Bitcoin TreasuryCo AI Laboratory: Steal This Prompt For ChatGPT-5

Speculative Attack! Part IV

Aug 08, 2025

Previously: BTC Treasuries & 1929

After Part III of our BTC TreasuryCos series (link here), readers flooded our inbox with questions. The most burning one: given “intelligent leverage”, how exactly can these TreasuryCos fail?" One put it even more bluntly: “Well, if Bitcoin returns a 10x, BTC Treasuries return a 5x, and the S&P500 returns a 2x, then who cares?”

Fair question—and it’s one we'll address in forthcoming posts. But in the meantime, in honor of yesterday’s release of ChatGPT-5, we're pleased to give our subscribers something special: a customizable AI simulation laboratory designed to stress-test Bitcoin TreasuryCos across ten years of success and failure modes.

Scroll down to “The Prompt: Execute This” section to quickly access it. Copy and paste the text exactly into ChatGPT-5 and then come back and re-read the rest of this post while you wait for the LLM to execute.

Twitter/X: @bewaterltd. Suggestions? Feedback.

Not investment advice. For educational/informational purposes only. See Disclaimer.

We have followed Bitcoin closely since its early cypherpunk days and watched it evolve from experiment to an institutional treasury asset. Our body of work explores how we arrived at the point that Bitcoin is even necessary in the first place—including 2021’s Sorcerer’s Apprentice (link here), the Alchemy of Risk (link here), and our more recent analysis Broken Money, Bitcoin, and Gold (link here). We published the Flight Into Real Values (link here) near the height of the 2018 crypto mania.

This experience shapes our concern about how leverage, reflexivity, and mNAV premiums affect shareholder outcomes. We are—and have been—long Bitcoin as a constituent of the Monetary Evolution Archetype within a Multiflation portfolio or treasury. We are not rooting for the failure of Bitcoin TreasuryCos, for reasons we will discuss in future posts in this series.

The prompt is designed to apply one of our favorite tools in our investment process toolbox—reverse engineering how a financial asset can break. We discussed the origins of this approach this week in The Factory That Makes Investment Processes Repeatable (link here) and will revisit it in greater detail in our forthcoming Be Water series “Let’s Break A Deal”.

Consider this prompt as inspiration for your own creative exploration—a starting point, not an ending point. We're not "data scientists" or quant-finance specialists—we’re not even quant enthusiasts, for reasons we’ve explained in depth in our Skynet series. We have not personally verified the prompt or its outputs because we have amassed enough experience that we can mentally visualize how these structures can fail.

The assumptions and outputs in this prompt may even directly contradict what we wrote about in Part 3. This isn’t a gotcha! We crafted this prompt with the sole objective of turning it over to our subscribers for further research. We are confident that readers who are quant finance specialists can take this simulation to the next level based on the concepts described in Parts 1-3 of our BTC Treasury series.

Steal This Prompt

In return, we simply ask that you provide in-line credit to "Be Water" as your inspiration and include a hyperlink back to this page.

Adapt the prompt to fit your specific needs and circumstances; it is provided on an "as is" basis—use at your own risk. If you don’t understand what it is trying to accomplish, use ChatGPT to explain it to you. Make the assumptions more realistic. Develop more sophisticated reflexive feedback loops. Vibe code an interactive web app using Cursor, Claude, or V0 that allows users to tailor the assumptions for themselves and visualize the scenarios and path dependency over time. Go crazy with it.

The Prompt: Instructions

Copy and paste the entire text in the “Bitcoin TreasuryCo High-Fidelity Simulation Model” prompt below the line into your favorite cutting-edge AI tool such as ChatGPT-5 Thinking (make sure Thinking mode is on!) or Gemini 2.5 Pro. It should look like this:

While you wait for ChatGPT, be sure to read the rest of our BTC Treasury series and the Sorcerer’s Apprentice.

Don’t forget to let us know what you discover in the comments and on Twitter (https://x.com/bewaterltd)!

