Research shortlist0 candidates
Protected owner workspace
Research first. Capital later.
Market data, wallet activity, creator hypotheses, model evidence, and risk controls stay separated so a persuasive story cannot quietly become a trade command.
Server credentials initializing. Run sudo bash scripts/fix-web.sh on your server to generate and display your owner access key.
No wallet secrets.Never enter a seed phrase, private key, exchange password, two-factor code, or API secret here.
Owner research terminal
Market intelligence with execution risk kept in view.
Charts, liquidity, forecasts, wallet evidence, model state, and risk controls share one dense workspace. Research and paper validation remain separate from live execution.
Point-in-time research
Live execution locked
StageLoading
CoinbaseLoading
DEX ScreenerLoading
NewsLoading
HeliusLoading
UpdatedLoading
Multimodal market workspace
Interactive history, forecast fan, structure, and risk
Completed Coinbase candles are inspected alongside an empirical probabilistic forecast, deterministic technical and ICT/SMC mapping, Hyperliquid derivatives context, strategy agreement, and the fail-closed Risk Engine.
Only provider-supported range and timeframe combinations are exposed.
OHLC history + volume + probabilistic fanSelect a market
OHLC + volume
q25-q75
q10-q90
Median scenario
Mouse or touch selects the nearest point. Arrow keys move the locked inspection point. Shift plus arrow pans. Plus and minus zoom. Zero resets the view. Forecast bands are uncertainty ranges, not a certain future path.
Load a market to inspect completed OHLC candles and volume.
No timestamp selected.
Forecast fanScenario distribution by horizon
Each horizon is evaluated separately. The local empirical distribution remains uncalibrated until enough true forward observations mature.
Technical + Market Structure + liquidityDeterministic, timestamped research mapping
BOS, CHoCH, displacement, gap, sweep, and order-block hypotheses appear only when reproducible completed-candle rules are satisfied. Terminology does not create alpha or execution authority.
Derivatives and contextual modalitiesMissing evidence stays visible
Current derivatives context uses public Hyperliquid information. Fundamentals, creator, and wallet evidence never receive fabricated fallback values.
Decision synthesisModel agreement, strategy agreement, and Risk Engine veto
SETUP is still a research state. Directional paper use requires the separate signed eligibility gate, and live authority remains NONE.
Risk Engine has not evaluated this workspace.
Pre-Buy ResearchStructured questions before action
PASS means the current research control has evidence. REVIEW and unavailable states remain explicit instead of being interpreted as favorable.
Overview
Research posture before any trade idea
The deterministic policy combines market state, opportunity evidence, blockers, and risk controls. Paper and live authority remain none unless a separate future milestone explicitly changes that state.
Market statesProbabilistic context
Each market keeps model version, uncertainty, and risk flags visible.
Avoid list0 candidates
Decision blockers0 controls
Risk veto and control coverage remain visible before any paper decision is considered.
Opportunity Board
Compare evidence before opening a thesis
Core markets and discovered tokens share one comparison surface. Missing probability, depth, forward evidence, or point-in-time modalities stay unavailable instead of being replaced with neutral scores.
Opportunity evidence is loading.
| Asset | Direction | Probability | Expected move | Net edge | Execution | Evidence conflict | Regime | Decision | Research snapshot |
|---|---|---|---|---|---|---|---|---|---|
| Opportunity evidence is loading. | |||||||||
DEX pair liquidity is not treated as executable depth. A missing directional forecast stays unavailable. Sorting never converts missing evidence into a zero score.
Pre-Buy Decision PassportContent-addressed research snapshot
Select an eligible row above to freeze the evidence currently available for that asset.
No Decision Passport selected.
Generate a passport to inspect durable storage readiness.
Market context
Regime and probability board
Daily Coinbase candles feed a transparent baseline. Probability is explicitly uncalibrated until OOS evidence supports stronger use.
Cross-asset regime is loading.
Event intelligence
News hypotheses before price attribution
Publisher headlines are tagged by asset and event type. Positive or negative labels are hypotheses for event-study follow-up, not causal proof and not trade commands.
Waiting for timestamped headline evidence.
Waiting for structured catalyst evidence.
Discovery
Altcoin research screener
Loading market guardDEX Screener candidates are scored using liquidity, activity, turnover, flow balance, pair age, valuation proxy, and explicit risk penalties. BTC regime can downgrade a research candidate to monitor without hiding its base score.
| Asset | Chain | Score | Liquidity | 24h volume | 1h flow | Age | Risk | Decision |
|---|
Paid token boosts are treated as promotional context and a risk flag. They do not increase the research score.
