervol microcaps (e.g. --yolo-cap 4.0). Only meaningful with --yolo.	3.00
--anchor	Anchor all comparisons to a specific chain. Without LABEL: use the FURTHEST chain available (max DTE, gives the thesis most runway). With LABEL: match by DTE ('625d') or expiration prefix ('2028-01'). Restricts the Best Fortress and Best Pure picks to that chain so the Verdict / P&L / Stop Profile tables compare apples-to-apples within one expiry. Adds an 'Anchor Cost Analysis' card showing what anchoring costs (or saves) vs the free-choice score-best winners.
--all	Run NORMAL + BREAKOUT + YOLO in one go and generate a single combined dashboard with a Master Verdict summary at the top + collapsible per-mode sections below. One IBKR fetch shared across all three scans. Use this when evaluating a name and you want all three lenses on one page. Composes with --anchor for cross-mode chain consistency.
--margin	MARGIN MODE: per-chain Fortress comparison normalized to a single contract count (the max across chains) so every chain is apples-to-apples. Two use cases only: Stock replacement (Bucket A) + Moderate-bull (Bucket B). Columns: Delta, Sh Eqv (share-equivalent delta), IC, ML, Margin (Reg-T put-spread, NA outside RTH), BE, and a clickable PnL panel (-50% to +300%). Best pick per bucket = highest weighted P&L. Ignores --all and --anchor.

Output Location


runs/leaps/{TICKER}/latest.html
runs/leaps/{TICKER}/{TIMESTAMP}.html
runs/leaps/{TICKER}/audit_{TIMESTAMP}.json

deploy/leaps/MSFT.html

↗ open full

⚖️

LEAPS vs Fortress

leaps_vs_fort.py — ~1350 lines

Head-to-head comparison of fortress_v26 (4-leg income Fortress) vs leap_builder (3-leg leveraged long) for ONE ticker at ONE max-loss budget. Equal-DTE tables, normalized breakeven-strike CC income for both structures, P&L curve from -50% to +300%, scored recommendation. Read-only.

HTML dashboard IBKR or CSV

Common Usage

Live head-to-head (needs RTH for true TIMS margin, Reg-T fallback otherwise)

python3 leaps_vs_fort.py MU:100k

Size to a TIMS maintenance-margin budget instead of max loss

python3 leaps_vs_fort.py DELL:150k --margin-budget 80k

Offline: one fortress-format CSV feeds both tools

python3 leaps_vs_fort.py MU:100k --csv ~/fortress/runs/CSVs/MU_20260620.csv

With AI narrative on the recommended pick

python3 leaps_vs_fort.py TSLA:200k --ai

Key Flags

Flag	Description	Default
position	Ticker and max-loss budget, e.g. MU:100k (consistent with --tickers / --contracts elsewhere).
--margin-budget	Optional maintenance-margin budget, e.g. 100k. When set, structures are sized to fit this TIMS budget instead of max loss.
--csv	Chain CSV (fortress-export format) used for BOTH tools instead of live IBKR. A single fortress CSV works for both.
--csv-fort	Override the Fortress CSV (must be fortress-export format). Falls back to --csv.
--csv-leaps	Override the LEAPS CSV (its own native chains CSV, for max fidelity). Falls back to --csv.
--fort-contracts	Contract count for the raw Fortress dashboards. Default: estimated from the LEAPS run so the raw dashboards are budget-relevant.
--account	IBKR account for the TIMS margin whatIf in BOTH tools. Maintenance margin varies by account; default is the first managed account.
--no-margin-curve	Skip the real-TIMS margin curve (whatIf at each displayed count +/- 2). Faster, but the inline editor then shows linear margin estimates only.
--ai	Add an Anthropic-generated narrative (needs ANTHROPIC_API_KEY).

Output Location


runs/leaps_vs_fort/{TICKER}/latest.html
deploy/leaps_vs_fort/{TICKER}.html  ← CF staged

deploy/leaps_vs_fort/MU.html

↗ open full

🔍

Covered Call Scanner

cc_scanner.py — ~3070 lines

Sigma-band scoring for CC strikes above safe strike. IER (Income Efficiency Ratio) ranking. Single ticker or batch portfolio scan.

