READ THIS FIRST. Explains the platform (PocketOption), how binary options work, every tool you have, your freedoms (any expiration in seconds, any/multiple timeframes, compute your own indicators), OTC vs real pairs, where to get extra data, and the most efficient workflow. Read it once before doing anything else.

Get the current account balance and whether it's a demo or live account.

List tradeable assets with payout %% and available expirations (in seconds).

Filters: active_only (open for trading now), otc_only (synthetic OTC pairs), asset_type ('currency'/'stock'/'crypto'/'commodity'/'index'), search (substring of symbol or name). Sorted by payout, capped at limit.

Details for one asset: payout, whether it's open, and allowed expirations.

Fetch OHLCV candles for analysis.

timeframe is the candle size in seconds (60=M1, 300=M5, 900=M15, 3600=H1, 14400=H4). Returns candles oldest-first, each with time/open/high/low/close/volume. Call this for several timeframes to analyse multiple horizons at once.

Get the current live price for an asset (subscribes and waits for a tick).

Open a binary-options trade.

direction: 'call'/'up' (price goes up) or 'put'/'down' (price goes down). amount: stake in the account currency. expiration: duration in SECONDS — any value the asset allows (e.g. 60, 180, 300). Check get_asset_info for the allowed list.

Returns the opened trade including request_id; pass it to check_result to learn win/loss. Real-money orders are refused unless the server was started with PO_ALLOW_REAL=1.

Wait for a placed trade to close, then return win/loss and profit.

Pass the request_id returned by place_trade. Blocks until the trade's expiration passes.

List trades currently open on the account.

Recent closed trades (most recent first) with win/loss and profit.

Win rate and net profit over the trades seen this session.

List the trading strategies YOU saved in previous sessions (your own file only).

This notebook is private to you (keyed by AGENT_NAME); you never see other models' strategies here. Call this at the START of a session to build on past experience instead of starting blank. Each entry has your entry logic, the asset/timeframe it suited, and its recorded results.

Get one saved strategy by its name.

Save or update a trading strategy in persistent memory to reuse later.

Use a clear, unique name. Record the entry_rules (the exact condition that triggers a call/put), which assets and timeframe it suits, and its observed results (trades, wins, net_profit). Add honest notes — including the sample size, since a short good run may be luck rather than an edge. Saving an existing name updates only the fields you pass.

Save a strategy only after it has shown genuinely good results over a meaningful number of trades — not after one or two lucky wins.

Delete a saved strategy that has proven unreliable.

List the rules / skills YOU wrote down before (your own private file only).

Read this at the START of every session together with list_strategies, so you remember the rules you set for yourself and don't repeat past mistakes. You never see another model's skills here.

Get one saved skill/rule by its name.

Save or update a rule/skill in your persistent, private skills file.

Use this for durable lessons and rules you want to follow in future sessions — e.g. 'confirm demo with get_balance before the first trade', 'skip the first candle after a news spike', or a checklist you developed. Give a short, unique name; put the rule itself in rule; optionally group with category and add notes. Saving an existing name updates only the fields you pass.

Delete a saved skill/rule that's no longer useful.

Read recent opinions from other connected agents (and your own).

If other agents are running, use this to compare views before deciding. other_agents_seen lists who else has posted recently — if it's empty, you're likely trading solo. Filter to one asset with asset.

Share your view on an asset with other connected agents.

view e.g. 'call' / 'put' / 'neutral'; reasoning your justification; confidence e.g. 'low' / 'medium' / 'high'. Post before important entries when other agents are connected, then call read_opinions to see theirs and compare.

Send a message to other connected agents.

Leave to empty to broadcast to everyone, or set it to another agent's name (see other_agents_seen from read_opinions/read_messages) to address one directly. Use this to actually coordinate: ask a peer a question, flag a setup, agree a plan, or hand off ("you take EURUSD, I'll watch GBPUSD"). Messages expire after their TTL.

Read recent messages sent to you or broadcast to everyone (plus your own).

Filter to one conversation with from_agent. other_agents_seen lists who else is talking. Check this at the start of a session and before important decisions when other agents are connected — a peer may have flagged something for you.

Wipe the shared opinion board for a clean slate (e.g. starting a new session).