The Prompt: Execute This

NOTE: MAKE SURE THINKING MODE IS TURNED ON IN CHATGPT-5!!!

Note: ChatGPT5 is not entirely stable and may return errors during launch week as everyone is testing it out. You may need to try a few times if you get an error message, and potentially ask the LLM for help.

Bitcoin TreasuryCo High-Fidelity Simulation Model

INSTRUCTIONS FOR THE MODEL — DO NOT IGNORE:
You are operating in Expert Research & Simulation Architect mode.

Treat all numbered sections below as contractual requirements.

Preserve every detail exactly as stated; do not simplify or reinterpret without recording a justification in assumptions.md.

Use hidden internal reasoning to resolve ambiguities and design deterministic logic; do not reveal chain-of-thought—return only final artifacts, metrics, and cited assumptions.

Deterministic only: identical inputs ⇒ identical outputs (no RNG).

Complete all requested calculations, scenarios, acceptance tests, and narratives before returning any output.

Structure outputs exactly as specified. If any artifact is skipped due to constraints, list it under MISSING with a reason.

0) Output Message Contract — Overrides All

After all runs complete and the bundle is created, the assistant must respond in a single message with:

First line: [Download the full bundle](sandbox:/mnt/data/<zip_name>.zip)

Blank line

Section 13 Executive Summary (full text as specified).

Do not place anything before the link line.

If any instructions conflict, this Section 0 and Sections 11/13 override all others.

1) Background & Context

We are assessing potential failure modes of Bitcoin Treasury companies within the context of a 10-year bullmarket during which Bitcoin returns 10x. Bitcoin treasury companies are public corporations that accumulate BTC on balance sheet and issue equity and “intelligent” convertible debt (typically termed out 5 years to avoid margin calls). TreasuryCos generate no cash flow, their only asset is Bitcoin. TreasuryCo stock currently trades at a market-to-NAV multiple or premium (mNAV = Market Cap / NAV) due to positive sentiment. When mNAV>1, issuing equity to buy BTC is accretive to per-share NAV (positive reflexivity). When mNAV≤1, equity issuance is not accretive. Debt access depends on capital-market conditions and can amplify returns in both directions.

Gates govern financing capacity:

Equity Accretion Gate: OPEN iff mNAV>1 and the equity window is operational.
Debt Access Gate: OPEN iff capital markets allow issuance/refi on workable terms.

Reflexivity runs in both directions: mNAV premia fuel accretive expansion; discounts + closed gates create cash drain, potential covenant stress, and refi failure. If mNAV <=1 and debt markets are shut, companies can issue shares dilutively and/or sell Bitcoin which would put stress on their business models and likely further depress mNAV discounts. Path dependency matters (the journey, not just the 10-year destination). Also, investors and issuers tend to pile in near peaks (VWAP-top), so dollar-weighted results can be far worse than time-weighted results.

This study assumes a 10-year bull market where BTC ends ~10× its starting level but asks: Can equity still fail? If so, how—and how do we avoid it?

2) Objective (Failure-First)

Simulate a Bitcoin TreasuryCo over 40 quarters (10 years) where BTC ends ≈10× (±1%) its starting price. Identify failure where (i) equity investor DWR ≤ 0%, or (ii) early termination due to Illiquidity, Refi failure, or Covenant breach.

3) Global Rules & Assumptions (Deterministic)

