AI Insights
Only approved general research context and your own simulated trading results.
User Management
Command Center
Your virtual portfolio, SOL monitoring and research activity in one place. No real orders.
Portfolio performance PAPER ONLY
Real recorded values. Missing history stays visible as missing.
Portfolio Data Health Checking recorder…
Data quality Checking observations…
No observations inspected.
Awaiting verified application state.
Research lenses · measured signals
Research lenses will populate from market observations.
These are descriptive screens of currently available tokens, not independent AI trading agents or buy recommendations. Demo tokens are synthetic.
Evaluation progress
Closed simulated positions only
Highest-ranked token research
With source logos and reported market caps · Heuristic scores — not investment advice
System activity
Recent events
Token explorer
Market cap and original coin logos, when supplied by data providers. FDV is separate from market cap.
Independent research groupings, not official Pump.fun “New Pairs / Final Stretch / Migrated” stages. Pool age and bonding state are not inferred from REST receipt time. Only positive provider-reported market caps are listed; missing caps are hidden, not deleted. Market cap ≠ FDV. Shows up to 35 per column, sorted by provider-reported fields. Updates use the existing paper-app state stream and provider polling, not Axiom's proprietary feeds.
| ASSET / LOGO | PRICE / USD | MARKET CAP | LIQUIDITY | 24H VOLUME | 24H CHANGE | RESEARCH | RISK |
|---|
Axiom-compatible Solana watchlist
Copy a token's Solana mint from Axiom Pulse and track it here using public market feeds. No Axiom account access needed.
No tracked tokens yet.
Tracking is not a direct Axiom API connection. Quotes come from supported public liquidity pools and may be unavailable for new launches.
Favorites and narratives
Open any token for a documented research summary; favorites stay on this device.
Emerging Narratives · Research Only
Observed public evidence, never token purchase recommendations.
X & news research · read-only
Detailed heuristic research signals
Paper trade history and token research ▾
Independent rug-risk evidence
Research signals
Transparent rules-based research. X and news are separate research feeds, not automatic trading signals.
X & news intelligence
Recent public posts and indexed articles. Source presence is not independent fact-checking.
Saved Public X References
Personal bookmarks, not shared AI training. Saving and opening links never calls the paid X API.
Clicking “Load official X embed” shares browser information with X. Widgets load only after your click. Protected posts may be unavailable.
Connect your X account
Official read-only OAuth 2.0 PKCE. Developer app access and an approved callback are required.
Checking authorization…
Connection diagnostics and callback URL
For the cloud deployment, configure your X OAuth 2.0 Web App, its exact HTTPS callback and read-only API budget privately on Hetzner. No X credentials belong in the browser. A separate local Windows wizard is available for local development.
The redirect-only test never exchanges tokens or connects an X account. X may still ask you to approve read-only permissions; approval can appear in your X account settings even when the Terminal discards the authorization code.
Your X password is never entered into this app. Connecting does not automatically grant search access. No trading actions are authorized.
Personalized X research feeds
Following uses authorized account context; Discover is a custom research search, not X For You.
X API credentials and endpoint access are required. Cached data is labeled as such.
No posts retrieved yet.
X account watchlist
Track public posts from selected handles, including Ansem. Prioritize relevant posts by simple engagement and Solana keywords.
Checking X integration…
No retrieved X posts yet. An official X API token is required.
X watchlist posts are public research only; they never authorize or trigger real-money trades. Scores are heuristics, not proof of importance or credibility. Accounts must be verified independently.
v0.9 Verified-mint social coverage
Counts from stored authorized X posts with an exact Solana mint mention, not ticker guesses. The feed is read-only and may be incomplete.
No recent verified-mint posts stored.
Emerging narrative phrases
KEYWORD RESEARCHRepeated phrases in retrieved headlines and X posts. Neither token matching nor a model summary verifies a rumor or predicts prices.
Available articles and posts
0 SOURCESSimulation desk
Estimated fills, explicit fee and slippage assumptions, no wallet signing.