Token Intelligence
Explain one candidate before acting on the score
Choose a token from Altcoin Radar to inspect the score components, raw market metrics, risk flags, and regime context behind its current research decision.
Solana on-chain evidenceCurrent ledger context with explicit missing groups
Supply distribution is measured from finalized token-account snapshots. Exchange flows, whale activity, active addresses, bridge flows, staking, protocol usage, and treasury movement remain unavailable until their own reliable histories and entity mappings exist.
On-chain research
Wallet activity lab
Recent wallet activity stays separate from replay evidence. Activity alone cannot establish profitability, timing quality, or replicability.
Tracked wallet research set0 / 5 wallets
Keep the set bounded and device-local. A wallet must be researched first before it can enter consensus analysis.
| Token | Direction | Consensus | Wallets | Status |
|---|
Rantai Replay
Cost-basis-aware wallet replay
Replay values only single-pair USDC swaps that can be reconstructed from wallet transfer legs. Unsupported routes stay unpriced, so incomplete evidence reduces coverage instead of creating synthetic PnL.
| Time | Token | Side | Quantity | Price | Gross USD | Cost basis | Realized PnL | Latency reference |
|---|
Timestamp-bounded latency referencePost-hoc replication evidence
Delayed market references are research-only observations. They do not prove that a copied wallet trade was executable at the displayed price.
| Delay | Target time | Observed time | Reference price | Staleness | Adverse move | Status |
|---|
Point-in-time prediction replayOriginal evidence stays frozen
Replay verifies the signed prediction chain and keeps later outcome evidence in a separate layer. Future information cannot rewrite the original data cutoff or feature snapshot.
Community intelligence
Ideas enter a verification queue
Creator content can suggest strategies, datasets, narratives, and tools. It cannot become model evidence from authority or screenshots alone.
Reviewed creator intelligence0 reviewed
Evidence status and incentive risk stay visible. Direct performance proof is required before a creator claim can influence a model.
Claim audit trail0 claims
Claims remain blocked from model use until their timestamp, data, costs, incentives, and replication test are strong enough.
Creator Skill MatrixEvidence coverage without follower-count scoring
Domain coverage is visible, but numeric skill and consistency scores remain withheld until point-in-time outcomes exist.
Creator point-in-time governancePublication, revision, outcome, and alpha-decay controls
Only exact publication timestamps and matured independently sourced outcomes may support classification. X and Telegram stay unavailable without authorized point-in-time ingestion.
X and Telegram evidencePublic or authorized point-in-time sources only
Missing social access remains unavailable. Popularity never upgrades confidence, and promotion context is risk evidence only.
Creator research queue0 sources
Opportunity map
Automation ideas ranked by research readiness
Each idea must explain where return comes from, who pays for it, what data it needs, and what can cause failure. This is a research backlog, not a profit menu.
Strategy RegistryDeterministic challengers and evidence gates
No strategy becomes champion from return alone. Cost realism, walk-forward stability, forward evidence, paper evidence, and risk controls are required.
Strategy tournamentCost-aware walk-forward challengers on current Coinbase history
Every challenger uses next-open entries, the same cost assumptions, conservative ambiguous intrabar handling, and a NO TRADE baseline. The current leader is not automatically a champion.
Method MinerNormalize explicit rules before backtesting
Missing rules are never invented. A method must specify deterministic entry, exit, risk, timing, liquidity, and execution constraints before it can enter data review.
Creator claim method audit is loading.
Agent boundary
Research team lanes
Each lane explains a different part of the evidence. None can place an order or override the risk gate.
Validation
Cost-aware out-of-sample replay
The baseline fits only on the training period, executes after the signal candle, models round-trip costs, and reports OOS evidence. Historical results remain hypotheses.
Paper validation
Signed decision ledger
Paper decisions are stored on this owner device and signed server-side. The hash chain detects edits, but device-local storage is not deletion-proof or cross-device storage.
Automated paper cycleRun the same evidence and Risk Engine gates without capital
The cycle can record NO TRADE, open a signed paper position, append a current research mark, or close a paper position when its declared horizon matures. Strategy stop and take-profit rules remain explicit research definitions, but this owner-triggered browser cycle does not claim intrabar stop execution. It never sends an exchange order.
Simulated executionRecord paper fills against signed signals
Open and close fills must reference an existing signed signal. Position size, chronology, fees, and chain integrity are validated server-side before the event is signed.
Open position valuationRecord a signed paper mark
A mark updates simulated open PnL from a declared observation price. It is not an exchange fill, and future close cost or slippage is not included.
