More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    AI Trading Risk & Position Sizing Guardian

    1

    AI Trading Risk & Position Sizing Guardian is a professional pre-trade and intraday risk-control agent designed for systematic, discretionary, and agentic trading workflows.

    $9.99

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    AI Trading Risk & Position Sizing Guardian

    AI Trading Risk & Position Sizing Guardian

    Example session with this skill installed

    CANDIDATE TRADE **Trade ID** MNQ-LONG-2026-08-22-001 **Instrument** MNQU26 **Instrument Type** Micro E-mini Nasdaq-100 Futures **Direction** Long **Entry** 24,850.00 **Stop** 24,825.00 **Target** 24,912.50 **Requested Quantity** 5 contracts **Tick Size** 0.25 index points **Tick Value** $0.50 per tick per contract **Point Value** $2.00 per index point per contract **Minimum Quantity** 1 contract **Quantity Step** 1 contract **Account Currency** USD ACCOUNT STATE **Balance** $50,420 **Equity** $49,800 **Peak Equity** $52,000 Realized P&L Today: -$320 Unrealized P&L Today: -$180 **Used Margin** $4,200 **Free Margin** $45,600 **Current Consecutive Losses** 3 **Trades Today** 6 **Account State Timestamp** 2026-08-22 10:31:00 America/New_York EXISTING POSITIONS Position 1: **Instrument** MESU26 **Direction** Long **Quantity** 3 contracts **Entry** 6,420.00 **Current Price** 6,412.00 **Stop** 6,405.00 **Risk to Stop** $225 **Notional Exposure** $96,180 **Correlation Bucket** US Equity Index Futures **Strategy** TrendPulse Position 2: **Instrument** MGCZ26 **Direction** Short **Quantity** 2 contracts **Entry** 3,485.00 **Current Price** 3,479.00 **Stop** 3,495.00 **Risk to Stop** $200 **Notional Exposure** $69,580 **Correlation Bucket** Precious Metals **Strategy** GoldMeanRev EXISTING PORTFOLIO TOTALS **Existing Open Risk** $425 **Existing Gross Exposure** $165,760 **Existing Open Positions** 2 Existing US Equity Index Futures Correlation-Bucket Risk: $225 RISK POLICY **Policy Name** Agentic Futures Conservative Risk Policy **Policy Version** 2.3 **Risk Basis** Current Equity **Maximum Risk per Trade** 0.50% of current equity **Maximum Daily Loss** $800 **Daily Loss Basis** Realized + Unrealized P&L **Maximum Total Open Risk** 2.00% of current equity **Maximum Gross Leverage** 4.0x equity **Maximum Open Positions** 4 **Maximum Correlated Positions** 2 positions per correlation bucket Maximum US Equity Index Futures Correlation-Bucket Risk: $500 **Maximum Consecutive Losses** 5 **Loss-Streak Rule** After 3 consecutive losing trades, reduce maximum risk per trade to 50% of the normal risk limit. At 5 consecutive losing trades, STOP TRADING. **Maximum Drawdown** 5.0% from peak equity **Drawdown Rule** At 4.0% drawdown, maximum risk per trade becomes 50% of normal. At 5.0%, STOP TRADING. **Minimum Reward/Risk** 1.50 **Maximum Position Size** 10 contracts per instrument **Maximum Daily Trades** 10 **Margin Buffer Requirement** At least 20% of equity must remain available as free-margin buffer after the trade. **Correlation Rule** Trades within the same correlation bucket must respect both the maximum correlated-position count and correlation-bucket open-risk limit. **Risk Rounding** Always round down to the nearest valid contract. **Fail-Closed Rule** BLOCK TRADE when a safety-critical trade input is missing. STOP TRADING when broker account state is contradictory or unreliable. REQUEST Do not analyze market direction. Evaluate only risk permission and maximum position size. **Calculate** Risk distance Ticks to stop Risk per contract Requested trade risk Normal risk budget Active loss-streak-adjusted risk budget Current drawdown Drawdown risk tier Daily P&L Remaining daily-loss headroom Existing open risk Post-trade open risk Gross exposure Gross leverage Open positions Correlation-bucket risk Reward/risk Maximum permitted contracts Binding constraint **Finish with exactly one final state** ALLOW REDUCE SIZE BLOCK TRADE STOP TRADING