Open positions · observed market quotes
PnL, market cap, 24h movement and price change since entry. Updates when usable observations arrive, not from invented ticks.
Awaiting portfolio state.
Detailed engine controls and positions table
Open simulated positions
Countdowns update every second. An overdue position requires a fresh, verified quote to close.
| TOKEN | ENTRY | MARKET PRICE | MARKET CAP | DATA AGE | EXIT COUNTDOWN | STOP / TAKE | VIRTUAL NOTIONAL | ACTION |
|---|
Recent engine activity
Closed simulated trades
Export CSV ↓| TIME | TOKEN | ENTRY | EXIT | NET P&L | REASON |
|---|
Trading Intelligence & System Health
See what blocks simulated entries, check provider freshness and compare independently simulated Shadow models.
Market Data Accuracy Center · v1.3.0
Source-specific timestamps, observed latency, liquidity pools, and optional manually entered Axiom reference values.
Observed price points appear after quotes arrive.
Enter a mint to inspect observed prices and source timestamps. No Axiom account connection is used.
Reference values and observation time are entered manually; they are not fetched from Axiom. Pool selection here affects research comparisons only, not the global paper-simulation price source. Circulating supply and Axiom equivalence are not assumed.
High-Fidelity Market Evidence · v1.5.6.4
Optional read-only parsed Solana swaps. Disabled by default: no billable provider is enabled by an update.
Provider-parsed confirmed swaps are provisional. Standard RPC logs are not swaps. No supply or USD conversion is inferred from unknown data. The existing Market Cap sample chart remains available.
Trading Decision Monitor
Most recent evaluated paper candidates, with every applicable rejection reason.
Most common blockers
Run the paper simulation to record candidate evaluations.
Recent coin decisions
No decisions recorded yet.
Counts may overlap because coins can fail several filters. These records document baseline decisions without changing them.
System Health & Market Data
Real-time status from stored observations, not simulated provider activity.
Optional on-chain mint verification
Historical execution replay (research only)
AI Performance & Experimental Paper Portfolios
Owner-approved validated Shadow model required; no experimental trading before validation.
Loading portfolio comparisons…
Different starting dates and estimated fills mean returns are not controlled experiments. Baseline position and risk rules are unchanged.
AI Control Center
One centralized AI · evidence-based research · simulated results and clearly scoped settings.
Central AI — current evidence
Observed paper results are not proof that a model is improving.
Risk Intelligence · v1.5.6.13
Account-scoped loss forensics, rug-risk uncertainty and experimental re-entry protection.
Observe mode blocks confirmed critical contract risks and same-tick re-entry. Cooldown mode selectively blocks repeat buys after exits; durations are unvalidated. Strict mode may prevent all new paper entries if safety and edge evidence are missing. No proven performance improvement.
Rug-Pull Investigation · v1.5.6.13
Historical recorded paper-price crashes. Suspicion is not proof of fraud; independent transaction and pool history may be unavailable.
An alert counts as pre-entry detection only when its source event AND its reception time preceded the simulated entry. Current contract checks are not historical proof.
Performance interpretation
Awaiting completed simulated trades.
Results describe past simulated outcomes; improvement has not been demonstrated without controlled validation.
Learning lifecycle
Stage progress is reported only from recorded data.
Central AI Research
CHECKINGLoading shared public-data model registry…
Market Data Reliability
SOURCE TIME ≠ RECEIPT TIMEChecking recorded quote quality…
Decision Quality Analysis
ACCOUNT-ISOLATEDComparing documented observations…
Central AI Data Eligibility
Real persisted market evidence, source timestamps and research readiness.
Checking source-event provenance and training eligibility…
Shared model validation history & data limitations
Live Market Discovery
Observed current public Solana token records · live paper engine unchanged.
Historical AI Research
Separate finalized pool OHLCV history for offline research only.
Historical Market Data & AI Research Readiness
GeckoTerminal public historical candles. A candle is not an individual on-chain trade.
