analyze_setup
ChatGPTAll trades carrying a given tag/setup name (any tag category) plus aggregate stats for that setup. stats uses exactly the definitions of get_stats: counted trades only (closed, with a P&L), every money figure net of fees per trade, break-even trades NOT in the win-rate denominator, one profit factor (net wins ÷ |net losses|, null without losses), and not_counted = matching trades left out because they are open or have no result. Each trade carries strategies as a LIST, not a single value: a trade can belong to several of the trader's strategies. An empty list normally means it belongs to none — but ONLY if the response does not also carry strategies_incomplete; when it does, an empty list means unknown, and you must not say the trade is unassigned. The per-strategy figures from get_stats are built from the same assignment, so the two agree. FIELD MEANINGS (the names are misleading): max_up_move is the MFE — how far the trade ran IN FAVOUR of the position while open (LONG: high - entry, SHORT: entry - LOW). It does NOT mean "the price went up": on a short a big positive value means the market FELL, which was good. max_down_move is the MAE, the move AGAINST the position, normally negative. A negative max_up_move means the trade was never in profit. post_exit_mfe/post_exit_mae are the same after the exit (use for "exited too early/late", never for the result). All four are PRICE POINTS, not currency. r_multiple is the result as a multiple of the risked amount. trade_high/trade_low are real price levels — the favourable one is trade_high for a long, trade_low for a short. A null value means the trade came from a source without a live connection.
get_accounts
ChatGPTThe trader's trading accounts: id, name, type, broker, realized P&L from the journalled trades and trade count. id is the exact account id — pass it as account_id to get_stats/get_trades instead of guessing a name substring. type is the trader's own classification, one of "prop" (prop-firm account), "live" (own broker account) or "demo" (simulation/paper). It does NOT by itself prove whether a figure is real money: a prop-firm EVALUATION is traded on a simulated account, and replay or backtest trades may sit on an ordinary account under any type (they are usually recognisable only by the account NAME). If the difference matters for the answer, name the accounts you used and ask the trader rather than assuming. Balances are deliberately NOT available — they are never synced to the cloud. Account numbers and digits inside names are masked: only the last 3 characters are real (e.g. "****567"). Use them to tell accounts apart; never report them as corrupted and never reconstruct them.
get_journal
ChatGPTThe trader's per-day notes, merged by date: the free-form reflection PLUS the trade-tab day — pre-market plan, intraday journal and voice note. Everything the trader writes around a trading day. Optional ISO date range. Days may also carry key_levels, hypotheses and custom_sections ([{title, text}]) — the trader's own sections from their note template, in template order; read them like the rest of the day's notes.
get_playbook
ChatGPTThe trader's own definitions: strategies, setups and trading rules (each with the trader's own description of what it means), plus setup criteria. Use this to understand what the trader INTENDS, then compare against the actual trades (which reference these by name). A description of null means the trader wrote none — do not invent one and do not read the name as a definition.
get_prop_accounts
ChatGPTThe trader's prop-firm challenge/funded accounts: firm, phase, status, the rule set (drawdown/target/daily-loss/consistency), the linked trading account, and the transaction history (fees/payouts by kind, category and date). Use to assess how the trader stands in their prop challenges. Account SIZE (starting balance) is deliberately not available — it is never synced to the cloud. The rules are RELATIVE: …Pct fields are fractions of the account size (profitTargetPct: 0.06 = 6%), never dollar amounts. Transaction sums are omitted for the same reason. Reason in percentages; never multiply the fractions out and never estimate the account size.
