Everything GTS Trading does

360+ APIs for agents. One key-authed API for everything — expressions, handlers, strategies, Lua / C++ code, history runs, deployment, real-time monitoring, alerts and channels, account and reports. The agent fetches the live API reference itself. CLI or HTTP — no UI required. Send the key as X-API-Key: gts_... or Authorization: Bearer gts_... from a terminal, IDE script, CI job or hosted agent.

• Built-in GTS AI — writes, validates and loads the expression, and helps write your Lua / C++ code. Daily AI-token allowance by plan (up to 1,000,000 / day on Private).
• Bring your own agent — it can fetch SDK metadata, generate Lua / C++ code, submit it for review, test it, attach dependencies and deploy approved strategies.
• Self-describing SDK — the same getters for users and models: getInd(), getCdl(), getTrend(), getChannel(), getABC(), getOpenPos(), getAccount(), getFund(). Agents query names, fields and members first with members() and hasMember().
• Auto-provisioned dependencies — calls such as getInd("RSI","30Min") or getFund("revenue","ttm") resolve into the platform expressions they need and are attached to the script automatically.
• Code in, reviewed, live — agents ship whole strategies in Lua or C++ (full custom handlers, not just single rules). Every script is validated and reviewed before it can run live, then joins alerts, handlers, backtests and deployment like any built-in block.
• Mass generate & backtest — permutation runs sized by plan, with reports the agent reads back.
• Deploy, pause & monitor — deploy / undeploy strategies, pause expressions, read open and closed positions, account snapshots, signals and alert history, stats and reports.
• Build via JSON or plain English — your model emits an expression object from a copy-ready, self-contained schema prompt and the platform validates and creates it; or it describes the signal and a fetchable grammar reference keeps it accurate.

Agent workflow: point the agent at the API Docs and the schema endpoint GET /api/v1/expression-schema → validate with POST /api/v1/expressions/validate → save with POST /api/v1/expressions and /api/v1/strategies → deploy with POST /api/v1/deploy → read results back from the monitor and signals endpoints.

Example agents

• Codex — in your repo, read the GTS API docs, generate a strategy, submit it through the API, then report what passed or failed.
• Claude — turn a plain-English trading idea into JSON or Lua SDK code you paste into GTS for review.
• Gemini — brainstorm 20 strategy variants, then pick 5 for GTS history backtests.

Copy-ready things to ask — simple to complex

• Same engine as live — history runs the exact strategy logic, not a re-implementation. Many strategies per request, including permutation packs.
• Deep analytics — channel & trend analysis (e.g. every channel’s ending legs and their distribution, or all channels and their durations); the extreme value each expression reaches, event by event; object-state capture (channel-done, trend-done, bars-to-event, extreme values) logged per event; derived probability trees and channel / pattern behavior.
• Permutation builder — expression-handler wiring, symbol / range sets and parameter sweeps (RSI thresholds, SL/TP, units, your own variables). Type-driven: only valid fields per expression type can be swept. Preview count = params × symbols × expressions. Retained permutation sessions by plan.
• Reports — per run: trades, signals, P&L, win rate, profit factor, drawdown and top permutations.
• By API — agents submit history runs and read the reports back.

• AND · OR · SEQ in handlers — a handler watches a list of expressions and the relation between them, then emits its action: RSI<30 AND spread OK → alert only.
• Compound conditions — Quick Create builds the legs plus the combiner; 3+ terms and mixed AND / OR with parentheses: (a OR b) AND c. Intermediate legs are created on Save.
• Crossings — build their two legs automatically: EMA(50) crosses above EMA(200), RSI(14) crossed 30.
• Repeat-within-window — EMA(200) > EMA(100) · repeat 3 · 1H.
• Unary ops & composition — MA / EMA / abs and other transforms on any value, even another indicator: EMA(50) of RSI(14) < EMA(20) of RSI(14).
• N-bars-back & sliding windows — Close[1] > Close[2], EMA(9) of RSI(14): last 3 values a > b > c.
• Account conditions — symbol PnL < 50 AND positions < 30, combined like any other expression.
• Consensus & state machines — multi-signal consensus objects and multi-step state machines are expression types too.
• Deep nesting — use JSON Import for deeply nested logic.

• Expression — one binary node, Left OP Right, e.g. RSI(14) < 30. One uniform node for any source: indicator, pattern, price move, account state, AI output, custom code, strategy state or handler result. 30+ element types (indicators, candles, price moves, OHLC, quotes, fundamentals, SEC/EDGAR, account state). Usable as a value (a getter) or a true/false signal. Unary ops (MA/EMA/abs), N-bars-back, repeat-within-window. Parametrized, named, reusable and composable.
• Handler — a list of expressions combined by AND · OR · SEQ, bound to an action. Five handler modes: Open Bid, Open Ask, Close Bid, Close Ask, Alert only / no trade.
• Strategy — symbol + position size + SL/TP, wiring open & close handlers. Position controls: units, max positions, SL/TP in ticks or %, TP tick reversal, hedging (allow opposite), close opposite on open.
• Account-aware logic — Account Snapshot formulas (balance, equity, open-position count, per-symbol P&L, daily closed P&L & drawdown) as operands; e.g. open positions < 3, symbol PnL < 50 AND positions < 30, open another when profit rises 2%, or average balance over the last 20 4-hour snapshots > X.
• Custom code — Lua as a first-class operand inside an expression; C++ custom handlers on the same code-expression API. Code reads saved expressions and SDK getters as method calls; drives exits & risk management like entries; sizing, SL/TP & hedging are built in.
• JSON import — paste an expression object, or copy the AI schema link into ChatGPT / Claude / Gemini and paste the result back. Quick Create spends no AI tokens.

• Fundamentals as operands — fundamentals and SEC/EDGAR data are expression elements: marketCap, trailingPE, revenue TTM, or from code with getFund("revenue","ttm").
• Example filter — marketCap > 10B AND trailingPE < 20 AND revenue TTM rising AND 1Hour trend bullish.
• Explorer reports — income statement, balance sheet, cash flow, statement of equity, analysis reports, 10-K, 10-Q, insider holdings; for a year, quarter or the last X years. Export CSV or use it via API.

The API Docs are the v1 reference: quick start, agent context links, authentication, expressions, SDK getters, strategies, history, signals, monitor and settings APIs. GTS University also covers SDK-powered algo systems and the full agent API.