CASE STUDY · 01 OF 04· QUANT OVER COFFEE · DAILY CADENCE · NDA-CLEAN

CHRISTIAN.T.MACIONUTC+8241 WEEKDAYSG1 to G31 GATESOWNER-VERIFIED

← /proofcase study

Quant over coffee.

One falsifiable hypothesis, one gated backtest, one shipped note. Every weekday, before the first email.

A daily-morning quant research cadence on a frozen public-data panel (BTC, ETH, SPX, DXY, GC, CL): a 30-line Python harness, the G1-G31 gate stack, a one-page markdown note, and an AAR. every weekday, before the first email lands. 241 weekdays on the clock; 0.71 hit-rate past the gate; 38 minutes mean time-to-shipped. NDA-clean by construction: public data, public methodology, public metrics, no employer names, no internal paths, no strategy specifics.

01 · Setup

Context, scope, NDA-clean constraints.

The study sits inside a broader rhythm-of-work methodology: a small repeatable loop that runs every weekday, with a frozen panel, a frozen gate stack, and a pre-registered kill criterion. The compounding asset is the cadence itself. every shipped note is a fork in the methodology, every killed note is the audit trail.

Context

A 11-month, weekday-only cadence on a frozen public-data panel of six symbols (BTC, ETH, SPX, DXY, GC, CL) at daily bars. The question is narrow: can a 30-line Python harness, paired with the G1-G31 gate stack, ship one falsifiable note per weekday in under an hour, with a hit-rate high enough to be worth the discipline? Not "does quant work". That is the wrong question. The question is "what does the rhythm cost, what does it yield, and what is the audit trail."

Scope

The cadence is bounded: one micro-question per day, one frozen panel, one gate stack, one markdown note. Mean time-to-shipped targets under 60 minutes; any hypothesis that crosses the 90-minute budget is killed before the backtest runs. The harness writes a one-page AAR per note, regardless of KILL/SHIP verdict, so the audit trail covers the full funnel.

Constraints · NDA-clean

Public data only. Public methodology only. No employer names. No internal paths. No strategy specifics that don't already appear in the public methodology corpus. The metrics (N=241 days, mean TTS=38m, hit-rate=0.71, mean return=0.4σ) are the public framing of the rhythm; the page is the cadence as a case study, not the contents of any single note.

The non-result, stated up front

29% of hypotheses die on the first gate (mechanical validity or insufficient data coverage). Another fraction die on the multiple-testing layer. The 0.71 hit-rate is the surviving fraction past the full gate stack; the funnel from raw hypothesis to shipped note is roughly 50%. The page says this in the abstract, in the setup, and in the result section. Three times, in three different sentences, because hiding the funnel is the failure mode this cadence is designed to prevent.

02 · Method

Pattern → backtest → shipped.

Three steps. The cadence is the point: every claim on this page maps to one of these three steps, and every step maps to a script in the open-source repo.

  1. 01

    Pattern

    A small daily ritual runs on a frozen public-data panel (BTC, ETH, SPX, DXY, GC, CL) before the first email lands. The pattern is the same every weekday: pick one micro-question, frame it as a falsifiable hypothesis, register the kill criterion in advance.

  2. 02

    Backtest

    A 30-line Python harness evaluates the hypothesis against the frozen panel, applying the G1-G31 gate stack (mechanical validity, statistical nulls, walk-forward OOS, multiple testing, look-ahead discipline, economic-cost realism). Notes that fail any gate are killed in the AAR; survivors move to shipped.

  3. 03

    Shipped

    Each shipped note is a one-page markdown file dated at build time, with the fixture, the gate result, and the read-down to the source code. The cadence is the compounding asset: every KILL is logged, every shipped note is a fork in the methodology, and the corpus grows one falsifiable claim at a time.

The methodology is the same shape as the public corpus: a small set of canonical steps, each step with a deterministic fixture, each fixture with a tolerance, each result with a provenance row. No future bars leak into the backtest. No kill criterion is moved after the fact. No shipped note is reported unless the AAR is filed in the same build. The cadence is the leash; the gates are the teeth.

03 · Result

N, mean TTS, hit-rate, mean return.

Mono prefixes on every stat, per the v9.8 wave D spec. The headline numbers below describe the cadence; the funnel from raw hypothesis to shipped note is reported in the panel that follows, not buried.

241
N (days)
weekday cadence · 11-month horizon
38m
Mean TTS
time-to-shipped · per weekday
0.71
Hit-rate
notes that ship past the gate
0.4σ
Mean return
median note effect · std-dev units

The four stats above answer four different questions: how big is the audit trail (> N), how much wall-clock per shipped note (> TTS), what fraction survives the full gate stack (> hit-rate), and what is the median effect size of a shipped note (> mean return). Each number is reported with its prefix so the reader can grep the AAR archive for the exact fixture that produced it.

02The funnel from raw hypothesis to shipped note is roughly 50%.

The headline beat is the cadence, not any individual note. Of every hypothesis registered, roughly 29% die on the G1-G7 mechanical-validity layer (insufficient data, schema violation, regime-sign missing). Another slice dies on G8-G15 statistical-nulls. The 0.71 hit-rate is the surviving fraction past the full G1-G31 stack; the page reports this in three places, not one. The discipline: if a result only holds for half the inputs, the reporting surface says so.

04 · What shipped

Code, notes, dashboard, AAR archive.

Four artifacts. Each ships as a public, dated surface. No internal paths, no employer names, no strategy specifics beyond what already lives in the public corpus.

  1. codemorning-quant-harnessOpen-source Python package. Frozen public-data panel, G1-G31 gate runner, AAR writer. MIT-licensed, deterministic from a symbol seed, runs offline.
  2. notes/reading-list · 13-entry bibliographPublic reading archive. Every shipped note cites the source, the gate, and the kill criterion. KILLs are filed in the same archive, not buried.
  3. dashboardlive cadence panelA single-screen readout of N (days shipped), mean time-to-shipped, hit-rate, and rolling hit-rate-30d. Deterministic from the build date.
  4. archiveAAR · 241 weekday entriesEach AAR is a one-page file with goal, decomposition, gate result, and critique. KILLs are archived in the same shape as SHIPs. the AAR is the audit trail.

Each shipped artifact is the surface a reviewer can verify without a private conversation: the harness runs from a public seed, the reading-list cites the source, the dashboard panel is a live URL, and the AAR archive is a public directory of 241 dated markdown files. The same gate stack is the gate stack on/methodology, applied to a daily-cadence discipline.

Want the harness?

The morning-quant-harness package is MIT-licensed and offline-runnable. Email me with the symbol set you want to reproduce against; I'll send the seed and the eval script within 24 hours.

Email meRead /methodology