01Operational
Research contracts and evidence identity
Source provenance, timestamps, immutable manifests, canonical hashes, and failure boundaries define what a later research claim is allowed to depend on.
Multi-Asset Research Lab
Research infrastructure and its historical lineage.
The Lab documents data contracts, deterministic replay, frozen comparisons, R15 gross-candidate validation, and the source-role runtime now waiting for prospective evidence. ASRA is the current research program; historical findings and execution authority remain separate.
What this infrastructure demonstrates
- 01Verified evidence
- 02Deterministic replay
- 03Funding-adjusted PnL
- 04Frozen validation
- 05Prospective observation
R15 model development
From signal design to comparative validation
R15 brings raw-return prediction, volatility-aware targets and ensemble design into a shared research framework. Results are hypothetical and before all trading frictions; they are distinct from the personal account record.
- R15_RAW_1X
- +1,463.45%
- +108.63% / 1.719
- 3 Oct 2026 / offline verified
Historical model researchR15_RAW_1XLong / short / passZero-friction grossSource-role runtime
R15_RAW_1X returned +1,463.45% over the FULL zero-friction replay with a 1.955 daily Sharpe and -46.03% minute-open maximum drawdown. RECENT returned +108.63% with a 1.719 Sharpe and -27.55% MDD. Commission, spread, slippage, and funding are zero in these cited payoff results, so this remains exposed historical development rather than execution-confirmed profitability.
System progression
Research progression with a separate execution boundary
Each layer can add evidence or operational discipline. None can silently grant the authority of the layer after it.
02Operational
Deterministic replay and PnL
A shared strategy contract replays completed BTCUSDT bars with explicit next-open timing, authenticated Funding, transaction costs, deterministic outputs, and independent reconciliation.
03Retained
R15 signal and payoff research
A raw-return reference connects multi-timeframe predictors to a 15-minute target under explicit timing and hypothetical zero-friction accounting.
04Completed
Volatility-aware and ensemble model development
Target normalization and scale-aligned signal blending produced positive gross results. Fixed paired comparisons kept model selection separate from headline returns.
05Offline validated
Bound source-role runtime
R15 is bound behind a WSS-primary, mandatory asynchronous REST-audit contract with fail-closed quarantine and deterministic replay.
06Not yet passed
Prospective validation
The 3 Oct 2026 checkpoint records readiness for separately governed observation, with zero forward-valid origins at that checkpoint and no live-trading approval.
07Not approved
Paper or live execution
Outside this research system. A research result, retained strategy, or forward state never creates broker authority, sizing, leverage, or an order.
Research operating rules
The result matters because the path to it is inspectable
Simple trading logic is not the differentiator. The harder part is proving what data, timing, costs, state, and decisions actually produced the result.
Freeze before looking
Scientific definitions, source identities, cost assumptions, partitions, and decision gates are frozen before the accepted run begins.
One accounting engine
Historical strategy comparisons share the same causal timing, Funding convention, cost treatment, benchmarks, and reconciliation surface.
Failures stay visible
Discarded features and failed strategy archetypes remain part of the research history instead of disappearing after a survivor emerges.
Forward means after activation
Historical and bridge context cannot be relabeled as prospective evidence. Forward observation begins only after an explicit authenticated handoff boundary.
Non-negotiable boundary
Research state is still not execution authority
These restrictions apply to the research Lab and retained research runtime. Separate execution-system telemetry does not grant research findings automatic authority.
01
No live trading in the research runtime
02
No paper trading approval
03
No private Binance or broker integration
04
No order submission, cancellation, or routing
05
No entry or short permission
06
No leverage or position sizing
07
No automatic strategy approval from historical performance
08
No claim that retrospective return is a live track record
09
No claim that one forward observation validates future profitability