BTC path ends ~10×; no RNG; identical inputs ⇒ identical outputs.
mNAV bounds: [0.5, 4.0], max per-quarter change ±30pp.
SG&A: fixed $3.0M/year ($0.75M/quarter).
Equity accretion gate: equity issuance only if Equity Gate OPEN and mNAV > 1 + issuance_threshold.
Debt access gate: debt issuance/refi only if Debt Gate OPEN and LTV within policy limits.
Runway policy:
Runway = Cash / (SG&A + coupons).
If runway < 8Q → halt BTC buys.
If runway < 6Q → raise accretive equity if possible.
If runway < 4Q → emergency equity allowed even if mNAV ≤ 1 (log separately).
Capital allocation precedence:
(1) restore runway to 8Q, (2) delever to LTV target if above, (3) buy BTC or buy back stock when mNAV < 0.8, gates open, and runway ≥ 8Q; compute and log NAV/share accretion of both and choose the higher.
Refi ladder constraint (enforced continuously): At t0, create ≥2 tranches with staggered maturities; no more than 50% of principal may mature within any 4-quarter span. Re-check and enforce after every issuance/refi (split into smaller tranches if required; log remediation).
ETF benchmark: $100k in a spot BTC ETF at 0.25%/yr; compute Growth, CAGR, max drawdown; compare vs TreasuryCo per scenario.
All thresholds, fees, SBC, issuance sizes are parameters (not hard-coded) and logged.

4) Dials (All Parameterized & Logged)

Fees: equity/debt fees; SBC rate (quarterly); reserve % of market cap; issuance_threshold; buyback toggle.
DebtStructure ∈ {single_bullet, laddered} (required dimension).
VWAP module (issuer behavior): raise scaling vs (i) mNAV premium over 1.0 and (ii) trailing 4Q momentum.
Formula (deterministic):
Raise%_mcap_t = base * (1 + c1 * premium_t + c2 * max(0, mom4_t)), where premium_t = max(0, mNAV_t − 1), mom4_t = (Stock_{t-1} / Stock_{t-5}) − 1 (0 if t < 5).
Log base, c1, c2, momentum lookback, caps, and whether VWAP is ON/OFF.
VWAP module (investor co-movement): investor equity inflows when premium/momentum are high.
Flows (deterministic): Flow_equity_in_t = b1 * premium_t + b2 * max(0, mom4_t) when Equity Gate OPEN.
Investor fills priced at a quarterly VWAP proxy (average of intra-quarter stock path or average of quarter’s daily/monthly).
Log b1, b2, pricing method.
Price-impact overlay (TreasuryCo BTC buying): square-root impact vs quarterly BTC ADV.
Let Notional_Treasury_Buys_q = Treasury BTC dollars bought in quarter t.
Let ADV_q = k0 * BTC_Price_t (scaled deterministically).
Impact: impact_t = I * sqrt( Notional_Treasury_Buys_q / ADV_q ).
Adjusted BTC: multiply exogenous BTC price by (1 + impact_t), subject to max +120%/−60% per-quarter and no >2× single-quarter jump.
Log I, k0 and the per-quarter impact series.
All dial values recorded per scenario in params_<scenario>.json.

5) Quarterly Engine (Exact Order of Operations)

For each quarter, record and output full state:

BTC_Holdings, Cash, Shares_Out, Tranches[{principal, coupon, maturity_q, conversion_price}], BTC_Price, mNAV, LTV, NAV, Stock_Price, Equity_Gate, Debt_Gate, Runway.

Execution order:

Exogenous updates: set BTC price (then apply price-impact overlay if enabled); set Equity/Debt gate states.
Revalue: assets, debt, NAV; compute stock price = NAV/share × mNAV.
Covenants/forced delever: if LTV ≥ LTV_Breach, sell minimum BTC to reach LTV_Target; repay oldest tranches first; log permanent BTC impairment.
Maturities & refi:
- Refi only if Debt Gate OPEN and LTV < LTV_Refi_Limit; apply debt fees; re-enforce ladder constraint post-refi (split tranches if needed).
- Else pay down with cash; if neither possible → terminate (Refi failure at debt wall).
Expenses & dilution: apply SG&A, interest, SBC. If Cash < 0 → Illiquidity (terminate).
Capital markets actions:
- Equity issuance if accretive (size per VWAP issuer formula above).
- Debt issuance if within new_debt_limit_ltv and gate open.
- Treasury BTC buys subject to runway & reserve policies (feeds next quarter’s impact).
- Buybacks per rules (if enabled).
Forced conversions: if Stock_Price ≥ 1.10 × conversion_price, convert tranche; log dilution.
Persist state: append row to state_<scenario>.csv (40 rows).