    • Read your context and instructions
    • Compiled the ai trading risk

    === AI TRADING RISK & POSITION SIZING GUARDIAN === **Decision** REDUCE SIZE **Decision Severity** HIGH **Candidate** MNQU26 Long **Requested Size** 5 contracts **Maximum Permitted Size** 2 contracts **Primary Binding Constraint** Loss-Streak-Adjusted Per-Trade Risk Limit **Secondary Constraints** Daily Loss Headroom US Equity Index Futures Correlation-Bucket Risk **Policy** Agentic Futures Conservative Risk Policy **Policy Version** 2.3 **Account State Timestamp** 2026-08-22 10:31:00 America/New_York === 1. ACCOUNT VALIDATION === **Balance** $50,420 **Equity** $49,800 **Peak Equity** $52,000 **Status** VALID No contradictory account-state information is present in the supplied data. === 2. CANDIDATE TRADE GEOMETRY === **Direction** Long **Entry** 24,850.00 **Stop** 24,825.00 **Target** 24,912.50 **Long stop validation** 24,825.00 < 24,850.00 **Status** VALID **Price Risk** 24,850.00 - 24,825.00 = 25.00 index points === 3. INSTRUMENT ECONOMICS === **Tick Size** 0.25 **Tick Value** $0.50 **Point Value** $2.00 **Ticks to Stop** 25.00 ÷ 0.25 = 100 ticks **Risk per Contract** 100 × $0.50 = $50.00 **Cross-Check** 25 points × $2.00 = $50.00 **Status** VERIFIED INTERNALLY FROM SUPPLIED SPECIFICATION === 4. REQUESTED TRADE RISK === **Requested Quantity** 5 contracts **Risk per Contract** $50 **Requested Planned Risk** 5 × $50 = $250 === 5. NORMAL PER-TRADE RISK BUDGET === **Maximum Normal Risk** 0.50% of $49,800 = $249.00 === 6. LOSS-STREAK ADJUSTMENT === **Current Consecutive Losses** 3 **Policy** At 3 consecutive losses: Maximum risk per trade becomes 50% of normal. **Normal Risk Budget** $249.00 **Adjusted Risk Budget** $249.00 × 50% = $124.50 **Risk-Based Raw Maximum Contracts** $124.50 ÷ $50.00 = 2.49 contracts **Whole-contract conservative rounding** 2 contracts **Maximum Risk-Based Size** 2 contracts Risk at 2 Contracts: $100 Risk at 3 Contracts: $150 Three contracts would exceed the active adjusted risk budget. **Therefore** Maximum = 2 contracts === 7. DRAWDOWN === **Peak Equity** $52,000 **Current Equity** $49,800 **Drawdown Amount** $52,000 - $49,800 = $2,200 **Drawdown Percentage** $2,200 ÷ $52,000 × 100 = 4.23% **Policy** Below 4%: Normal risk 4% to below 5%: 50% risk 5% or greater: STOP TRADING **Current Drawdown** 4.23% **Status** REDUCED-RISK TIER This independently confirms that the active maximum risk per trade must be reduced to 50% of the normal limit. The loss-streak rule and drawdown rule therefore point to the same reduced-risk state. === 8. DAILY LOSS CONTROL === Realized P&L: -$320 Unrealized P&L: -$180 **Policy Basis** Realized + Unrealized Current Daily P&L: -$320 + -$180 = -$500 **Maximum Daily Loss** -$800 **Remaining Daily Loss Headroom** $800 - $500 = $300 **Requested Trade Risk** $250 Requested trade would not independently exceed daily-loss headroom. Reduced 2-Contract Risk: $100 Post-Stop Daily Loss Scenario at Planned Stop: -$500 - $100 = -$600 Status at 2 Contracts: PASS Remaining buffer to daily stop after planned stop loss: $200 **Important** This calculation assumes execution near the planned stop. Gap and slippage risk beyond the stop are not quantified in the supplied policy input. === 9. TOTAL OPEN RISK === **Current Equity** $49,800 **Maximum Total Open Risk** 2.00% × $49,800 = $996 **Existing Open Risk** $425 Candidate Risk at Requested 5 Contracts: $250 **Requested Post-Trade Open Risk** $425 + $250 = $675 **Status** PASS Candidate Risk at Reduced 2 Contracts: $100 **Reduced Post-Trade Open Risk** $425 + $100 = $525 **Open-Risk Utilization** $525 ÷ $996 = 52.71% **Status** PASS === 10. OPEN POSITIONS === **Existing** 2 **Candidate** 1 new position **Post-Trade** 3 **Maximum** 4 **Status** PASS === 11. CORRELATION REVIEW === **Candidate Correlation Bucket** US Equity Index Futures **Existing Position in Same Bucket** MESU26 Long **Existing Correlated Positions** 1 **Post-Trade Correlated Positions** 2 **Maximum** 2 **Position-Count Status** PASS AT LIMIT **Existing Bucket Risk** $225 Candidate Risk at Requested 5 Contracts: $250 **Requested Post-Trade Bucket Risk** $225 + $250 = $475 **Maximum Bucket Risk** $500 **Requested Size Status** PASS, but only $25 below the bucket limit. Candidate Risk at Reduced 2 Contracts: $100 **Reduced Post-Trade Bucket Risk** $225 + $100 = $325 **Bucket Utilization** $325 ÷ $500 = 65% **Status** PASS **Risk Observation** The existing MES long and proposed MNQ long both belong to the same US equity-index correlation bucket. The portfolio would reach the maximum permitted number of positions within this bucket. No additional new position in the same bucket should be admitted while both remain open unless the policy state changes. === 12. REWARD / RISK === **Long Reward** 24,912.50 - 24,850.00 = 62.50 points **Risk** 24,850.00 - 24,825.00 = 25.00 points **Reward/Risk** 62.50 ÷ 25.00 = 2.50 **Minimum Required** 1.50 **Status** PASS **Important** This verifies structural reward/risk compliance only. It does not estimate win probability or trade quality. === 13. GROSS EXPOSURE / LEVERAGE === **Existing Gross Exposure** $165,760 Candidate notional exposure must be calculated using the appropriate verified futures notional methodology. The supplied data includes price and point value but does not explicitly define whether the risk policy wants contract notional calculated directly as: Price × Contract Multiplier or whether broker-reported futures exposure should be used. **Therefore** **Gross Leverage** REQUIRES POLICY-CONSISTENT NOTIONAL DEFINITION This item is not silently inferred. Because the supplied policy requires a 4.0x gross leverage limit, execution authorization should use the broker or operator's defined futures exposure methodology before final order submission. **Current Decision Impact** No stricter size is derived from leverage because the required notional methodology is not explicitly supplied. **Revalidation Required Before Execution** YES === 14. MARGIN BUFFER === **Current Free Margin** $45,600 **Required Buffer** 20% × $49,800 = $9,960 Current Free Margin is substantially above the required buffer. However, candidate contract margin requirement was not supplied. **Therefore** **Post-Trade Margin Status** REQUIRES BROKER MARGIN INPUT This does not change the current calculated risk-based maximum size, but the 2-contract authorization must not be executed if actual post-trade free margin would fall below $9,960. === 15. SIZE-CAP COMPARISON === **Requested Size** 5 contracts **Normal Per-Trade Risk Cap** 4 contracts maximum after conservative rounding from $249 / $50 **Loss-Streak-Adjusted Risk Cap** 2 contracts **Drawdown-Adjusted Risk Cap** 2 contracts **Daily-Loss Headroom Cap** 6 contracts by planned-stop arithmetic **Total Open-Risk Cap** 11 contracts by open-risk arithmetic, before other constraints **Correlation Bucket Risk Cap** 5 contracts by bucket-risk arithmetic **Maximum Position Size** 10 contracts **Open-Position Count** Allows candidate position **Margin Cap** Requires broker margin verification **Leverage Cap** Requires policy-consistent futures notional verification === 16. MAXIMUM PERMITTED SIZE === Maximum Currently Permitted by Verified Quantitative Constraints: 2 contracts **Planned Risk** 2 × $50 = $100 **Risk as Percentage of Equity** $100 ÷ $49,800 × 100 = 0.201% **Active Adjusted Maximum** $124.50 ÷ $49,800 × 100 = 0.250% **Status** COMPLIANT === 17. BINDING CONSTRAINTS === **Primary** Loss-Streak Reduced-Risk Tier **Secondary** Drawdown Reduced-Risk Tier Both independently reduce maximum risk per trade from: 0.50% to: 0.25% === 18. REASON CODES === RG-004 RISK_PER_TRADE_EXCEEDED **Requested risk** $250 **Active maximum** $124.50 **Additional Control** DRAW-DOWN REDUCED-RISK TIER ACTIVE **Current drawdown** 4.23% **Additional Control** LOSS-STREAK REDUCED-RISK TIER ACTIVE **Current consecutive losses** 3 === 19. RISK UTILIZATION AT 2 CONTRACTS === **Trade-Risk Utilization** $100 ÷ $124.50 = 80.32% **Daily-Loss Utilization After Planned Stop** $600 ÷ $800 = 75.00% **Open-Risk Utilization** $525 ÷ $996 = 52.71% **Correlation-Bucket Risk Utilization** $325 ÷ $500 = 65.00% **Open-Position Utilization** 