HTML per ticker + portfolio IBKR

Common Usage

Single ticker with safe strike

python3 cc_scanner.py NVDA 480 --contracts 10

Batch portfolio scan from CSV (v3.0 mega-batch, one IBKR connection)

python3 cc_scanner.py --csv positions.csv

Reload previous scan (no IBKR)

python3 cc_scanner.py --from-scan ~/fortress/runs/CSVs/cc_scan_data.csv

Drawdown mode (tighter delta cap 0.25)

python3 cc_scanner.py GLD 165 --contracts 20 --drawdown-mode

Key Flags

Flag	Description	Default
ticker	Stock ticker symbol (single-ticker mode)
safe_strike	Safe Strike of the fortress (single-ticker mode)
--csv	Path to CSV position file for batch/portfolio scan (requires IBKR)
--from-scan	Path to previously exported scan CSV (no IBKR needed, dashboards only)
--output-dir	Output directory for all files (default: ./outputs)	./outputs
--contracts	Number of contracts (single-ticker mode, default: 10)	10
--max-delta	Maximum delta filter (default: 0.40)	0.40
--max-dte	Maximum DTE to scan (default: from CSV per-ticker, or 45 in single-ticker mode)
--min-dte	Minimum DTE to scan (default: from CSV per-ticker, or 2 in single-ticker mode)
--host	IBKR TWS host (default: 127.0.0.1)	127.0.0.1
--port	IBKR port (default: auto-detect. Tries 4001, 4002, 7496, 7497)	0
--client-id	IBKR client ID (default: 30)	30
--drawdown-mode	Enable drawdown mode (tighter DTE preference)
--no-browser	Do not auto-open dashboards in browser
--output	Output HTML filepath (single-ticker mode only)
--parallel	Number of parallel IBKR connections for batch mode (default: 4). Each worker gets its own connection at clientId=base+i. Do NOT raise above 4: parallel=6 was tested and caused silent Gateway deadlock (IBKR stops responding without emitting 354/420). 4 workers x 50 batch size = 200 peak concurrent subscriptions, which this account tolerates reliably.	4
--tickers	Filter --csv to only these tickers (e.g. --tickers MSFT NVDA TSLA)
--score-ledger	Fix 5: resolve matured picks in the calibration ledger via yfinance and print predicted-vs-realized calibration by sigma band. No IBKR needed.
--skip-cc-timing	Skip the automatic cc_timing.py refresh at start of run. By default cc_scanner runs cc_timing first (uses yfinance, no IBKR conflict, ~15-30s) so the per-ticker weekly gate panel reflects today\'s close. Use this flag for fast re-renders when you trust the existing telemetry.

Output Location


runs/cc_scanner/{TICKER}/latest.html   ← per ticker
runs/cc_scanner/cc_scan_PORTFOLIO.html ← master view
runs/CSVs/cc_scan_data.csv             ← reusable scan data

deploy/cc_scanner/PORTFOLIO.html

↗ open full

🎯

CC Manager

cc_manager.py — ~2550 lines

Covered-call income HEALTH CHECK and goal tracker, not an order/strike tool. Reuses fortress_rebuild's cached chains plus the trade-log sheet: 3-month sold-CC behaviour, goal tracking (floor vs ideal), open-CC buyback book, forward candidates and regime overlay. Read-only, no live option scan.

HTML dashboard Cached chains + light IBKR

Common Usage

Standard run (fortress_rebuild must have run first)

python3 cc_manager.py

Pure offline: skip the small live fetch of open-CC marks

python3 cc_manager.py --no-live-marks

Custom goal band + longer lookback

python3 cc_manager.py --floor 30000 --ideal 50000 --months 6

Run and push the dashboard to Cloudflare

python3 cc_manager.py --deploy

Key Flags

Flag	Description	Default
--rebuild-cache	fortress_rebuild last_scan.json to reuse for chains (default: runs/rebuild/last_scan.json)
--no-live-marks	skip the small live IBKR fetch of open-CC marks for Insight 1 (use the cached marks even if the gateway is up)
--csv	active fortresses CSV (default: active_fortresses.csv)
--host	IBKR host for the Insight 1 live mark fetch	127.0.0.1
--account	IBKR account filter (default: all)
--floor	monthly income floor in dollars for the goal tracker (default 30,000)	30000.0
--ideal	ideal monthly income in dollars for the goal tracker (default 50,000)	50000.0
--months	lookback window in months for sold-CC history (default 3)	3
--spread	per-ticker cap (0-1) for the Mix variation; lower = more diversified (default 0.67)	0.67
--bullish	comma-separated longer-term bullish holds to exclude from sell recommendations (default: GLXY,RIOT)	GLXY,RIOT
--max-target	legacy/no-op: Max now sells every available contract, so this no longer bounds it	75000.0
--no-html	console only
--deploy	push the staged dashboard to Cloudflare (runs deploy_cf.sh) after the run
--score	grade past projections vs realised income from the ledger (no IBKR); advisory calibration read