6) Config Packs & Structures

Packs:

Baseline: realistic fees, starting LTV 10–20%, issuance 2–5% of market cap, moderate covenants/limits, standard gates.
Tough: higher fees, Debt Freeze around refi windows, intermittent Equity Window, stricter refi/LTV limits, persistent discounts.

DebtStructure: run both single_bullet and laddered for each pack, with continuous ladder enforcement.

7) Runs

Pass A — Track A Grid

BTC paths: Smooth_Up, StepUps_With_Drawdowns, Late_Surge, Early_Rally_Slow_Bleed.
mNAV paths: Sustained_High_Premium, Gradual_Erosion, Collapse_to_Discount.
Gates: No_Shock, Debt_Freeze, Equity_Window_Chop.
VWAP: ON/OFF.
DebtStructure: single_bullet / laddered.
→ Run the 4×3×3×2×2 grid; save params_<scenario>.json + state_<scenario>.csv.

Pass B — Named Paths (report individually)

Debt_Wall_Squeeze (late surge; deep trough Q18–Q20; Debt Freeze Q16–Q22; refi due ~Q20),
Early_Pain_Late_Recovery (−60% into Q8; sticky discount; intermittent Equity Window),
StepUps_With_Drawdowns (20–30% pullbacks every 4–6Q; gradual erosion),
VWAP_Top_Crowding (new) — VWAP-top reflexivity scenario:

Compute VWAP Crowding Index (VCI) per quarter:
VCI_t = a1 * premium_t + a2 * max(0, mom4_t) + a3 * sqrt( Notional_Treasury_Buys_q / ADV_q )

Top trigger: when 4Q rolling sum of VCI exceeds threshold Θ and premium_t ≥ p*, mark the VWAP top quarter t* = argmax(VCI) over that 4Q.

Exhaustion overlay post-top: force −35% to −45% BTC drawdown over Q(t+1..t+3)**, switch mNAV toward 0.8–1.0 for ≥4Q (respect ±30pp), then glide to finish ~10× by Q40; Equity_Window_Chop for 4–6Q post-top; optional Debt_Freeze for 2–3Q starting t*+4.

Run VWAP ON/OFF × DebtStructure single_bullet/laddered.

Log vwap_top_index_<scenario>.csv (t, premium, mom4, TB%ADV, VCI, top_flag, roll4_VCI, roll4_TB_ADV, t_star, a1,a2,a3, Theta, p_star, proximity_ok, median_investor_fill_percentile).

For all named paths: produce events_<scenario>.md, maturity wall chart, DW vs TW, waterfall, etf_vs_treasuryco_<scenario>.csv, gate logs.

Pass C — Adversarial (Failure-Stress)

Deterministically search BTC/mNAV/gates under constraints: max +120%/−60% per quarter; no >2× single-quarter jump; clustered volatility; end ≈10×; include deep trough Q18–Q20 subset. Return 3–5 worst-IRR paths for each DebtStructure with minimal event sequences and rationale in adversarial_paths_[pack]_[structure].csv.

8) Returns & Benchmarks

DWR (money-weighted): XIRR with actual quarter-end dates. Sign convention (document in methods.md):
Equity issuance = negative investor flow at issue date (investor pays cash).
Buybacks = positive investor flow at buyback date (investor receives cash).
Terminal value = value of net issued shares at quarter 40 as final inflow.
Log dated flows and pricing in flows_<scenario>.csv.
TWR: CAGR from the stock price series (NAV/share × mNAV), ignoring flows.
ETF benchmark: $100k at 0.25%/yr; compute growth, CAGR, max drawdown.