Demo execution reconciliationCompare paper intent with manually imported demo fills
Generate a template from signed paper fills, replace the template values with actual demo-account fills, then compare execution delay, adverse slippage, quantity, and fees. Never paste an API key or account credential.
| Paper fill | Asset | Expected side | Delay | Adverse slippage | Quantity diff | Fee diff | Status |
|---|
Demo reconciliation measures forward execution differences only. It never grants live trade authority and does not prove future fill quality.
Signed demo execution journalFreeze reconciled demo evidence before capital review
Signing binds the current paper intent and imported demo fill into an append-only device journal. It does not prove broker authenticity, place an order, or grant live authority.
| Recorded | Paper fill | Asset | Side | Delay | Adverse slip | Order ref | Status |
|---|
Only system-signed reconciliation events can enter the DEMO_EXECUTION durable stream. Capital Readiness reads that stream server-side when durable storage is enabled.
| Recorded | Event | Asset | Decision | Size | Price | Paper result | Integrity |
|---|
Realized paper PnL uses signed simulated fills after recorded fees. Open paper PnL uses the latest signed mark after the recorded opening fee and before any future close fee or slippage. These records are not exchange or demo-account execution evidence.
Durable research evidence
Preview before importing device records
The owner can compare a verified device-local chain with managed storage before any append. Conflicts fail closed, retries resume from a matching prefix, and device reads remain active until a separately verified cutover.
Preview has not run.
Preview has not run.
Preview has not run.
Managed storage readiness has not been checked.
Import is an idempotently resumable event sequence. It does not delete device records, switch the active read source, grant paper authority, or unlock live execution.
Model Lab
Immutable prediction journal
Snapshot a current BTC, ETH, or SOL research state before the outcome is known. Evaluation is appended later as a separate signed event, so model changes cannot rewrite the original prediction.
Model RegistryBaselines, challengers, licenses, and benchmark status
External model code and weights are not executed merely because a model is popular. Source, license, security, compatibility, benchmark, forward, and paper gates remain explicit.
Local model tournamentLeakage-safe benchmarks on current Coinbase history
This owner-triggered run compares the local logistic research model and mandatory simple baselines on protected historical targets. The empirical distribution forecast is evaluated separately in the workspace and true forward journal. External foundation-model weights remain out of process until pinned security and license review.
Multimodal ablationOnly point-in-time eligible modalities may enter historical comparison
The current evaluator compares market-only and market-plus-technical variants on a chronological maturity-safe holdout. Liquidity, derivatives, on-chain, wallet, creator, social, narrative, and full-multimodal variants stay unavailable until timestamp-correct history exists.
Outcome evaluationAppend evidence, never overwrite the prediction
Actual return, MAE, and MFE are entered after the declared horizon using point-in-time outcome evidence.
| Recorded | Event | Asset | State | Probability up | Data cutoff | Outcome | Integrity |
|---|
Journal entries are device-local and server-signed in this prototype. They are append-only evidence, not live signals.
Capital Readiness
Evidence gate before any real-money review
This gate combines signed true-forward outcomes, calibration, cost-complete paper trades, durable evidence, champion governance, execution reconciliation, and the Risk Engine. Passing it would only allow a separate small-capital review.
Capital readiness evidence is loading.
Promotion blockersEvery blocking gate stays visible
There is no readiness score that can hide a failed critical control. Sample size, calibration, paper execution, durability, champions, and demo execution must each pass independently.
Thresholds are governance floors for review, not proof of profitability. Live trade authority remains NONE even if every readiness check eventually passes.
Automation Dashboard
Research jobs fail closed before paper or capital
This dashboard distinguishes implemented owner-triggered work from scheduled or durable jobs that still require production configuration. A job marked blocked never silently becomes a trading signal.
Automation readiness is loading.
Daily Brief
Point-in-time evidence summary
Deterministic summary of source state, cutoff, market regime, decision context, opportunities, risk blockers, and automation state. Missing evidence stays unavailable.
Daily Brief evidence is loading.
| Job | Mode | Status | Last success | Last error | Next run | Duration | Freshness | Provider health | Failure behavior | Failure class | Write state | Retry policy | Next action |
|---|
Live trading automation is intentionally absent. The kill switch is logically active because live authority is NONE.
Autonomous Operations
Autonomous Live and Auto Trade Control Center
Web-based remote management and execution observability without needing Remote Desktop. Control autonomous background workers, monitor open positions, and enforce fail-closed risk controls.
Performance analytics0 trades evaluated
Realized profit, win rate, and profit factor calculated across closed live and paper trades.
Real-time cumulative equity curve
0 closed trades
Awaiting closed trade executions to plot equity curve...
Remote configurationRisk limits and execution mode
Changes apply immediately to the background worker without service restart.
Manual order dispatchDirect web execution
Place immediate market orders directly from the web interface without opening terminal or desktop apps.