Output Location


runs/cc_manager/PORTFOLIO/latest.html
deploy/cc_manager/PORTFOLIO.html  ← CF staged

deploy/cc_manager/PORTFOLIO.html

↗ open full

✅

CC Check

cc_check.py — ~390 lines

Reads active_fortresses.csv, scans all IBKR accounts, and reports missing or partial covered call coverage.

HTML dashboard IBKR

Common Usage

Default scan (all accounts, active_fortresses.csv)

python3 cc_check.py

Custom CSV

python3 cc_check.py --csv my_list.csv

Specific account only

python3 cc_check.py --account U1234567

Specific port

python3 cc_check.py --port 4001

Key Flags

Flag	Description	Default
--csv	CSV file with Ticker, Contracts columns (default: active_fortresses.csv)
--port	IBKR port (0 = auto-detect)	0
--account	IBKR account ID (omit to scan all accounts)
--max-dte	Max DTE for short calls to count as coverage (default: 180)	180
--no-survival	Skip the stock-price + CC-survival fetch (faster, no market data)

Output Location


runs/cc_check/PORTFOLIO/latest.html
deploy/cc_check/PORTFOLIO.html  ← CF staged

deploy/cc_check/PORTFOLIO.html

↗ open full

🧭

Fortress Rebuild (Lifecycle Manager)

fortress_rebuild.py — ~3100 lines

Read-only lifecycle analyzer. Scans every fortress position across all IBKR accounts and recommends one action per fortress: SKIP, WAIT, ROLL, UNCAP, or REBUILD. Surfaces ride-to-expiry math (stock flat vs stock at CC strike), proactive roll candidates, and stranded-WAIT escalations. Routes all IBKR fetches through module_ibkr_fetch for identical reliability and logging (FROZEN→DELAYED→SNAP cascade, OI backfill, deep-ITM historical fallback).

HTML dashboard IBKR

Common Usage

Scan every fortress across every account

python3 fortress_rebuild.py

Bind to active_fortresses.csv (authoritative Entry / Safe Strike / Net Debit)

python3 fortress_rebuild.py --csv

Filter to specific tickers

python3 fortress_rebuild.py --tickers COIN,MSTR,IREN

Horizon cutoff — only fortresses with HP expiring by date; CC scan window tightens to match

python3 fortress_rebuild.py --expiry 2026-09-30

Single account

python3 fortress_rebuild.py --account U18827291

Fast mode — skip proactive roll fetches (useful for triage on slow IBKR)

python3 fortress_rebuild.py --no-rolls

Re-render from cached scan (no IBKR) — iterate on decision logic offline

python3 fortress_rebuild.py --cached

Decisions Produced

          SKIP — structurally safe, no action needed.

          WAIT — margin OK but borderline; monitor. Borderline variants show theta/trip-wire narrative.

          WAIT (MARGINAL) — sigma is thin; roll candidates fetched proactively to find an escape.

          ROLL CC — concrete roll candidate identified with credit, new sigma, and survival.

          UNCAP — close the short call (strike capped, no forward roll viable).

          REBUILD — structural damage; suggests going to leap_builder or fortress_repair.

Ride-to-Expiry Math

          Every decision shows the "do nothing" baseline as explicit P/L:

          CLOSE P/L — close fortress now at market mids.

          STOCK FLAT — hold through CC expiry, stock unchanged. CC extrinsic decays to zero.

          STOCK @ $strike — hold through CC expiry, stock rallies to cap. Adds fortress gain using full synthetic delta (LC long + SP short − HP long).