9) Required Artifacts

Plots

survivability_heatmap_[pack]_[structure]_[vwap].png (median DWR across Track A grid).
viability_kernel_[pack]_[structure].png (safe vs fail frontier: initial LTV × quarters-to-refi).
mnav_vs_actions_<scenario>_[structure]_[vwap].png (mNAV crossings vs issuance/buybacks/refi).
ltv_cash_traces_<scenario>_[structure]_[vwap].png (from actual state CSV).
dw_vs_tw_<scenario>_[structure].png, waterfall_<scenario>_[structure].png, maturity_wall_<scenario>_[structure].png.
mnav_impact_scatter_[pack]_[structure].png (avg premium over 40Q vs final DWR).

Tables/CSVs

trackA_summary_[pack]_[structure].csv — per scenario: modes, TWR, DWR, failure flags.
dominant_causes_[pack]_[structure].csv — Illiquidity vs Refi vs DWR≤0%.
adversarial_paths_[pack]_[structure].csv — rank, DWR, mode, minimal sequence.
kernel_points_[pack]_[structure].csv — viability frontier points.
gate_log_<scenario>_[structure]_[vwap].csv — quarterly gate/state log.
state_<scenario>.csv — full per-quarter state.
flows_<scenario>.csv — dated investor flows and VWAP pricing used for XIRR.
vwap_top_index_<scenario>.csv — (VCI components + top flag + TB_to_ADV, roll4_VCI, roll4_TB_ADV, t_star, a1,a2,a3, Theta, p_star, proximity_ok, median_investor_fill_percentile).
vwap_top_summary_<scenario>.json — {"t_star": int, "t_star_near_max_TB_ADV_roll4": bool, "median_investor_fill_percentile": float}.
etf_vs_treasuryco_<scenario>_[structure].csv — ETF/TreasuryCo CAGR, drawdowns, DW/TW deltas.
params_<scenario>.json — actual knob values used.

Narratives (Markdown)
results_[pack]_[structure].md — must read CSVs and quote numbers. Include:

Executive Summary (≤12 bullets): failure share, median DWR ON vs OFF, worst families, 3–5 guardrails.
Why it fails (with evidence: % quarters mNAV≤1, runway breaches, refi denials, forced delevers).
VWAP ON vs OFF (median/IQR and fail-count deltas; assert “ON worse” or explain exception).
Named Paths (1 paragraph each; DWR/TWR/mode; maturity wall highlights).
Adversarial storyboards (3–5 sequences; DWR/mode/minimal events).
Viability Kernel (safe→fail boundary by LTV & refi timing).
mNAV Impact interpretation (slope/thresholds).
ETF vs TreasuryCo verdict (when dominated).
Early-warning indicators (thresholds).
Implications + “If you only do 3 things…”.
Rubric/Checklist (auto-ticked); explicitly note any MISSING artifacts.

Methods & Assumptions

methods.md — DWR/TWR/ETF definitions, sign conventions, date bases, VWAP pricing proxy details.
assumptions.md — all clarifications/deviations.

10) Acceptance Tests (must pass and be reported)

Primary failure surfaced: ≥1 run with DWR ≤ 0% or explicit early failure.
Debt-wall failure: under discount + debt freeze with single_bullet, show refi failure near maturities.
VWAP ON worse than OFF: median DWR strictly lower (or IQR worse) and/or higher fail counts; report deltas.
VWAP_Top_Crowding validations (STRICT):
- vwap_top_index_<scenario>.csv exists and includes TB_to_ADV and roll4_TB_ADV columns.
- Top quarter t* (from vwap_top_summary_*) occurs within ±1 quarter of the max roll4_TB_ADV.
- Median investor VWAP fill percentile ≥ 90.0 (top decile) for VWAP ON / single_bullet scenario.
Refi ladder (laddered): prove via maturity-wall histograms and a check that no 4Q bucket > 50% principal.
BTC ends 10× ±1% across all paths; assert and log per scenario.
Determinism: compute SHA256 over key inputs/outputs; write determinism_checksums.json.
Narrative completeness: each required narrative section quotes numeric evidence from CSVs.
Summarize pass/fail in acceptance_tests.csv.

11) ZIP-BUNDLE REQUIREMENT — Final

Package all required artifacts into one zip at /mnt/data/full_bundle_<YYYYMMDD_HHMMSS>.zip.