Notification channelsTelegram and webhook alerts
Receive real-time alerts on order fills, stop-loss triggers, and emergency stops without keeping any app open.
Dispatched alert log0 alerts
Recent Telegram and Webhook notifications dispatched by the execution engine.
| Time | Event | Alert message |
|---|---|---|
| No notifications dispatched yet. | ||
Exchange credentialsUnconfigured
Coinbase Advanced Trade and Hyperliquid L1 credentials for LIVE_ACTIVE order routing. Stored securely in protected local state.
Active positions0 open
Real-time mark-to-market valuation, dynamic stop loss, and take-profit status.
| Asset | Side | Entry price | Current price | Quantity | Unrealized PnL | Stop loss | Take profit | Action |
|---|---|---|---|---|---|---|---|---|
| No active positions open. | ||||||||
Autonomous Signal Radar0 assets scanned
Live market scan across active universe with spread quality, signal confidence, and one-click execution.
| Asset | Signal | Price | Spread | Confidence | Regime | Scanner Status | Action |
|---|---|---|---|---|---|---|---|
| No scanner results available yet. Run a cycle or wait for daemon tick. | |||||||
Execution stream and risk audit0 orders
Complete history of filled, closed, and risk-vetoed order attempts.
| Time | Order ID | Asset | Side | Price | Notional | Fee | Status | Reason / Note |
|---|---|---|---|---|---|---|---|---|
| No order history recorded yet. | ||||||||
Autonomous worker activity0 events
Live stream of autonomous daemon cycles, market scans, and execution events.
| Time | Event | Cycle | Details |
|---|---|---|---|
| No daemon events recorded yet. | |||
Influencer and Social Pulse RadarActive Stream
Real-time market sentiment and alpha calls tracked across X (Twitter) and Telegram crypto communities.
Monitored Influencer Channels (X and Telegram)
Online Internet Model AdaptationAdaptive Online
Models continuously ingest real-time market data, order book depth, and internet sentiment to recalibrate decision boundaries.
Internet Telemetry Feeds:
Coinbase Spot L1/L2, Hyperliquid Perps OI and Funding, X Influencer Stream, Telegram Alpha, Alternative.me Fear and Greed
Macro Batch Multi-Asset ExecutionEXPANSION_BULLISH
Evaluates unified multi-asset macro baskets (BTC, ETH, SOL, altcoins) and macro indicators in every autonomous run.
Macro Multi-Asset Indicators:
BTC Dominance: 56.8% | Aggregate Exposure: 1.00x | Macro Veto: INACTIVE
Hugging Face Foundation Models HubSynchronized
Tracks open-source time-series foundation models and crypto NLP sentiment classifiers from Hugging Face.
Open Model Registries:
Amazon Chronos-2, Google TimesFM 3.0, Salesforce Moirai-2.0, IBM Granite TTM, ProsusAI FinBERT, ElKulako CryptoBERT
Daily Stable Profit Walk-Forward EnginePASS_STABLE_PROFIT
Evaluates models with rigorous backtesting and rolling out-of-sample walk-forward slices to target consistent daily profit.
Walk-Forward Out-Of-Sample Slices:
Rolling 14-day forward test windows verified across independent market regimes
VPS host diagnosticsStandby
Real-time system health, process uptime, memory usage, and storage integrity.
Autonomous worker console and live terminalConnected
Live stdout and stderr stream from background autonomous daemon on the VPS without opening SSH or RDC.
VPS Process Tail (worker.log and server.log)
Initializing live terminal connection...
Governance
Risk and deployment controls
Automation is useful only when failure states are designed before live capital is exposed.
Regime, volatility, liquidity, candidate quality, data freshness, and event risk. Vercel Hobby scheduling is limited to a daily research cadence.
Calibration, OOS performance, walk-forward stability, cost sensitivity, drift, and feature logic are reviewed before any model promotion.
Default simulation guards cap declared loss at 1% per idea, aggregate pending plus open risk at 3%, aggregate notional at 1x paper equity, and leverage at 1x. These are research controls, not live-capital advice.
Future execution must stop on corrupted data, API failure, abnormal spread, abnormal size, loss-limit breach, security incidents, or unexplained behavior.
Portfolio riskSigned paper exposure with explicit evidence gaps
Asset concentration is measured from active signed paper exposure. Chain, stablecoin, beta, correlation, narrative, and liquidity concentration remain unavailable unless explicit metadata or aligned histories support them.
Portfolio risk evidence is loading from the signed paper ledger.
Production configurationSecret readiness without secret exposure
This owner-only evidence reports whether required server-side capabilities are configured. It never returns secret values and cannot grant live execution authority.
Production configuration evidence is loading.