Establish truth
Normalize books, trades, funding, OI, liquidations, wallet activity, freshness, and source provenance.
Research system
Research designed to fail honestly.
TheDuckTrader does not optimize for the number of strategies that trade. It reconstructs market truth and market state, then tests whether any apparent edge survives economics, risk, and time.
Latest program snapshot
Market State is live. Alpha remains unpromoted.
On 26 August 2026, the observe-only Market State canary started on fresh server data with no change to trading behavior.
System governance42 / 21 / 0
42 mapped research components, 21 active observers or inputs, and zero paper or live candidates.
Market State EngineCanary
Liquidity, volatility, funding/leverage, and execution-confidence states are generated separately from alpha.
Wallet identityOOS pending
The primary test is incremental future-risk information beyond anonymous microstructure, not copy trading.
Liquidation truthCanonical feed
Continuous event collection is separated from legacy sampled data before liquidation-state research is allowed.
The research loop
Truth. State. Validate. Reject or retain.
A repeatable process keeps attractive historical results from becoming unearned confidence.
Reconstruct state
Describe liquidity, volatility, leverage, and execution conditions without turning states into hidden trade signals.
Validate
Measure costs, MAE, false positives, tails, clustering, chronological OOS, and external-source agreement.
Decide
Kill, archive, observe, or continue offline research. Paper and live permissions remain separate and blocked.
Current architecture
Not every useful signal is a strategy.
The 42 mapped components are increasingly classified as infrastructure, state features, benchmarks, offline hypotheses, or backlog — not 42 trading bots.
Funding & basis carryBenchmark
Retained to measure executable economics; current evidence remains below realistic cost requirements.
Funding / leverage informationMarket State
Funding, basis, OI, volatility, and liquidation context are being consolidated into economic-state research rather than separate alpha claims.
Order-flow imbalanceFeature
Maintained as conditioning and microstructure input; current directional and basis-response variants are not standalone alpha.
Cross-sectional relative valueOffline
Future work focuses on factor stability and structure change before adding complexity or new pair searches.
Wallet intelligenceResearch
Peer-adjusted markout, persistence, specialization, metaorders, and pre-positioning are evaluated causally before external-provider comparison.
External market truthValidation
Independent public datasets are used to challenge collector completeness and event agreement, not replace raw venue truth.
Current experiment
Does market state separate future risk?
The active canary is intentionally non-directional. The question is whether current market states meaningfully separate future adverse excursion, spread widening, depth deterioration, and volatility before any strategy consumes them.
Liquidity State
Measures depth, spread, thinning, and recovery conditions.
Volatility State
Separates normal, elevated, high, and shock activity regimes.
Funding / Leverage State
Combines funding, basis, OI, volatility, and leverage-stress context.
Execution Confidence
Tests whether visible liquidity is likely to remain usable when execution conditions change.
Evidence gates
What must be true before promotion.
Passing a model metric is not sufficient. The complete decision must remain economically and operationally coherent.
Data integrity
Fresh inputs, explicit timestamps, viable depth, stable schema, source reconciliation, and no silent collector substitution.
Statistical validity
Chronological splits, frozen hypotheses, embargo, calibration, persistence, placebo tests, and independent evidence.
Economic validity
Positive response after conservative costs, controlled false positives, acceptable tails, and sufficient sample retention.
Operational validity
Reproducible versions, explicit permissions, stable outputs, observability, and no silent router promotion.
Current decision: promote nothing.
The platform is functioning correctly by collecting evidence without converting research-state outputs into execution.
Research partnerships
We want independent challenges, not confirmation.
We are interested in conversations with institutional operators, market-microstructure researchers, wallet-data teams, fintech product teams, and quantitative risk specialists.
Independent venue and data validation
Strategy-validation pilots
Wallet-intelligence benchmarks
Risk and market-state workflows