Tool reference

Search the Arc Research knowledge graph for concepts and their relationships. Use for questions about market drivers, company exposure, and published research themes.

Parameters

• query string required Plain-language search query
• hops integer optional Neighborhood expansion depth (1-2)

Look up one concept in the Arc Research knowledge graph by slug or name and return its neighborhood of relationships and supporting sources.

Parameters

• concept string required Concept slug or name
• hops integer optional Neighborhood expansion depth (1-2)

Trace how two concepts in the Arc Research knowledge graph are connected, including relationship rationales from source research.

Parameters

• from string required Starting concept slug or name
• to string required Destination concept slug or name

Resolve a commodity by slug, name, or futures product code and return hub and stream paths on Arc Research.

Parameters

• query string required Commodity slug, name, or product code

Resolve a ticker or futures root symbol to its Arc Research page and any related commodity hub.

Parameters

• symbol string required Ticker or futures root symbol

Latest outright futures quote. A product root or slug (NG, natural-gas) returns the front month. A contract symbol (NGZ5) returns that outright. is_stale is true when the contract is expired or the bar is older than three trading days.

Parameters

• symbol string required Futures contract symbol

Latest equity quote (OHLC) for a ticker such as EQT. A futures root may collide with an equity (NG is NovaGold, not Henry Hub); the payload includes symbol_warning and company_name. is_stale is set when the bar is older than three trading days.

Parameters

• symbol string required Equity ticker symbol

Recent daily OHLC bars for an equity ticker. Server-computed — do not invent bars.

Parameters

• symbol string required Equity ticker symbol
• limit integer optional Number of bars (default 30, max 120)

Server-computed SMA, EMA, or RSI for an equity ticker. kind is required: sma, ema, or rsi. Never invent indicator values.

Parameters

• symbol string required Equity ticker symbol
• kind string required sma, ema, or rsi
• period integer optional Lookback period

List commodities the authenticated user has pinned on Arc Research.

No parameters.

List commodity-adjacent equities (producers, miners, fertilizer, utilities) for a commodity, with latest screen scores when available.

Parameters

• commodity string required Commodity slug, name, or product code
• limit integer optional Max members (default 25, max 100)
• role string optional Optional role: producer, midstream, utility, sic_match
• include_non_common boolean optional Include preferreds, notes, and units. Default false.

Screen commodity-adjacent equities. preset: value | breakout | flow | hybrid. Returns a short ranked table with scores and one-line reasons.

Parameters

• commodity string required Commodity slug, name, or product code
• preset string optional value | breakout | flow | hybrid
• min_dollar_volume number optional Minimum dollar volume
• max_results integer optional Max hits (default 15, max 50)
• include_non_common boolean optional Include preferreds, notes, and units. Default false.

Equity institutional-flow proxy from daily price/volume (not 13F/dark pool). Optionally attaches linked commodity COT positioning.

Parameters

• symbol string required Equity ticker symbol
• lookback_days integer optional Lookback days (default 60)
• commodity string optional Optional commodity slug for COT context

Single-name equity dossier: quote, screen features, flow proxy, fundamentals, graph neighborhood, and explicit data gaps.

Parameters

• symbol string required Equity ticker symbol
• commodity string optional Optional commodity slug for feature context

Latest Finnhub basic financials snapshot (P/E, P/B, margins, debt) plus optional peer-relative comparison within a commodity universe.

Parameters

• symbol string required Equity ticker symbol
• commodity string optional Optional commodity for peer medians

Get the latest CFTC Commitments of Traders (COT) positioning snapshot for a commodity: open interest, trader-category levels, week-over-week changes, and a managed-money sentiment score. Use for marketing/spec positioning questions.

Parameters

• commodity string required Commodity slug, name, or product code (e.g. natural-gas, NG)

Return recent week-over-week CFTC/COT position changes for a commodity (managed money, producers, swap dealers, open interest). Useful for tracking marketing and speculative flow.

Parameters

• commodity string required Commodity slug, name, or product code
• weeks integer optional Number of weeks (1-26, default 8)

Compute the managed-money COT oscillator (0-100) for a commodity over a lookback window. High values mean managed money is near lookback net-long extremes; low values mean near net-short extremes.

