Engine topology
Knowability, transforms, portfolio construction, controls, execution, and output gates.
/00ALPHAC / open research system
CANLI CAPITAL
Four independent sleeves. One governed book. Every decision leaves evidence.
ALPHAC combines equity momentum, managed-futures trend, funding carry, and point-in-time macro surprise at equal-quarter weights. Research is walk-forward, net of costs, and trial-aware. Failed sleeves stay in the public ledger; published states are signed and reproducible from source.
Forward evidenceNot yet established16 paper days accrued
/01Governing status / machine-derived
Ambition, without borrowed certainty.
The target is a verified forward Sharpe of 1.5 with expected maximum drawdown near 11%. The current paper record is too short to establish either. Targets govern admission; they are not displayed as achievements.
- Paper return to date
- -1.64%
- 2026-08-07 to 2026-08-23
- Sleeve frontier
- 4 / 14
- 0 new admissions
- Trial identities
- 228 / 320
- Immutable budget accounting
- Verified trial packets
- 2 / 228
- 228 verified family bindings; 197 candidate matches
Execution3 sleeves use dedicated Alpaca paper accounts; AlphaForge uses the local AlphaForge PaperBroker against live exchange order books; ALPHAC is a derived composite and has no direct broker account.
AttributionDesigned, researched, and published by Arhan Canli.
Standingtargets not yet achieved. paper only.
/01What is honestly true
READING THE HONEST FIGURES
/02The open book
Inspect the system, not the story.
The living publication follows ALPHAC from point-in-time inputs to broker execution and public evidence. It includes the uncomfortable parts by design.
Sleeve monographs
Economic premise, cadence, evidence lineage, weaknesses, and exact place in the book.
Failure ledger
Net results, rejected hypotheses, corrections, trial deflation, hashes, and reproduction.
/03Access
The objective is ambitious. The claims stay narrow.
The governing objective is a verified forward Sharpe of 1.5 through genuine breadth, never leverage or a prettier backtest. The implied in-sample support band is 2.25–3.0, not an earned return. Access reveals the same paper record used by the system, with no curated history and no promised result.