Key Flags

Flag	Description	Default
--host	TWS host	127.0.0.1
--port	TWS port (default 7496)	7496
--client-id	IBKR clientId	25
--account	Single account to scan (default: all managed)
--tickers	Ticker filter. Space-separated or comma-separated (e.g. --tickers COIN MSTR IREN or --tickers COIN,MSTR).
--exclude	Tickers to EXCLUDE from the run. Same syntax as --tickers (space or comma separated). Applied after --tickers, so you can include a set then drop a few (e.g. --exclude META or --exclude META,NEM).
--expiry	Horizon cutoff date. (1) Hedge-put (HP): only show fortresses whose HP expires by this date. (2) Short-call (CC): scan CCs whose DTE is within (cutoff - today). Without --expiry, HP is unfiltered and CC horizon defaults to 90d. Accepts 2026-09-18, 20260918, 17122027 (DDMMYYYY), 18-sep-2026, 2026-Sep-18, 18/09/2026, YYYY-MM, etc.
--csv	Use active_fortresses.csv as structural authority (ticker → LC/SP/HP strikes + Entry/Safe Strike/Net Debit/Contracts). Binds CSV rows to live IBKR legs. Without a PATH, defaults to ~/fortress/scripts/active_fortresses.csv. Without --csv, the script falls back to fuzzy-matching legs in IBKR positions.
--cached	Reuse cached scan data (skip IBKR)
--no-rolls	Skip roll-chain fetches (faster)
--get-cc-timing	Run cc_timing.py first to refresh the weekly-gate telemetry JSON (BB ladder + BB ZONE source). Use when running ad-hoc mid-day so the gate matches current weekly bars. Skip on nightly cron runs — cc_timing is already a separate job there.
--cc-suggest	Append a section at the bottom of the dashboard with short-call candidates for both UNCAPPED fortresses (fresh CC to sell) and SKIP+CC fortresses (roll options). Natural-ladder view with up to three algorithmic picks highlighted: BEST EV, SAFEST VIABLE, and (drawdown rolldowns only) RISKY HARVEST — a below-SS pick that explicitly trades a locked-loss-on-assignment for premium credit.
--cc-suggest-max-dte	Max DTE for CC suggestions (default 20, matches the 'rarely sell beyond 20d' convention).	20
--cc-suggest-scope	Which fortresses to include: BOTH (default), UNCAPPED (gear-mode only), or CAPPED (SKIP+CC only).	BOTH
--cc-suggest-max-rows	Optional explicit cap on candidate rows per expiry. Default: no cap (every strike that survives the 0.05-0.40 delta band is shown and considered for marker selection). Pass an integer to truncate by delta descending.
--json	Also write raw JSON to this path
--html	Also write HTML to this explicit path (OutputManager always runs)

Output Location


runs/rebuild/{TICKER_LIST}/latest.html
deploy/rebuild/{TICKER_LIST}.html  ← CF staged

deploy/rebuild/PORTFOLIO.html

↗ open full

⚡

FIGHT Income Recovery

fortress_fight.py — ~2320 lines

Aggressive drawdown recovery via near-ATM short-dated calls. P&L curves at 8 price points, gap-up scenarios, income target tracking.

HTML + JSON IBKR

Common Usage

From fortress_results.json (most common)

python3 fortress_fight.py --from-results ~/fortress/runs/fortress/NVDA/fortress_results.json

Manual ticker + position detail

python3 fortress_fight.py --ticker NVDA:10:480:20260620 --fight-dte 7 --fight-otm 0.01

Multiple tickers from results

python3 fortress_fight.py --from-results fortress_results.json --tickers NVDA GLD COIN

Custom gap-up + income targets

python3 fortress_fight.py --ticker MSTR:5:350:20270117 --gap-pct 0.15 --targets 0.25 0.50 0.75 1.00

