Blaque Baux Balanced

Margin of safety — is value investing still a real edge, or a lost decade?

Balanced is a member of the Blaque Baux family. The core repo is the engine and blueprint — a governed, systematic platform (Julia) with a venue-agnostic execution controller and a Layer-3 live-money safety gate. Balanced points that engine at the oldest question in the book — does cheap beat dear? — and inherits the governance wholesale.

Not investment advice. Educational/research software. Nothing here is validated. See LICENSE.

git clone --recursive https://github.com/blaquebaux/balanced.git
julia --project=engine -e 'using Pkg; Pkg.instantiate()'   # one-time engine setup

The thesis

Graham's margin of safety — buy below intrinsic value so you're protected when you're wrong — is the foundation of value investing. But value spent the 2010s losing to growth (the "value is dead" decade), then roared back in the 2022 rate shock. So the honest question isn't "is value good?" but "is value still crucial to safety, and when?" This is the value factor (cheap vs expensive), distinct from the buffett sleeve's cheap-safe-quality blend — here we isolate value vs growth itself, and ask whether it's an unconditional edge or a regime-conditional one (a natural conditional-keeper case: value tends to win when rates rise).

Data honesty — clean, ETF-based. The value/growth factor is directly priceable: IVE/IVW (S&P value/growth), VTV/VUG (Vanguard), and the pure factor RPV/RPG (Invesco pure value/growth) which strip the blend. Rates via IEF/SHY. No private data; the only limit is that ETF factor definitions are coarse vs a true fundamental cheapness screen — flagged.

Research plan (Path A)

• Value vs growth, risk-adjusted. RPV vs RPG (and IVE/IVW) over the full cycle: Jensen's alpha, M², Jarque-Bera, drawdown — does value earn a risk-adjusted premium, or just a different beta?
• The margin-of-safety test. Does value's cheapness cushion drawdowns (shallower left tail) even when it lags on return — i.e. is "safety" the real product rather than alpha?
• Regime-conditional value. Value vs growth conditioned on the rate regime (IEF trend): is value a conditional keeper — on when rates rise, off when they fall?

Research — first pass done

Full detail in research/README.md. Scorecard (Alpaca SIP, 2016–2026, vs SPY):

#	Question	Verdict
1	Does value earn a risk-adjusted premium?	❌ value lags growth — a lower/different beta, not alpha (RPV α −2.3% / M² −4.9% vs RPG −3.4% / −3.9%; spread −1.9%/yr, corr +0.68)
2	Is "margin of safety" real (shallower tail)?	⚠️ mostly a myth — pure value's maxDD −51% vs growth −37%, worse skew (−0.59); value crashed harder (2020 cyclicals). Its only safety: lower downside vol + cushioning growth-specific selloffs (diversification, not absolute downside)
3	Is value a conditional keeper (rising rates)?	✅ clearly regime-conditional — value−growth +12.3%/yr rising rates, −12.3%/yr falling; but a rate-timed rotation (α +1.8% / M² −0.5%) beats both styles yet doesn't clear the SPY hurdle net of cost
4	Does the published signal earn its keep on a naive growth book?	✅ naive QQQ PASS 3/3 — DD −35%→−29% (18% cut), Sharpe +0.95→+1.04, keeps 90% (managed broad's QQQ failed — it self-manages). The overlay earns it where the book is unmanaged and its worst DD is rate-driven

The synthesis: "Is value still crucial to safety?" — not the way the folklore says. As a factor this decade, value was a lower-vol-with-worse-crashes rate bet in disguise: it lagged growth risk-adjusted, crashed harder than growth (−51% vs −37% — the margin-of-safety claim inverted), and its one clean, strong signal is rate-conditional (+12/−12%/yr) but too coarse to beat the index net of cost. A conditional-keeper ingredient, not a standalone edge. Margin of safety belongs at the security level (Graham's bottom-up discipline), not the top-down value ETF — which is exactly what separates this from the buffett blend and the bogle hurdle.

Status

Research: first pass complete — value is a rate-regime bet, "margin of safety" mostly a myth at the factor level. Value lagged growth risk-adjusted, crashed harder than growth (−51% vs −37%), and its strong clean signal is rate-conditional (+12/−12%/yr) but too coarse to beat the index net of cost. A conditional-keeper ingredient, not a standalone edge. No trading driver — it publishes its signal for the family (below).

Live — publishes the rate regime (rate_regime.txt)

The rotation isn't a standalone keeper, but the signal underneath it is clean and strong — so, exactly as bonds/brics/benchmark publish their regime reads even when their own trading edge is marginal, balanced publishes the rate regime as its real product. live/balanced_rate_emitter.py writes ~/.config/blaquebaux/rate_regime.txt — the rising/falling-rate state from IEF's 100d trend, with the validated implication (value_tilt=value when rates rise, growth when they fall). Any value-sensitive sleeve can consume it; read-only on prices, it writes only the regime file.

python3 live/balanced_rate_emitter.py           # publish ~/.config/blaquebaux/rate_regime.txt
BB_DRYRUN=1 python3 live/balanced_rate_emitter.py   # print only, write nothing

The published signal is the family's 4th regime read (alongside bonds' stock-bond correlation, brics' dollar trend, and benchmark's market internals). Honestly labeled: the regime is real; the coarse rotation it drives doesn't beat the index alone — it's an ingredient for a consumer to combine.

The signal is validated to earn its keep — on the right consumer. research/balanced_4_naive_growth.py tests it on a naive growth book: de-risking buy-&-hold QQQ when rates rise passes the full family bar (DD −35%→−29%, a 18% cut; Sharpe +0.95→+1.04; keeps 90% of return). The same signal on managed QQQ (broad, trend + vol-target) cut 0% off drawdown and shipped opt-in — because broad already spent the drawdown protection. The law (benchmark #4): a de-risking overlay's value ∝ how unmanaged the book is — and here, also whether the book's worst drawdown is actually rate-driven (QQQ's was 2022; RPG's was 2020, so RPG is mixed).

About Blaque Baux

Blaque Baux is a quantitative research initiative and a subsidiary of Carter Warrens. BlaqueBaux.com is the home for the work; the code lives here on GitHub — open to study, test, and build bespoke strategies on top of.

Anyone can point an AI at a market. The edge is understanding what the data actually says — and turning it into something you can act on. We test relentlessly and put most of it on the record as rejected, with the reason; what survives is built, governed, and validated before it is ever called real. That combination — honest research, reproducible evidence, and execution you can trust — is why Carter Warrens leads on strategy and implementation, not merely uses the tools everyone now has.

The Blaque Baux family

This repo is one sleeve of the Blaque Baux family — a single governed engine steered in many directions. The core repo is the base/blueprint and holds the full family roster.

Layout

engine/     the Blaque Baux platform (git submodule -> blaquebaux/base)
research/   _balanced_common.py (loaders + JB/Jensen/M² toolkit) + balanced_1_value_vs_growth / _2_margin_of_safety / _3_regime / _4_naive_growth + scorecard
live/       balanced_rate_emitter.py (publishes rate_regime.txt) + run_balanced_rate.sh + plist
live/       governed live drivers (once a sleeve graduates to paper A/B)

License

MIT. (c) 2026 Carter Warrens.