YUCLAW v5.3.3 Signal Review
Validation LabSMH Evidence LensXLK Evidence LensCanada Resources EvidenceForward TrackingSignal ReviewExplorerSectorsGitHubPyPILedgerMethodologyHome

Bring your signal.
YUCLAW tries to break it before you trust it.

A registered, reproducible research review — statistics locked before computation, adverse results delivered as measured, a bundle that reruns every number without us.

pip install yuclaw && yuclaw intake-check your.csv
start here — runs on your machine, sends nothing
Research and education only — not investment advice. Signal Review produces research classifications, never recommendations. It will not tell you what to buy or sell, and it will not predict tomorrow's price.

How it works — five steps, no uploads

1🔍
Pre-check your file locally
One command, on your machine. Sends nothing.
more

pip install yuclaw && yuclaw intake-check your.csv — the exact intake rules, run on your machine. Your data stays yours until you engage; this page has no upload box by design.

2📅
Request a slot
Contact + a short scope call.
more

Working path today: open a GitHub issue titled "Signal Review slot request" with no data attached — [CONTACT CHANNEL PLACEHOLDER — published after counsel sign-off on the delivery path].

3🔒
Protocol locked before computation
You receive the spec and its hash.
more

Your engagement's statistical specification is registered in a client-namespace tamper-evident chain first; you receive the spec and its hash before any number is computed.

4🖥️
Sovereign analysis, custody-enforced
Local hardware. Your data never enters our repo.
more

All inference runs on our own local hardware. Client data never enters the public repository or backups — enforced by an automated custody gate, not policy prose.

5📦
Memo + bundle + findings session
Every number cites its protocol.
more

Every number in the memo cites its protocol; the reproduction bundle reruns the analysis end-to-end without us.

yourCSVlocalintake-checkscopecallprotocol locked (hashto you)sovereign analysis(custody gate)memo + bundleto youyou rerun everything without us

Founding pilot tiers

Founding Pilot A
Signal validation core
CAD 2,500 fixed
  • ✓Locked protocol registered before computation (you receive the spec and its hash)
  • ✓Signal decomposition suite: rank association with later returns with confidence intervals, quantile ordering, churn, horizon decay, placebo
  • ✓Methodology note (how to read every number)
  • ✓Reproduction bundle (rerun our numbers without us)
  • ✓Research memo covering the suite results
FULL REVIEW
Founding Pilot B
Full signal review
CAD 5,000 fixed
  • Everything in A, plus:
  • ✓Basket event-study panel with cluster-robust intervals
  • ✓Falsification battery (date-shuffle, sign-flip, placebo)
  • ✓Coverage and exclusion anatomy (what was measurable and why)
  • ✓30 days of written questions + a findings session
fixed fees only · no contingent, performance-linked, or asset-based fees · no payment is collected on this site — invoicing follows the scope call

What you receive

The memo — first page
Signal Review — Research Memo
protocol <hash> · registered before computation
1 · Data Integrity
2 · IC with CIs
3 · Quantile Ordering
4 · Falsification
5 · Limitations
6 · Reproduction

Layout illustration of the memo's section structure — every number in a delivered memo cites its registered protocol.

The bundle — actual contents
client_deliverable.zip ├── CLIENT_MEMO.pdf ├── CLIENT_MEMO.md ├── METHODOLOGY.md ├── bundle/ │ ├── client_input.csv │ ├── client_signals.csv │ ├── manifest.json │ ├── registry_client.jsonl │ └── rerun.sh ├── SHA256SUMS └── README_VERIFICATION.md

bash bundle/rerun.sh recomputes the delivered numbers from the bundled inputs and prints REPRODUCTION OK on an exact match.

One honest chart
0k=1-0.006k=2-0.029k=5-0.004k=10+0.021k=20+0.028
SYNTHETIC DEMONSTRATION — a null signal, correctly diagnosed as null

Rank-association (IC) by horizon with bootstrap confidence intervals, rendered from the synthetic dry-run's actual output. Every interval straddles zero — exactly what an information-free signal should show, and exactly what the suite reported.

The honest ceilings

🚫 Standings are not purchasable. Client engagements cap at EXPLORATORY (CLIENT); the public record's higher standings cannot be bought — at any price.
📏 Under 15 names per date = underpowered, disclosed up front — honesty about sample size, not a defect in your signal.
⚖️ Adverse results are delivered exactly as measured. Nothing is softened, decorated, or withheld.
🧹 Unrequested columns are rejected automatically. We never carry client data we did not ask for.
A defensible null — "your signal shows no measurable information at this sample size, and here is the power analysis that says so" — is a result we are proud to deliver. It may save you far more than the fee.

Who this is for

For

Anyone holding a signal, a screen, or a systematic idea who wants an independent, registered, reproducible read of what the evidence shows — and what it cannot show at the current sample size.

"If you want tomorrow's stock price, we are deliberately the wrong vendor."

Not for anyone seeking predictions, price targets, trade timing, portfolio construction, or advice. No standing above EXPLORATORY (CLIENT) is available, and no recommendation will ever be part of a deliverable.

Request a slot

Contact + a short scope call. Working path today: open a GitHub issue titled "Signal Review slot request" with no data attached — [CONTACT CHANNEL PLACEHOLDER — published after counsel sign-off on the delivery path].

Research and education only — not investment advice. Signal labels and review outputs are research classifications, not buy/sell recommendations. YUCLAW is not a registered investment adviser. Past results — in-sample or forward-tracked — do not predict future performance. Engagement terms are subject to counsel review before any engagement is accepted.

YUCLAW · Home · GitHub

Updated 2026-08-12