Key Flags

Flag	Description	Default
csv_file	Chain data CSV (skips live chain fetch, still reads IBKR portfolio)
--tickers	Ticker symbols. Space-separated or comma-separated (e.g. --tickers NOW COPX or --tickers NOW,COPX). Contracts, Safe Strike, and ND are read from the --csv file.
--csv	Path to active_fortresses.csv (ticker → contracts, Safe Strike, legs, ND). Defaults to active_fortresses.csv in the scripts directory.
--legs	Optional leg detail (e.g. GLD:HP:410:20260630:7.214). Repeatable.
--from-results	Read positions from fortress_results.json
--all	Portfolio FIGHT mode: scan every ticker in the --csv roster, keep only UNCAPPED fortresses (no active short calls in IBKR, i.e. cc_check NONE), analyze the UNDERWATER ones, and write a PORTFOLIO master dashboard with per-ticker click-throughs.
--host	IBKR TWS host (default: 127.0.0.1)	127.0.0.1
--port	IBKR port (default: auto-detect. Tries 4001, 4002, 7496, 7497)	0
--client-id	IBKR API client ID (default: 99). Parallel workers use client_id+100+N to avoid collisions.	99
--account	IBKR account ID (e.g. U1234567). If omitted, uses the first managed account.
--legacy-fetch	Use the old per-ticker worker fetch (one connection per worker, per-group tier sleeps) instead of the single-connection mega-batch. For parity comparison; the mega-batch is the default.
--parallel	Parallel IBKR connections (default: 4)	4
--eager-escape	Fetch escape-path chains for EVERY flagged call (incl. capped-only WATCH names like GLD/META). Complete but slow: those chains are otherwise deferred to on-demand --tickers T runs to keep the portfolio run fast.
--expiry	HP (hedge put) expiry date. Accepts: 20260630, 30-Jun-2026, 30-jun-26, Jun-2026. Selects which fortress structure to analyze. If omitted, uses the nearest active HP per ticker.
--min-dte	FIGHT CC DTE floor (range 1-45). Default auto-anchors the nearest gamma-safe upcoming Friday (>= 2 DTE): the floor drops from 4 to include it (Tue Fri = 3d, Wed Fri = 2d), else stays 4. Pass a value to override: BELOW 4 catches a near weekly (pin/gamma, warned); ABOVE 4 opens fetch + pick to the full band above the floor (e.g. --min-dte 30 = 30-45d, longer-tenor fewer-rolls).
--html	HTML output path (default: fight_TICKER_TIMESTAMP.html)
--no-html	Suppress HTML output entirely
--json	JSON output path
--force	Show roll decisions for ALL current CCs regardless of sigma
--roi	Override annual ROI on ML for normal income ref (e.g. 0.80 for 80%). Default: IV-based (LOW 45%, MED 75%, HIGH 95%)
--nd	Original entry ND per share (immutable, from trade creation). Overrides IBKR-computed ND after HP rolls. Bare number form applies to all --tickers (e.g. --nd 107.37). Per-ticker form: --nd MSFT:107.37 --nd GLD:137.34

Output Location


runs/fight/{TICKER}/latest.html
runs/fight/{TICKER}/{TIMESTAMP}.html

deploy/fight/PORTFOLIO.html

↗ open full

🔧

Fortress Repair Analyzer

fortress_repair.py — ~2800 lines

Analyzes SP roll-down paths to lower the safe strike. Recommends HP repairs with payback analysis. Post-repair CC sustainability included.