Parameters

• commodity string required Commodity slug, name, or product code
• lookback integer optional Lookback weeks (4-104, default 52)

Return latest plus prior weeks of CFTC/COT rows for a commodity (levels and changes). Use when you need a longer positioning history beyond a single week.

Parameters

• commodity string required Commodity slug, name, or product code
• weeks integer optional Prior weeks to include (1-52, default 12)

Fetch the published Arc Research COT / marketing-positioning narrative for a commodity stream. Optionally pass a YYYY-MM-DD reporting date; otherwise returns the latest published brief.

Parameters

• commodity string required Commodity slug, name, or product code
• date string optional Optional reporting date YYYY-MM-DD

Get the latest EIA Weekly Natural Gas Storage Report snapshot for Lower 48 working gas: storage level (Bcf), net change, and 5-year average context. Defaults to the natural-gas commodity hub.

Parameters

• commodity string optional Commodity slug (default natural-gas). Must have EIA storage linkage.

Fetch the published Arc Research EIA Weekly Natural Gas Storage narrative. Optionally pass a YYYY-MM-DD reporting date; otherwise returns the latest brief.

Parameters

• commodity string optional Commodity slug (default natural-gas)
• date string optional Optional reporting date YYYY-MM-DD

Get the latest NOAA Climate Prediction Center population-weighted HDD/CDD degree-day snapshot (national + regions vs normal). Key input for U.S. natural gas weather demand research.

Parameters

• weeks integer optional Number of recent weeks (1-12, default 1)

Get the latest GFS/ECMWF multi-hub U.S. demand-weighted HDD/CDD forecast: weighted cumulative degree days, model convergence, and run-over-run deltas. Use for forward-looking natural gas weather/demand research.

Parameters

• include_locations boolean optional Include per-hub location detail (default false; can be large)

Return point-in-time natural gas weather ML features (Helios): demand-weighted HDD/CDD from GFS and ECMWF, model spreads, convergence, and run deltas as of a decision date. Defaults to today.

Parameters

• as_of string optional Decision date YYYY-MM-DD (default today)

Fetch the latest published Arc Research weather & demand narrative for natural gas (NOAA degree days and/or GFS/ECMWF hub forecast briefs). Optionally pass a YYYY-MM-DD reporting date.

Parameters

• commodity string optional Commodity slug (default natural-gas)
• date string optional Optional reporting date YYYY-MM-DD

List the caller's Arc Research stories (investment theses) with id, title, description snippet, sentiment, tickers, and path. Use before get_story or journal tools when the story id/title is unknown.

No parameters.

Fetch one of the caller's stories by id or title, including description, sentiment, tickers, and recent journal entries.

Parameters

• story string required Story id or title
• journal_limit integer optional Recent journal entries to include (1-50, default 10)

List journal entries for one of the caller's stories (newest first). Pass story id or title.

Parameters

• story string required Story id or title
• limit integer optional Max entries (1-100, default 20)

Semantically search the caller's stories and journal entries (StoryChunk embeddings). Use for "what did I write about X?" style questions. Falls back to title/description match when embeddings are not ready yet.

Parameters

• query string required Natural language search query
• limit integer optional Max chunks (1-20, default 5)

Add a journal entry to one of the caller's stories (by id or title). Use when the user wants to record an insight or update on an investment thesis. This mutates the caller's data and re-embeds the story for search.

Parameters

• story string required Story id or title
• content string required Journal entry body

Admin only. List all blog posts including drafts: id, slug, title, published_at, draft/published status, and tags.

No parameters.

Admin only. Fetch one blog post by id or slug, including full markdown body and metadata.

Parameters

• blog_post string required Blog post id or slug

Admin only. Create a blog post (markdown body). Defaults to a draft unless published_at is set. Slug is parameterized from the title when omitted.

Parameters

• title string required Post title
• body string required Markdown body
• slug string optional URL slug (lowercase letters, numbers, hyphens)
• tags string optional Comma-separated tag names
• excerpt string optional OG description / excerpt
• og_title string optional Open Graph title (defaults to title)
• published_at string optional Optional ISO8601 publish time; omit to leave as draft

Admin only. Edit title, markdown body, slug, tags, or excerpt. Does not publish or unpublish; use publish_blog_post / unpublish_blog_post for that.

