Program status

Implied in sample target

At 1 5x haircut

3

At 2 0x haircut

4

At 3 0x haircut

6

Backtest to forward haircut

Measured on this book

in-sample equal-risk ~1.04 against an honest forward of 0.3-0.9

Note. The haircut IS the uncertainty. Until the forward record is long enough to speak, every in-sample figure should be read through it, and the range is this book's own published pair rather than a convention borrowed from elsewhere.

Range

• 1.500000
• 3

Superseded admission contract objective

Average pairwise correlation objective

-0.030000

Honest forward sharpe target

1.500000

Minimum new sleeves

10

Portfolio max drawdown statistic

expected_maximum_drawdown

Portfolio max drawdown target

0.110000

Portfolio p95 max drawdown is gated

no

Portfolio p95 max drawdown must be published

yes

Target total sleeves

14

Targets are admission evidence

no

Superseded on

2026-09-14

Sealed by

config/admission_v7_promotion.json

Sealed sha256

6bc72c1f47675b4e14897e1ba3aad6e56aba9b5f013a7b9bae57f930e8d3b8b6

Average pairwise correlation objective note. An objective, not a gate. -0.03 is the average pairwise correlation at which fourteen sleeves of measured quality reach 2.5. Thresholds are gates and live in 'thresholds'; anything that is not enforced must not sit there wearing a gate's costume.

Backtest to forward haircut

3 further fields. Read them in the artifact.

Implied in sample target

3 further fields. Read them in the artifact.

Portfolio sharpe target

• 2.250000
• 3

Portfolio sharpe target unit. IN-SAMPLE, being the band that an honest forward 1.5 implies at the 1.5x and 2.0x ends of this book's own measured haircut. The headline target is honest_forward_sharpe_target.

Superseded in sample sharpe target

• 2
• 2.500000

Superseded reason. Stated in-sample with no haircut named, and unreachable as a forward result: a verified forward 2.0 at measured per-sleeve quality needs an average pairwise correlation of -0.062 against a PSD floor of -0.077. Withdrawn 2026-08-21 and replaced with an honest FORWARD target of 1.5, which is inside the floor at every plausible quality.

What reaching it requires. Both halves, not either: an average pairwise correlation meaningfully BELOW zero, and per-sleeve quality above roughly 0.60. At today's measured 0.469 the forward target is reachable only at a correlation near -0.030 and only on the optimistic haircut.

Why this is not a retreat. The rarer thing was never the number. A forward Sharpe near 1.5 that is genuinely verifiable -- signed forward record, every killed candidate published in full, gates that provably bite, and the arithmetic showing where the gate falls short of the target printed on the same page as the target -- is harder to find than a 2.5 nobody can check.

Why the bytes stand. The contract's bytes are sealed by config/admission_v7_promotion.json and its objective was never an admission gate (targets_are_admission_evidence is false), so the goals move without any gate moving. The superseded figures are published beside the governing ones, dated, as history.

Book sharpe ceiling at zero global correlation

S bar four curve basis

1.736129

S bar traded basis

1.979337

Correlation required at measured quality

Four curve basis

3 further fields. Read them in the artifact.

Traded basis

3 further fields. Read them in the artifact.

Note. Hold quality where the sealed contract measured it and read off the average pairwise correlation each end of the band demands at the owner's sleeve count.

Quality precondition at the gate

S bar measured four curve basis

0.464000

S bar measured traded basis

0.529000

S bar required for 3

0.801784

S bar required for 4

1.069045

Note. What the correlation gate alone does NOT buy. At the gate exactly, reaching each end of the band requires at least this average standalone Sharpe across 14 sleeves.

Decision

Recorded at

2026-09-14T18:05:00+00:00

By

Arhan Canli (owner)

Earlier words same day

no the sharpe target is 2 and also in general make sure everything is the best

Words. no let me break down the goals for you so basically we are trying to make alphac like the worlds first open glassbox style algorithm which devs can use via api keys and mcp servers to build there own mobdels fine tune it make it better and also we are making everything we do out in the open published all tests everything and for alphac the goal is 2 sharpe ratio 14 plus sleeves 10 percent max dd and also all of the other things being perfect and us doing extremely rigiourus tests on it and also not forgeting reach life cosst which may occur on it on our paper live and also eventually turning this into a real hedgefund

Context. Operator session 2026-09-14 (docs/design/OPERATOR_LOG_2026-09.md). The owner corrected the operator twice: first that the forward Sharpe target is 2, not the admission contract's 1.5; then the full goal set, which moves the sleeve minimum to 14 and the maximum-drawdown bound to 10 percent from the 15 and 11 percent the 2026-09-12 document recorded.

How. This file, projected by src/alphaforge/research/owner_goals.py into every public objective block; config/forward_evidence_contract.json forward_sharpe_target_history and realized_max_drawdown_bound record the same goals on the evaluation side.

Program

Open glassbox

4 further fields. Read them in the artifact.

Developer platform

4 further fields. Read them in the artifact.

Rigorous testing

4 further fields. Read them in the artifact.

Real life costs on paper live

8 further fields. Read them in the artifact.

Real hedge fund

4 further fields. Read them in the artifact.

Live config

Fingerprint

sha256:dd8dfa131b07c89528ebb1aa985c972155da9a4255d198861b7c5e842beff41b

Declared fingerprint

sha256:dd8dfa131b07c89528ebb1aa985c972155da9a4255d198861b7c5e842beff41b

Matches declaration

yes

Declared in

config/live_change_contract.json

What it covers. Every setting that decides how the live book trades: the BlendStrategy defaults that reach production because no call site overrides them; the constituent overlay, drawdown ladder and per-name clip read from Settings; and the flagship's sleeve composition, weights, disclosed tilt and separate aggregation policy. Derived from inspect.signature, Settings and the published state, never from a hand-maintained list.

