NT-NinjaTrader

@nt-ninjatrader source repo

13 skills
0 followers
1 day ago last updated

▌
Risk Coach · nt-ninjatrader bundle
Advisory behavioral-context check on recent fill history. Flags revenge-trading (new entry soon after a losing exit). Plus consecutive-loss streaks, overtrading counts, size drift against the user's own recent median, and hour-of-day anti-edges. Does not evaluate whether a trade is good, right, or profitable — only surfaces behavioral context from recent activity. Never blocks. Always explains. The user decides. Hard limits live server-side, and update_risk_settings configures them. Use when the user asks "am I on tilt", "should I take this", "is this a good trade" (skill responds with behavioral context, not a yes/no recommendation), "is this a good setup", "worth taking", or "is this revenge". Also use when the user says "talk me out of this", or asks "is this my 4th trade today", "does this size feel right", "sanity check", or "check me". For retrospective rule-mining use trade-debrief. For descriptive streak/hold-time stats use trade-journal.

0
installs
▌
Event Watch · nt-ninjatrader bundle
Covers a high-volatility economic event calendar — FOMC, CPI, NFP, EIA, and OPEC. Also covers earnings-season awareness and session-boundary context — globex open, NY cash open, settlement, and EIA release windows. Reports upcoming events and filters them by the user's current positions. Explains a move after it happens (for example, "ES +24pts in 8min matches the hot CPI print"). Computes historical reaction-size magnitudes from prior releases. Proposes reaction-window bracket alerts via alerts-composer. It is strictly reactive, with no timer-based push. Use when the user asks "what's coming up", "economic calendar", "why did it move", "NY cash open", or "session open". Also use when the user names a specific event — CPI, FOMC, NFP, EIA, OPEC, or earnings.

0
installs
▌
Chart Render · nt-ninjatrader bundle
This skill renders three static PNG charts, with Matplotlib. Candlestick charts, with VWAP, fill-marker, and price-level overlays. Horizontal volume-profile histograms, with POC, VAH, and VAL lines. Cumulative-P&L equity curves, with optional drawdown shading. Chart data comes from market_history bars, fill_history fills, market-context analytics, and trade-journal round-trip trades. Use it when the user asks to "chart", "plot", "draw", "visualize", or "render". Also use it when the user wants a picture of the bars, fills, equity, or profile. Also use it for a qualified "show", such as "show the chart", "show me the bars", or "show my equity curve". A bare "show me my positions" is NOT a chart request. Route it to position-watchdog or my_portfolio. It produces PNG output at a specified path. It is never interactive. This skill requires matplotlib (install with `pip install matplotlib`).

0
installs
▌
Trade Replay · nt-ninjatrader bundle
A deep dive on ONE closed trade. It reconstructs the trade's MFE (maximum favorable excursion) and MAE (maximum adverse excursion) across its lifetime, from bar OHLC. Then it runs alternate-exit counterfactuals against the same bars. The counterfactual scenarios are a fixed R target, an N-times-ATR trailing stop, a breakeven-after-X-R pin, and a time-based exit. Outputs include the exit timestamp, price, and reason. They also include the P&L in points, dollars, and R, and the delta vs the actual exit. Use it when the user asks "replay that trade", or asks for "MFE", "MAE", or "excursion" on a specific trade. Also use it for "what if I had held for X", "what if I'd used a trailing stop", "what if I'd moved my stop to breakeven", "how far did it go", "could have had", or "should have held". This is NOT backtesting. It runs no rule simulation across market history. Counterfactuals use only the user's actual fills. For many-trades summaries, use trade-journal. For session-level rule-mining, use trade-debrief.

0
installs
▌
Pretrade Risk · nt-ninjatrader bundle
Sizes a proposed futures trade from risk-per-trade (% of netLiq), stop distance (ATR-based or explicit), and contract specs. Then proposes an OCO bracket (stop + target at R-multiple) for the chosen entry. Pending-qty-aware — it includes existing filled position and open Buy/Sell orders. Combined exposure then matches what the server's pre-trade risk check will evaluate. The estimate_order tool exposes that check. Flags exceeds_daily_loss_budget, exceeds_max_contracts, stop_too_wide_for_risk_budget. Use when the user asks to "size a trade", "how many contracts", "what's my risk on this trade", "buy ES", or "sell NQ", or similar bare-intent phrasings without a quantity. Also use when the user wants a bracket proposal. Never executes orders — produces proposals the user must approve. The final feasibility gate is the estimate_order MCP tool.

0
installs
▌
Scale Manager · nt-ninjatrader bundle
Scales into or out of an existing futures position. Actions include add to winner (pyramid), take partial at 1R, and move stop to breakeven. Actions also include trail stop, close on reversal, and ladder an entry across a price band. Computes weighted cost basis, before/after dollar risk, R-multiples, and P&L impact. Proposes the exact modify_order / place_order / close_position payloads for the user to approve. Never executes. Use when the user asks to "scale out", "scale in", "take partial", "add to winner", "pyramid", "trim", "add more", "move stop to breakeven", "trail stop", "ladder entry", "split the order", or "close half". For INITIAL entry sizing use pretrade-risk; for live-position monitoring use position-watchdog.