Loading recorded public OHLCV coverage…
Historical backfill state & API cooldown
Historical Research Intelligence · v1.5.4
SHARED PUBLIC RESEARCHIndependent historical pool discovery, verified data quality and research coverage. No trading signals or profit predictions.
Social Narrative Intelligence · v1.5.5.2
Google Trends, optional GDELT and Bluesky · public samples, no investment signals.
Attention counts from different platforms are not directly comparable. No AI training or performance improvement is inferred.
Social Data Infrastructure · v1.5.5.1
Official X API permissions, measured public observations and strict administrator request caps. No social trading signals.
X API usage, coverage and provider diagnostics
Candidate pools, coverage & data-quality diagnostics
Research-only token risk indicators & exclusions
Historical pool selection decisions
Recent research activity
ACTUAL EVENT RECORDSLearning and shadow experiments
Historical evidence and controlled experiments; no automatic strategy promotion.
Learning Laboratory and model reflections
Historical research and evidence
Advanced market accuracy, decision monitor, system health and experimental portfolios
Nightly research · paper-only
Checking stored research report…
Scheduler activation is managed by the server administrator through systemd. This browser cannot verify or change the timer state.
Stage 2 qualification — paper-only
Research experiments and shadow results do not qualify. Existing controls remain available in the 1,000 Trade Recap.
Personal AI display preferences
These preferences only control the presentation of research information for your account. They do not alter global learning, trading rules, or the nightly timer.
Central AI and administrator controls
Shared learning and the overnight scheduler are operated on the Hetzner server. Global controls are intentionally not exposed to ordinary accounts; no fake or insecure enable buttons.
For paper-trading risk settings, use the existing general Settings page. For the research timer, the administrator uses systemd on Hetzner.
AI Performance Center
Account-isolated, closed-paper-trade forensics and offline cohort research. No autonomous live strategy changes.
Overnight research status
OPTIONAL · ADMIN OPT-INNo research report loaded. The scheduled job is disabled by default.
Historical loss evidence
Research safeguards
All results are past paper-trading observations. Historical comparisons cannot establish an executable or profitable strategy. Backtests never modify virtual account balances or Stage 2 progress.
Open-position equity is unavailable to the offline worker; use Overview for live valuation.
Time-separated cohort comparison
RESEARCH ONLY · NOT AN EXECUTABLE BACKTESTAutomated morning research note
1,000 Trade Recap
Completing 1,000 trades does not activate real-money trading.
Milestone progress
Eligible completed simulated trades for the selected mode and strategy.
The full recap unlocks at 1,000 completed trades. No real orders.
Realized virtual equity curve
CLOSED TRADES ONLYStage 2 — Optional partial-exit experiment
Requires 1,000 completed Stage 1 paper positions AND verified +50% net return AND +50% realized net profit. All data, risk and validation checks must pass. Paper only.
Independent Stage 2 position history
Independent paper-only outcomes. A partial exit is not a completed position; prices and virtual fills are not guaranteed executions.
AI Learning Laboratory
Experimental statistical research on recorded paper trades. Baseline selection and execution rules stay unchanged.
AI Learning Center · Trade Reflections
Evidence-based summaries after closed paper trades. Notes remain hypotheses until independently validated.
Historical observations & evidence
No entry-time evidence available yet.
Recent completed paper trades
No completed paper trades recorded.
Reflections do not automatically promote models or change paper-trade orders. The next coin can be compared with similar historical conditions.
Learning readiness
Collecting entry-time observations
No historical observations reconstructed.
Candidate training
Uses a fixed logistic-regression baseline with regularization, chronological holdout and purged overlapping trade outcomes. Experimental results are never used to place paper trades.
Research safeguards
Performance on historical simulated positions is not evidence of future returns.
Historical feature groups
Descriptive associations on snapshot-linked paper trades only; small groups are omitted.
No sufficiently complete entry-time samples yet.
Model evaluation history
Saved reports can be archived without changing any trade records.