HTML dashboard IBKR

Common Usage

Analyze specific tickers

python3 fortress_repair.py NVDA MSTR COIN

All fortress positions in portfolio

python3 fortress_repair.py --all

Filter by HP expiry, limit payback period

python3 fortress_repair.py --all --expiry 2026-06 --max-payback 0.5

With CC income override

python3 fortress_repair.py XLE --cc-income 0.45 --hp-min-dte 45 --hp-max-dte 90

Key Flags

Flag	Description	Default
tickers	Ticker(s) to analyze (e.g. IREN IWM TSLA)
--all	Analyze all fortress positions in portfolio
--expiry	Filter to positions with HP expiring by this date (YYYY-MM or YYYYMMDD). Repair targets use --hp-max-dte independently.
--hp-min-dte	Min DTE for HP candidates (default: 60). --min-target-dte is an alias (repair_trades.py parity).	60
--hp-max-dte	Max DTE for HP candidates (default: 120)	120
--gap-mult	Max ratio of new_gap vs current gap (default: 2.0). HP candidates whose roll-down would widen the SP-HP gap beyond this multiplier are rejected. Use 1.5 for tight RM discipline, 3.0 for aggressive roll-down freedom.	2.0
--price-multiplier	Add pick_score ROI component at Nx current stock price. For parabolic movers (crypto miners, high-beta names) where default 4x/5x grid underweights tail outcomes. Must be > 1. Float values accepted (e.g. 7.5, 10).	0
--max-payback	Max payback period in years for repair candidates (default: 1.0). Candidates with payback above this are flagged but still shown.	1.0
--cc-income	Monthly CC income per share override (e.g., 4.00). If not set, auto-estimates from the active SC leg.	0
--maint-roll-dte	DTE threshold for healthy-fortress HP maintenance roll (default: 45). Healthy positions whose HP DTE drops below this trigger MAINTAIN analysis (roll-out to similar/smaller-ML strike) instead of REPAIR (roll-down).	45
--maint-cost-cap	Max ratio of new HP annual cost vs CC annual income for a maintenance roll candidate to qualify (default: 0.25). Keeps the rolled hedge easily covered by income.	0.25
--no-maint	Skip the maintenance-roll pass on healthy fortresses.
--max-analyze-dte	Skip fortresses whose HP DTE exceeds this (default: 120). A hedge with months of runway isn't an actionable roll today; they show as OK.	120
--account	Limit scan to this account ID. Default scans ALL managed accounts and concatenates positions.
--positions	CSV with actual position data (overrides IBKR avg_cost). Required col: ticker. Optional: nd_per_share, lc_strike, sp_strike, hp_strike, sc_strike, contracts, ic, safe_strike
--json	Export to JSON file
--html	Export HTML dashboard
--csv	CSV with safe strikes / ND / contracts / leg strikes per ticker (cols accepted: TICKER, SS, ND, CONTRACTS, LC, SP, HP, SC). Compatible with active_fortresses.csv. Defaults to ~/fortress/scripts/active_fortresses.csv. Pass --csv '' to disable. Values here take precedence over IBKR-derived metrics, AND restrict --all scope to the tickers listed.	DEFAULT
--nd	Net Debit per share override (e.g., 16.0). Single-ticker runs only. Overrides --csv ND and IBKR avg_cost.	0

Output Location


runs/repair/{TICKER}/latest.html
runs/repair/{TICKER}_{TICKER2}/latest.html  ← multi-ticker

deploy/fortress_repair/ALL.html

↗ open full

⚙️

Repair Trades Finder

repair_trades.py — ~3990 lines

Finds concrete 6-leg BAG repair trades (close 3 + open 3) with real IBKR prices. Exhaustive 1k-2k combo search, TIMS margin enrichment, P&L curves.

HTML + JSON Audit IBKR

Common Usage

Basic repair analysis

python3 repair_trades.py MSTR

Specify expiry + contracts

python3 repair_trades.py COIN --expiry 20270117 --contracts 5

With IBKR margin check + CC overlay

python3 repair_trades.py BMNR --margin --cc --gap-mult 3.0

Crypto miner: 3x price projection + FIGHT panel

python3 repair_trades.py MARA --price-multiplier 3 --fight --cc

Custom DTE range for target LEAPS

python3 repair_trades.py NVDA --min-target-dte 400 --max-target-dte 800

Key Flags

Flag	Description	Default
ticker	Ticker to analyze (e.g. MARA)
--expiry	Current position expiry to match (YYYYMMDD or 18-sep-2026). If omitted, auto-detects from portfolio.
--contracts	Override contract count (e.g. 200 for 200 contracts = 20000 shares)	0
--lc	Select fortress structure by LC strike (e.g. --lc 60). Required when multiple structures exist on the same expiry.	0
--html	HTML output path (default: repair_<TICKER>_<timestamp>.html)
--audit	Audit log JSON path (default: audit_<TICKER>_<timestamp>.json)
--gap-mult	Max gap multiplier vs current gap (default: 2.0x)	2.0
--margin	Run IBKR whatIf margin checks on top candidates (requires write access)
--min-target-dte	Minimum DTE for target LEAPS expiries (default: 365)	365
--max-target-dte	Maximum DTE for target LEAPS expiries (default: 1100)	1100
--price-multiplier	Add P&L projection at Nx current price (e.g., 10 for 10x). For parabolic movers like crypto miners.	0
--cc	Add CC income overlay to lens picks (scans DTE 5-50 call chains)
--fight	Add FIGHT income panel: near-ATM short-dated CCs for income while waiting on repair. Shows gap risk tiers at 5/10/20% of ML.
--nd	Override net debit per share (all-in cost basis). IBKR avg cost is wrong after repairs/rolls. Use your real all-in ND. e.g. --nd 2.21	0
--no-roll	Skip roll candidates entirely. Only restructure within current expiry.
--max-candidates	Cap total candidates after merge (default 250000). Trims lowest-score candidates to bound memory and audit JSON size on dense-strike chains. Set 0 to disable.	250000
--pnl	P&L curve only. Connects to IBKR, reads position, prints P&L at expiry to console. No repair analysis. Use with --expiry, --lc, --nd as needed.
--pnl-max	Max stock price for P&L curve (default: 5x current price)	0
--roll-short-put	Roll-down-short-put mode: keep LC/HP, find the optimal lower-strike short put to cut early-assignment risk on a deep-ITM SP. Runs this analysis only and exits.
--rsp-expiry	Expiry for the new short put (default: current SP expiry). Roll the strike down on the same date by default.
--rsp-min-strike	Floor for candidate new SP strikes (default: stock x0.5; always kept above the HP strike).	0
--rsp-max-strike	Ceiling for candidate new SP strikes (default: current SP strike, exclusive).	0
--rsp-cushion	Target extrinsic/carry cushion for assignment-safe (default 1.0; >=1 means the put holder has no early-exercise incentive).	1.0
--rsp-rate	Risk-free rate for the carry / early-exercise boundary (default 0.04).	0.04