get_stats
ChatGPTAggregate performance (win rate, net P&L, avg win/loss, profit factor, avg R) over an optional time range / instrument / account / strategy. SCOPE — READ BEFORE QUOTING ANY NUMBER: without an account filter this aggregates EVERY account the trader has, including demo, simulation, replay and backtest accounts, in ONE figure. A profit factor or P&L mixed across real and simulated trading looks like a genuine number and is not one. Call get_accounts first, then either filter by account_id or state explicitly which accounts the figure covers. Never present an unfiltered figure as the trader's real-money performance. DEFINITIONS (identical to the trader's dashboard): only COUNTED trades enter any figure — a trade counts when it is closed AND has a P&L; open trades and trades without a result are only reported in not_counted. gross_pnl is the sum of P&L before fees; net_pnl = gross_pnl − total_fees. EVERY other money figure is NET per trade (P&L − fees): avg_win, avg_loss (negative), best, worst, expectancy, profit_factor. A trade is a win if its net is above 0, a loss if below 0, and BREAK-EVEN if its net is exactly 0 to the cent (+$2 gross with $4 fees is a LOSS). win_rate_pct = wins ÷ (wins + losses): break-even trades are NOT in the denominator (null when there is no win and no loss). profit_factor = sum of net wins ÷ |sum of net losses| (null when there is no loss — say "no losing trades", never "infinite" or 0). expectancy = net_pnl ÷ trades (break-even counts). payoff_ratio = avg_win ÷ |avg_loss|. avg_r_multiple is the PLAIN arithmetic mean of r_multiple over only those trades that actually carry one; trades without a recorded risk are dropped from both numerator and denominator, so its sample is usually SMALLER than trades and it is not weighted by size, P&L or duration (null when no trade has an R — never describe it as 0 or as an average over all trades). trading_days = number of distinct calendar days on which a counted trade was ENTERED; green_days = days whose net sum is above 0 (a swing trade belongs to its entry day). Days are cut in the trader's own time zone, named in tag_zeitzone ("UTC" means the trader has not set one yet, so late-evening/night trades may sit on the neighbouring day). A null value means "no data", never zero. Use strategy together with get_playbook to answer "how is this strategy performing" — it aggregates over every trade assigned to that strategy, including trades assigned afterwards. If the result contains error: "stats_incomplete" instead of numbers, say so and do not substitute the overall figures.
get_trade_detail
ChatGPTDetail for one trade. Use detail to control the response size (default "summary" = core trade fields only, no bars/images). Ask for larger levels only when needed: "execution" adds fills, MFE/MAE and stop/target; "chart_data" adds OHLC bars; "screenshots" adds up to 4 chart images; "full" returns everything. Each trade carries strategies as a LIST, not a single value: a trade can belong to several of the trader's strategies. An empty list normally means it belongs to none — but ONLY if the response does not also carry strategies_incomplete; when it does, an empty list means unknown, and you must not say the trade is unassigned. The per-strategy figures from get_stats are built from the same assignment, so the two agree. FIELD MEANINGS (the names are misleading): max_up_move is the MFE — how far the trade ran IN FAVOUR of the position while open (LONG: high - entry, SHORT: entry - LOW). It does NOT mean "the price went up": on a short a big positive value means the market FELL, which was good. max_down_move is the MAE, the move AGAINST the position, normally negative. A negative max_up_move means the trade was never in profit. post_exit_mfe/post_exit_mae are the same after the exit (use for "exited too early/late", never for the result). All four are PRICE POINTS, not currency. r_multiple is the result as a multiple of the risked amount. trade_high/trade_low are real price levels — the favourable one is trade_high for a long, trade_low for a short. A null value means the trade came from a source without a live connection.
get_trades
ChatGPTList journal trades with optional filters (time range ISO, instrument root e.g. "MNQ", direction long/short). Newest first. Each trade carries strategies as a LIST, not a single value: a trade can belong to several of the trader's strategies. An empty list normally means it belongs to none — but ONLY if the response does not also carry strategies_incomplete; when it does, an empty list means unknown, and you must not say the trade is unassigned. The per-strategy figures from get_stats are built from the same assignment, so the two agree. FIELD MEANINGS (the names are misleading): max_up_move is the MFE — how far the trade ran IN FAVOUR of the position while open (LONG: high - entry, SHORT: entry - LOW). It does NOT mean "the price went up": on a short a big positive value means the market FELL, which was good. max_down_move is the MAE, the move AGAINST the position, normally negative. A negative max_up_move means the trade was never in profit. post_exit_mfe/post_exit_mae are the same after the exit (use for "exited too early/late", never for the result). All four are PRICE POINTS, not currency. r_multiple is the result as a multiple of the risked amount. trade_high/trade_low are real price levels — the favourable one is trade_high for a long, trade_low for a short. A null value means the trade came from a source without a live connection.
search_notes
ChatGPTFull-text search across trade notes and daily journal entries.