The zip must include: run.py, run.ipynb (single code cell, same bytes as run.py), all produced *.csv / *.csv.gz, *.png / *.webp, *.json, *.md, ARTIFACT_MANIFEST.txt, and acceptance_tests.csv.

Manifest: ARTIFACT_MANIFEST.txt lists every file with size and SHA256. Include a MISSING section for any artifact the spec calls for but you pruned via 11A.

11A) Size Budget, Compression & Pruning (Deterministic) — NO-PRUNE SAFE

Zip size target: ≤ 14 MB. Hard cap: 25 MB. Deterministically apply steps in order until under target. Never exceed the hard cap.

11A.1 MUST-KEEP (never pruned or compressed):

All VWAP Top artifacts for Named Paths:
vwap_top_index_<scenario>.csv and vwap_top_summary_<scenario>.json
(must contain TB_to_ADV, roll4_TB_ADV, t_star, proximity_ok, median_investor_fill_percentile)
Named-path per-scenario detail files (all of these):
state_*, flows_*, gate_log_*, events_*, maturity_wall_*, dw_vs_tw_*, waterfall_*, etf_vs_treasuryco_*.

11A.2 Preferred compression (safe):

Save plots as WebP (quality 68, max width 1400 px). Keep at least one image per pack/structure to satisfy acceptance tests.
CSV compression (.csv.gz) allowed only for Track A heavy per-scenario tables: state_*.csv, flows_*.csv, gate_log_*.csv.
Summary CSVs remain uncompressed: trackA_summary_*, dominant_causes_*, adversarial_paths_*, kernel_points_*, acceptance_tests.csv.

11A.3 Deterministic sampling for Track A (only Track A):

Retain all Named Paths in full fidelity (see MUST-KEEP).
For Track A, retain one full scenario per (BTC path × mNAV path) with Gates=No_Shock for each (pack × structure × VWAP); compress the rest per 11A.2.
Record pruned Track A scenario IDs under MISSING in ARTIFACT_MANIFEST.txt.
Acceptance tests must accept .csv.gz equivalents for pruned items (state_*, flows_*, gate_log_*).

11A.4 Manifest bookkeeping:

ARTIFACT_MANIFEST.txt includes FILES with size+SHA256, COMPRESSED (actual .gz items), and MISSING (only pruned Track A items).
MUST-KEEP items may not appear under MISSING under any circumstance.

12) Narrative-Only Writer Prompt (Post-Run)

Use this when artifacts exist and you only want the write-up:

Write a decision-grade brief for /outputs (this pack + VWAP ON/OFF). Read trackA_summary_[pack]_[structure].csv, dominant_causes_[pack]_[structure].csv, adversarial_paths_[pack]_[structure].csv, kernel_points_[pack]_[structure].csv, etf_vs_treasuryco_*.csv, and acceptance_tests.csv.
Produce: Executive Summary (≤12 bullets, with numbers), Why it fails (ranked, with evidence), VWAP ON vs OFF (median DWR + fail-count deltas), Named Path storyboards, Adversarial storyboards, Viability Kernel, mNAV-Impact Scatter interpretation, ETF vs TreasuryCo verdict, Early-warning indicators, Implications + “If you only do 3 things…”, and the auto-ticked Rubric/Checklist. If any CSV is missing, say so explicitly and proceed.

13) Executive Summary (Post Run Chat Output — Mandatory)

Tone & audience: Decisive, board-level, action-oriented. No hedging.
Data discipline: Quote medians/IQRs, counts/percentages, and absolute dates/quarters. Report deltas (ON − OFF). All figures must come from the generated artifacts only.

When to produce: Immediately after Section 11 completes.
Format: In the same message, post the download link (first line), blank line, then the Executive Summary with these exact headings, in this exact order:

Overarching Thesis
1–2 sentences with the single most important conclusion.