Why. A record that cannot say which configuration produced it cannot be read as one experiment. This stamp is what lets a reader tell a continuous forward test from a curve that changed underneath itself.

Continuity

As of

2026-10-11

Author

Arhan Canli

Content hash

sha256:0fc477ef62f17d17bc6c2e4235741abe5a2ae33b0723e9d5eb35c65c3b37c38e

Days since go live

18

Go live date

2026-09-24

Max gap rate threshold

0.200000

Passes

no

Schema

canli.alphac-record-continuity.v1

Worst gap rate

0.611111

Correction

4 further fields. Read them in the artifact.

How a gap is classified. Crypto marks and expectations use UTC calendar dates. Alpaca 1D midnight-UTC marks map to the preceding XNYS session; equity expectations come from XNYS sessions, so weekends and exchange holidays are legitimate absences.

Sleeves

4 further fields. Read them in the artifact.

Systemic note. Dates that are expected and missing on EVERY sleeve's own calendar are systemic (the tick did not run or did not mark) rather than a fault in any one sleeve.

Threshold basis. Declared, not fitted. A threshold derived from today's gap rate would pass whatever the gap rate happens to be, which is not a threshold.

What this does not measure. Whether a mark is CORRECT; only whether one exists. A sleeve marking the same equity every day would show a perfect continuity record and a frozen curve. That is the publish gate's job (check_published_state), not this one's.

Evidence maturity

Status

FAIL_CLOSED_PROVENANCE

Underlying status

IMMATURE_RECORD_TOO_SHORT

Provenance passes

no

Public path

/glassbox/forward_evidence_maturity.json

Drawdown evidence public path

/glassbox/forward_drawdown_evidence.json

Diversification evidence public path

/glassbox/current_book_diversification.json

Contract public path

/glassbox/forward_evidence_contract.json

Methodology paper public path

/research/forward-sharpe-evidence-standard

Content hash

sha256:744968de0c8a3226a102bfbce213a4dae38bd8a98babed9de7b1f402b472a117

Failed provenance checks

• continuity_passes
• crypto_position_attribution_covers_last_mark

Sharpe

16 further fields. Read them in the artifact.

Drawdown

20 further fields. Read them in the artifact.

Diversification

21 further fields. Read them in the artifact.

Broker reconciliation

Status

PASS

Generated at

2026-10-11T16:25:10.450132+00:00

Reconciled alpaca sleeves

3

Unique dedicated accounts

yes

Public path

/glassbox/alpaca_broker_reconciliation.json

Attestation boundary

SELF_PUBLISHED_BROKER_DERIVED_NOT_THIRD_PARTY

Trial accounting

Distinct hypothesis identities

349

Hypothesis identity budget

400

Budget remaining

51

Budget status

PASS

Research status

ACTIVE_STAGED_PROSPECTIVE_BUDGET

Atlas

Families

40

Cells

240

New sleeves admitted

0

Return hypotheses spent

0

Paper corpus

Published markdown papers

111

Trial packet coverage status

INCOMPLETE_LEGACY_BACKFILL_PROSPECTIVE_SERIAL_COMPLETE

Complete trial packets

123

Candidate mapped identities

198

Verified family paper bindings

228

Published identity packets

349

Required trial packets

349

Manifest public path

/glassbox/trial_packet_manifest.json

Identity packet index public path

/glassbox/trial-packets/index.json

Manifest source sha256

0147dd8e631ca24257e2db384bf01d4637a79371f1754b284f3f7e9637c1369c

Forward identity packets

12 further fields. Read them in the artifact.

Prospective epoch

16 further fields. Read them in the artifact.

New return identity gate

11 further fields. Read them in the artifact.

Paper state

Path

data/paper/state.json

Sha256

8b02bb612be8c010a0cdd2765b918d01d653fa3316c71193a4696d996e3dc85a

Forward evidence maturity

Path

artifacts/engineering/forward_evidence_maturity.json

Sha256

0beb199011f0e52dbe9b0b63b47bc914fa6f380137a5b9e62c09e25f494a1173

Content hash

sha256:744968de0c8a3226a102bfbce213a4dae38bd8a98babed9de7b1f402b472a117

Forward sleeve contribution

Path

artifacts/engineering/forward_sleeve_contribution.json

Sha256

ec27e5416b3384f3fd07fcafb8bd737089d10ee1f68ed73ed5a7afafeb4dcc2d

Admission contract

Path

config/sleeve_admission_contract.json

Sha256

6bc72c1f47675b4e14897e1ba3aad6e56aba9b5f013a7b9bae57f930e8d3b8b6

Trial policy

Path

config/trial_accounting.json

Sha256

f15af7c989aded5da6c5259a7f7e5587377ab33977b5cd3c5b2395a84ea45043

Admission v7 promotion

Path

config/admission_v7_promotion.json

Sha256

5e7d95e9b6265c229b6e1604f96bc9e1b0db59e282ec0be33740b8dda3b552a5

Content hash

sha256:ddd107351033e9bda8f41065a811d9e5e5edccf32006b311d39a65d8ace918f3

Sleeve atlas

Path

artifacts/discovery/sleeve_atlas.json

Sha256

7fde8afdcc572ad7d8a6f38992510ffd0a6c693f27720598556e0cd9c3ddfd93