Parameters

• blog_post string required Blog post id or slug
• title string optional New title
• body string optional New markdown body
• slug string optional New URL slug
• tags string optional Replacement comma-separated tag names
• excerpt string optional OG description / excerpt
• og_title string optional Open Graph title

Admin only. Set published_at on a blog post (defaults to now). Use unpublish_blog_post to clear it.

Parameters

• blog_post string required Blog post id or slug
• published_at string optional Optional ISO8601 publish time (default now)

Admin only. Clear published_at so the post becomes a draft again.

Parameters

• blog_post string required Blog post id or slug

Alpha — requires Commodity ML alpha access. List forecast models the caller can access, including visibility and whether each is platform core.

No parameters.

Alpha — requires Commodity ML alpha access. Fetch metrics, hyperparameters, and artifact metadata for one model version.

Parameters

• model_version_id integer required CommodityMlModelVersion id

Alpha — requires Commodity ML alpha access. List recent training exports (CSV snapshots used to train forecast models). Use the id as training_export_id for get_commodity_ml_training_export, analyze_commodity_ml_training_export, or create_commodity_ml_experiment.

Parameters

• commodity string optional Commodity slug or name (optional)
• feature_profile string optional Filter profile, e.g. natural_gas_storage

Alpha — requires Commodity ML alpha access. Inspect one training export: metadata, stored data_quality audit, CSV columns, date range, present targets, and head/tail sample rows. Use analyze_commodity_ml_training_export for stats and improvement ideas.

Parameters

• training_export_id integer required CommodityMlTrainingExport id
• sample_rows integer optional Head/tail sample size (default 5, max 15)

Alpha — requires Commodity ML alpha access. Analyze a training-export CSV: column stats, target summary, feature–target correlations, collinear pairs, and ranked suggestions (missing targets, stale schema, nulls, constant features, scale). Use the suggestion actions to pick the next experiment grid or re-export.

Parameters

• training_export_id integer required CommodityMlTrainingExport id
• sample_rows integer optional Head/tail sample size (default 5, max 15)

Alpha — requires Commodity ML alpha access. List parameter experiments for a model.

Parameters

• model_id integer required CommodityMlModel id

Alpha — requires Commodity ML alpha access. Get trial ranking and walk-forward metrics for one experiment.

Parameters

• experiment_id integer required CommodityMlExperiment id

Alpha — requires Commodity ML alpha access. Start a parameter sweep experiment on a training export. Default grid: reg_param [0.1,1.0,10.0] x standardize_features [false,true].

Parameters

• model_id integer required CommodityMlModel id
• training_export_id integer required CommodityMlTrainingExport id
• parameter_grid_json string optional Optional JSON object grid
• eval_window_days integer optional Walk-forward window days (default 180)

Alpha — requires Commodity ML alpha access. Publish your own model (subscribers or published visibility). Does not promote to platform core.

Parameters

• model_id integer required CommodityMlModel id
• visibility string optional subscribers or published (default published)

Alpha admin only. Promote a completed experiment trial to the platform core published model used by scheduled forecast pages. Non-admins cannot call this successfully.

Parameters

• trial_id integer required CommodityMlParameterTrial id

Alpha — requires Commodity ML alpha access. Queue a walk-forward backtest for a completed model version.

Parameters

• model_version_id integer required CommodityMlModelVersion id

Alpha — requires Commodity ML alpha access. Queue a Lower-48 storage training export and (by default) train the change + direction models. Same job as the Wednesday cron. After it completes, list_commodity_ml_models / get_commodity_ml_model_version for the new storage_regression version, then run_commodity_ml_backtest.

Parameters

• commodity string optional Commodity slug or name (default natural-gas)
• start_ruby_training boolean optional Train after export (default true)

Alpha — requires Commodity ML alpha access. Latest published platform forecasts and the model version that produced them.

Parameters

• commodity string optional Commodity slug or name (default natural-gas)