0
installs
▌
Trade Debrief · nt-ninjatrader bundle
Educational review of past trading sessions. This skill combines timeline_report, performance_summary, fill_history, position_history, and market_history into a structured report. It then routes the report through 7 prompt templates. These are mistake identification, pattern extraction, trade scoring, correction rules, daily debrief, pattern clustering, and expectancy analysis. Every output cites the report fields directly and leads with the dollar impact. This skill describes the user's own past trading patterns. It labels any proposed change as a hypothesis, not advice. Use it for "debrief", "analyze trading", "what mistakes did I make", "what went wrong", "find patterns in trades", "score a trade", "review today's session", "coaching", "improvement rules", "what should I change", or "was that a good trade". For plain "how did I trade today" summaries, or for descriptive TCA, use trade-journal. For single-trade counterfactuals, use trade-replay.

0
installs
▌
Trade Journal · nt-ninjatrader bundle
Retrospective, descriptive analysis of closed trades — transaction-cost analysis (TCA), win/loss streaks, and hold-time distributions. It also covers P&L by symbol and by hour. On demand, it also computes per-fill slippage against arrival mid and interval VWAP. This skill is descriptive only. It reports "what happened today", not what to change. Use it when the user asks "how did I trade today", "how was my day", "what was my net P&L", or "session summary". Also use it when the user names TCA, slippage, post-mortem, or journal. Also use it for P&L breakdown, win rate, best/worst trade, hold times, or streaks. For prescriptive rule-mining and coaching, such as "what should I change" or "what mistakes", use trade-debrief. For a single-trade deep dive with counterfactuals, use trade-replay. For live open-position monitoring, use position-watchdog.

0
installs
▌
Contract Intel · nt-ninjatrader bundle
Symbol discovery, front-month resolution, rollover-imminence classification, and product-family awareness (full ↔ micro pairs like ES/MES) for futures contracts. Resolves bare product codes like "ES" or "NQ" to the actual tradeable front-month contract (e.g., ESU6) by open-interest rank across maturities. Classifies rollover status as not_imminent / approaching / imminent / rolled, with calendar-spread candidate flag when both legs are liquid. Use when the user asks "what symbol", "which contract to trade", "front month", "roll" or "rollover", "micro" or "full", or "calendar spread". Also use when the user gives a bare product code (ES, NQ, YM, CL, GC, etc.) without a maturity. Also handles product-family metadata via search_contracts includeFamilySiblings flag.

0
installs
▌
Market Context · nt-ninjatrader bundle
Computes market metrics for futures from market_history output. Metrics include session VWAP, POC/VAH/VAL (value area), bar-level delta, cumulative delta, order-flow imbalance, ATR(N), realized volatility, and cross-symbol relative strength. It produces narrative summaries of market structure and the volatility regime. Use when the user asks about VWAP, POC, market profile, or value area. Also use for ATR, realized volatility, delta, cumulative delta, or order-flow imbalance. Also use for the volatility regime, or for intra-day relative strength between related symbols (e.g., ES vs NQ vs RTY). position-watchdog, pretrade-risk, scale-manager, chart-render, trade-replay, and trade-debrief consume this skill's output as input.

0
installs
▌
Alerts Composer · nt-ninjatrader bundle
Translates natural-language alert intent into a validated Tradovate alert-DSL expression. Submits the expression via create_alert.expression. Covers 37 numeric functions across Account (netLiq, dollarOpenPL, dailyLossLimit, ...) and Contract (lastPrice, percentChange, settlementPrice, ...) entities. Also covers Position (netPos, posOpenPLUsd, posInitMarginUsd, ...) and Currency (currentRate) entities. Supports arithmetic (plus, minus, times, divide, parens), comparison (gt, gte, lt, lte, eq, neq), and top-level AND/OR/XOR logic. Validates the expression offline (subject syntax, function catalog, paren balance, flat-logic rule) before submission. Use when the user says "alert", "notify", "notify me", "watch for", "wake me when", "tell me when", "heads up when", "ping me if", "set a trigger", or "let me know if". Also use when the user asks about a sibling skill's proposed alert expression. position-watchdog, event-watch, and pretrade-risk also submit their proposed expressions through this skill.

0
installs
▌
Correlation Hedge · nt-ninjatrader bundle
Compute historical Pearson correlation and OLS regression beta between two futures symbols, on log-returns. Turn that result into a dollar-neutral hedge proposal for an open position. Covers cross-underlying relationships — ES vs NQ, CL vs NG, GC vs SI, bonds vs stocks, and currencies vs DX. For same-underlying siblings like ES/MES, use contract-intel instead. Produces sized hedge candidates with raw qty, rounded qty, residual exposure percentage, and coverage percentage. Always informational — never emits order payloads. Use when the user asks about "correlation between X and Y", "hedge this", "how do I hedge", "what could I short to offset", "beta", "X vs Y correlation", "cross-asset", "co-move", "correlated exposure", or "am I already long the market".

0
installs
▌
Position Watchdog · nt-ninjatrader bundle
Reports the live health of open futures positions. Output covers open P&L in dollars and R, and distance to stop (ticks, dollars, % of the daily loss budget). It also covers R-achieved to target, time in trade, and margin utilization. It also lists guardrail flags (no stop attached, a stop moved against entry, correlated index exposure). It also gives informational what-if math — move stop, extend target, go breakeven. It also detects stop drift from order history. A trailed stop means less risk. A widened stop means more risk. Use when the user asks "how am I doing", "what's my P&L", "how much am I up", "am I up or down", "position health", "my positions", "where's my stop", or "how close to stop". Also use when the user asks "am I in trouble", "daily loss", "margin usage", "am I safe", "what if I move my stop to X", "what if I extend target", "should I go breakeven", or "has my stop been moved". For closed-position analysis use trade-replay. For exact order payloads to execute, use scale-manager.

0
installs

13 published skills