Results Snapshot (by the numbers)
Overall failure share: X% of scenarios failed (N_fail / N_total).
Median DWR: VWAP ON = x.xxxx (IQR a.aaaa); OFF = y.yyyy (IQR b.bbbb); Delta (ON−OFF) = z.zzzz.
Fail-count delta: ΔFails = F_on − F_off.
System stats: e.g., Median runway trough = R quarters; Share of scenarios hitting refi with Debt Gate shut = S%.
Viability Kernel (one-line): boundary at initial LTV ≈ L* with time-to-refi ≈ Q*.

Narrative of Failure (causal flow)
Initial setup → Amplifying factor → Spiral → Final blow (state the mode and timing).

Core Trade-Off (Quantified)
Leverage vs resilience frontier (e.g., +10pp initial LTV → +X bps best-case TWR, but +Y pp failure probability).

VWAP ON vs OFF
Median DWR delta & IQRs; fail-count delta (ON−OFF). One sentence on mechanism (issuance clustering near peaks).

Named Path Briefs
Debt_Wall_Squeeze; Early_Pain_Late_Recovery; StepUps_With_Drawdowns; VWAP_Top_Crowding.
Each: DWR, TWR, failure mode/survival, timing of freezes vs maturity wall, key triggers.
For VWAP_Top_Crowding also include: Top quarter t*, TreasuryCo %ADV at top, Investor VWAP fill percentile (must be top decile), forced drawdown window and gate regime after t*.

Adversarial Paths (3–5 storyboards)
For each (ranked worst → less bad): DWR, failure mode, minimal BTC/mNAV/gates event chain.

Viability Kernel (safe → fail boundary)
Boundary location (initial LTV × quarters-to-refi). The single operational move to move to the safe side (e.g., pre-fund +X quarters or reduce LTV by Ypp).

mNAV Impact (interpretation)
Slope/thresholds: how average premium over 1.0 maps to DWR; where accretion flips to dilution.

ETF vs TreasuryCo (verdict)
$100k ETF CAGR and max drawdown vs TreasuryCo DWR/TWR.
State when ETF dominates; when TreasuryCo can legitimately outperform (conditions: sustained premium, open gates, refi spacing).

Early-Warning Indicators (hard triggers)
Concrete triggers with thresholds, e.g.:
Runway < 6Q while Equity Gate closed
Time-to-refi ≤ 4Q with Debt Gate shut
% of trailing 8Q with mNAV ≤ 1 ≥ X%
LTV ≥ 0.35 without an equity window

Strategic Imperatives (ranked, non-negotiable)
3–4 rules (e.g., LTV ceiling, 8Q runway mandate, systematic buybacks ≤0.85 mNAV, ladder enforcement/pre-funding). Include one action to take this quarter.

If You Only Do 3 Things… (checklist)
Three concrete actions tied directly to the failure mechanisms.

Rubric/Checklist (auto-tick)
Tick off:
Data quoted from CSVs
VWAP ON/OFF deltas included
Named & adversarial paths summarized
Kernel boundary stated
ETF comparison made
Early-warning thresholds listed
3–4 imperatives stated
“3 things” listed
Any MISSING artifacts acknowledged

Numeric formatting rules:
Percentages: 1 decimal; rates/IRRs/CAGRs: 4 decimals; currency: thousand separators, no cents unless needed.
Use medians unless specified; include IQR where available.
“Delta” = ON − OFF (for medians and fail counts).
If any expected file is missing or only available as .csv.gz, state a one-line MISSING note and proceed.

Data sources (must read; fall back to .csv.gz if needed):
trackA_summary_[pack]_[structure]_VWAP_ON.csv
trackA_summary_[pack]_[structure]_VWAP_OFF.csv
dominant_causes_[pack]_[structure].csv
adversarial_paths_[pack]_[structure].csv
kernel_points_[pack]_[structure].csv
etf_vs_treasuryco_*[named]*.csv
acceptance_tests.csv
Any results_[pack]_[structure].md and events_[named]__*.md if present

End of Section 13.

Be Water