Output Location


runs/repair_trades/{TICKER}/latest.html
runs/repair_trades/{TICKER}/audit_{TIMESTAMP}.json

deploy/repair_trades/COIN.html

↗ open full

🧠

Stock Researcher

stock_researcher.py — ~3040 lines

4-pillar quantitative scoring: Fundamentals (40%) + Technicals (15%) + Fortress Fit (30%) + Valuation (15%). AI narrative via Claude API. Hard gates apply.

HTML report

Common Usage

Research single ticker

python3 stock_researcher.py NVDA

Batch research (multiple tickers)

python3 stock_researcher.py MSTR COIN BMNR RKLB HOOD AMD SOFI NVDA PLTR MU

Custom investment horizon

python3 stock_researcher.py GOOG --horizon "2-3 years"

No IBKR (yfinance IV only)

python3 stock_researcher.py MU AMD PLTR --no-ibkr

Scoring Pillars

Pillar	Weight	Key Metrics
Fundamentals	40%	Revenue growth, margins, FCF, debt
Fortress Fit	30%	IV, liquidity, options chain depth, trend alignment
Valuation	15%	P/E, P/S, EV/EBITDA vs peers
Technicals	15%	Momentum, moving averages, RSI

Key Flags

Flag	Description	Default
tickers	Stock tickers
--horizon	Investment horizon (default: '1-2 years')	1-2 years
--output	Output HTML path
--api-key	Anthropic API key (or ANTHROPIC_API_KEY env)
--no-ibkr	Skip IBKR, yfinance IV only
--json	Also dump raw data as JSON alongside HTML

Output Location


runs/research/{TICKER}/latest.html
runs/research/{TICKER1}_{TICKER2}/latest.html  ← multi

deploy/research/SNOW.html

↗ open full

💰

GS Put Seller

gs_put_seller.py — ~1200 lines

Goldman Sachs "Art of Put Selling" screener. Finds optimal puts by FCF yield targets (1x / 2x / 3x monthly FCF). Specific tickers or SPX/QQQ/IWM discovery mode.

HTML dashboard IBKR or yfinance

Common Usage

Specific tickers

python3 gs_put_seller.py --tickers MSFT AAPL NVDA GOOG

Discovery mode (screen index constituents)

python3 gs_put_seller.py --scan --top 10

No IBKR, specific expiry

python3 gs_put_seller.py --tickers QQQ SPY --no-ibkr --expiry 20260515

Capital sizing + force rescan

python3 gs_put_seller.py --scan --capital 500000 --collateral-pct 25 --rescan

Key Flags

Flag	Description	Default
--tickers	Space-separated ticker symbols
--scan	Discovery mode: screen SPX, QQQ, IWM constituents	False
--top	Top N stocks per index in discovery mode (default: 10)	10
--expiry	Target expiration (YYYYMMDD). Auto-selects nearest monthly 20-45 DTE if omitted
--capital	Total portfolio capital for position sizing
--collateral-pct	Collateral % per put (100=fully collateralized, default: 100)	100
--port	TWS port (default: 0 = auto-detect TWS/Gateway)	0
--no-ibkr	Use yfinance instead of IBKR for options data	False
--min-oi	Minimum open interest filter (default: 100)	100
--max-spread-pct	Maximum bid-ask spread as % of bid (default: 15.0)	15.0
--output	Output directory for HTML report
--rescan	Force fresh discovery scan, ignore cache	False

Output Location


deploy/gs_putseller/{TICKER1}_{TICKER2}.html
(also staged to deploy/ for CF Pages)

(no recent output)

↗ open full