No candidate models trained.
Learning Laboratory — Shadow & rejected tokens
New opportunities are recorded independently of Stage 1. Shadow predictions are not simulated trades.
Autonomous Learning — Laboratory 3.0
Background eligibility checks, chronological candidate evaluation, and validated shadow model selection. No changes to paper trades.
Loading eligibility and data quality status…
Learning 4.0 — Net return research
Independent chronological holdout research. This target never controls simulated entries or exits.
Research training history
No autonomous research runs yet.
Model selection & audit
Audit events
Automatic research can begin before 1,000 baseline trades when enough complete observations exist. Stage 2 and original scoring remain locked to their previous rules. Model comparisons are exploratory; simulated or live profitability is not guaranteed.
Observed accepted vs rejected tokens
Observed 1h price movement only. Not a hypothetical portfolio result.
Waiting for later verified observations.
Research decision trail
Only original recorded decisions, quote timestamps and supported social coverage are shown.
No evaluation records yet.
Market regimes
Awaiting observed market data.
Model drift
Stage 2 research
Counterfactual partial-exit outcomes require complete timestamped price history. Recorded Stage 2 simulation remains unchanged.
v0.9 Market observation coverage
Automatically recorded snapshots, including low-cap and declining tokens. Counts are not a complete Solana census.
No verified observations recorded yet.
Provider request time is recorded; actual market-trade time may be unknown. Synthetic demo results are labeled separately.
v0.9 Historical strategy laboratory
Seven read-only exit-policy comparisons on the same recorded paper entries. Missing price history is counted as unavailable, never invented.
No counterfactual comparisons calculated.
This is a retrospective, in-sample calculation using provider snapshots and estimated costs, not an out-of-sample test or an executable trade quote. No strategy is automatically selected.
Recorded trade exit analysis
Descriptive reports only: the original exit reasons, costs and holding durations remain unchanged.
No completed baseline trades recorded in this mode yet.
Walk-forward candidate research
Time-ordered training, overlapping-outcome purge, untouched final holdout. Research only.
No eligible shadow models yet.
Market Intelligence & Multi-Model Laboratory
Learn from both baseline-selected and rejected coins. Outcomes are recorded market observations, not tradable returns.
Observed outcomes by horizon
No observational labels yet.
Chronological model comparisons
Models appear after enough timestamped observed outcomes exist. No automatic paper-trade changes.
Forward shadow evaluation
No matured predictions recorded yet.
Explain a coin observation
Enter an exact mint address to inspect recorded, account-scoped observations and research predictions.
No coin selected.
Missing outcomes remain unavailable; external market timestamps are only verified when supplied by providers.
Phantom portfolio viewer
Inspect public Solana holdings. This application cannot sign or send transactions.
Paper balances never correspond to your actual Phantom holdings.
Track a public wallet address
Paste your Phantom SOL public address. Track any public address, or authorize your Phantom browser extension to share only its public address. Extension connection is browser-only; no signatures or transfers.
Never enter your recovery phrase or private keys. No permissions to trade are requested. Token valuations are estimates from cached market data; unpriced assets are excluded.
Settings & personalization
Choose your workspace independently of the simulation strategy. Visual changes apply immediately.
Choose your workspace
Three distinct dashboard layouts. Your choice is remembered on this device and can be changed anytime.
Portfolio chart preferences are saved privately to your signed-in account; existing layout and motion settings remain browser-local.
Appearance is stored in this browser on this computer; it doesn't change paper trades, API data or AI model behavior.
Free Social Research
Public-source collection runs once on the server. Only the server administrator can enable providers or change request limits; browser accounts cannot increase usage.
Google Trends: regional public RSS when enabled. GDELT and Bluesky: opt-in server collectors. X embeds: click-to-load only. Paid X API remains disabled separately.
Simulation configuration
Data and reliability
External integrations are unavailable without connectivity, valid provider access and their usage terms. This is a research MVP, not production trading infrastructure.