3 ÷ 4 = 75.00% **Correlated-Position Utilization** 2 ÷ 2 = 100.00% === 20. UNKNOWNS === **Gap Risk** Not quantified. **Stop Slippage** Not quantified. **Candidate Margin Requirement** Not supplied. **Post-Trade Free Margin** Cannot be verified until candidate margin requirement is known. **Gross-Leverage Methodology for Futures** Not explicitly supplied. **Current Broker Margin Specification** Not verified. These items should be resolved before execution where required by policy. === 21. EXECUTION REVALIDATION === **Required Before Execution** YES **Recheck** Account Equity Daily P&L Existing Positions Current Open Risk Loss Streak Drawdown Broker Margin Requirement Post-Trade Free Margin Gross Leverage Correlation-Bucket State **Reason** The risk decision is valid only for the supplied account-state snapshot. Another order fill or meaningful P&L change can alter the permitted quantity. === FINAL DECISION === REDUCE SIZE **Requested** 5 MNQU26 contracts **Maximum Permitted** 2 contracts **Primary Reason** The requested $250 planned risk exceeds the active $124.50 per-trade risk budget. The active budget has been reduced to 50% of normal because both: Consecutive Losses = 3 and: Drawdown = 4.23% place the account in the policy's reduced-risk tier. At 2 contracts: **Planned Risk** $100 **Reward/Risk** 2.50 **Post-Trade Open Risk** $525 **US Equity Index Futures Bucket Risk** $325 **Open Positions** 3 of 4 **Correlated Positions** 2 of 2 The trade is therefore conditionally compliant at no more than 2 contracts under the verified calculations. **Final State** REDUCE SIZE This is a risk-policy decision only. It is not a prediction that the trade will be profitable or safe.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    AI Trading Risk & Position Sizing Guardian is a professional pre-trade and intraday risk-control agent designed for systematic, discretionary, and agentic trading workflows. Its role is deliberately separated from market prediction. It does not decide whether the market is likely to rise or fall. It does not generate BUY or SELL signals. It does not increase risk because a strategy reports high confidence. Instead, it answers two operational questions: 1. Is this proposed trade permitted under the supplied risk policy? 2. If it is permitted, what is the maximum position size that can be taken without violating that policy? The agent can receive: Account Balance Account Equity Peak Equity Entry Price Stop-Loss Price Target Price Instrument Direction Requested Quantity Tick Size Tick Value Point Value Contract Multiplier Lot Size Pip Value Minimum Quantity Quantity Increment Existing Positions Existing Open Risk Gross Exposure Net Exposure Current Margin Available Margin Daily Realized P&L Daily Unrealized P&L Current Drawdown Loss Streak Correlation Data Strategy Exposure Asset-Class Exposure Risk Policy Broker or Exchange Constraints It evaluates the proposed trade against explicit operator-defined limits such as: Maximum Risk per Trade Maximum Daily Loss Maximum Total Open Risk Maximum Portfolio Exposure Maximum Gross Exposure Maximum Net Exposure Maximum Leverage Maximum Margin Utilization Minimum Free-Margin Buffer Maximum Open Positions Maximum Correlated Positions Maximum Correlation-Bucket Risk Maximum Single-Instrument Exposure Maximum Sector Exposure Maximum Asset-Class Exposure Maximum Directional Exposure Maximum Strategy Exposure Maximum Consecutive Losses Maximum Drawdown Minimum Reward/Risk Maximum Position Size Daily Profit Protection Rules Cooldown Rules Session Stop Rules Kill-Switch Rules The guardian follows a strict risk-processing sequence: Validate Inputs → Validate Risk Policy → Normalize Instrument Economics → Calculate Unit Risk → Calculate Requested Trade Risk → Calculate Maximum Risk-Based Position Size → Evaluate Daily Loss Headroom → Evaluate Existing Open Risk → Evaluate Portfolio Exposure → Evaluate Leverage → Evaluate Margin → Evaluate Correlation → Evaluate Position Count → Evaluate Drawdown → Evaluate Loss Streak → Evaluate Reward/Risk → Apply Hard Stops → Apply Trade-Level Constraints → Determine Maximum Permitted Size → Produce Final Decision → Record Audit Trail The final result is always one of four operational states: ALLOW The requested quantity complies with every applicable supplied risk rule. REDUCE SIZE The proposed trade is structurally permitted, but the requested quantity exceeds one or more risk limits. The agent calculates the largest valid quantity that remains compliant. BLOCK TRADE The account may continue trading, but this particular trade cannot be admitted. STOP TRADING An account-level or session-level risk stop has been reached and no additional risk-increasing trades are permitted. Decision precedence is: STOP TRADING → BLOCK TRADE → REDUCE SIZE → ALLOW This means a severe account-level risk event cannot be overridden by an attractive reward/risk ratio, favorable strategy signal, or small individual stop. The guardian calculates position size from actual monetary risk. For linear instruments, the core concept is: Unit Risk = Monetary loss from entry to stop for one position unit. If the risk policy defines a percentage-based trade budget: Risk Budget = Account Equity × Maximum Risk per Trade % Maximum Raw Position Size = Risk Budget ÷ Unit Risk The result is then constrained by: Daily Loss Headroom Total Open-Risk Headroom Gross Exposure Net Exposure Leverage Margin Correlation Position Count Instrument Limits Strategy Limits Quantity Increment Minimum Tradable Quantity The final maximum permitted size is the minimum of all applicable limits. The agent never rounds upward when doing so would breach policy. For futures, it can calculate risk using: Price Distance Tick Size Tick Value Point Value Contracts Example: Price Distance ÷ Tick Size × Tick Value = Risk per Contract For equities: Absolute Difference Between Entry and Stop × Shares For forex, it can use: Lot Size Pip Size Pip Value Quote Currency Account Currency Currency Conversion For crypto derivatives, it can account for: Contract Multiplier Linear vs Inverse Contract Structure Position Units Margin Mode Leverage Liquidation Information when those inputs are supplied. It does not assume that spot, futures, perpetual contracts, forex lots, CFD contracts, equity shares, and options use identical sizing mathematics. Instrument economics must be verified before monetary risk is calculated. The agent validates stop-loss geometry. For a conventional long trade: Stop < Entry For a conventional short trade: Stop > Entry An invalid or zero-distance stop can produce: BLOCK TRADE The guardian never silently creates a stop-loss when none has been supplied. It can calculate reward/risk. For a long trade: Reward = Target - Entry Risk = Entry - Stop For a short trade: Reward = Entry - Target Risk = Stop - Entry Reward/Risk = Reward ÷ Risk If the supplied risk policy requires a minimum reward/risk ratio, the trade must satisfy it. If a strategy legitimately uses trailing exits without a fixed target, the agent does not fabricate a target merely to create an R/R value. The guardian performs daily loss control. The policy may define daily loss using: Realized P&L Only or: Realized + Unrealized P&L The agent never mixes those definitions silently. It can calculate: Current Daily Loss Maximum Daily Loss Remaining Daily Loss Headroom If the daily stop has already been reached: STOP TRADING If the daily stop has not yet been reached but the full requested trade would exceed remaining headroom: REDUCE SIZE when a smaller valid position exists. Otherwise: BLOCK TRADE The guardian evaluates total open risk. A common policy definition is: Total Open Risk = Sum of planned losses from current prices or entries to active stop levels Post-Trade Open Risk = Existing Open Risk + Candidate Trade Risk It compares this result with the supplied maximum open-risk limit. The skill explicitly states that stop-defined risk is planned risk, not guaranteed realized maximum loss. Actual loss can exceed the stop-defined amount because of: Gaps Slippage Liquidity Rejected Orders Exchange Interruptions Broker Failures Market Dislocations The agent calculates gross and net portfolio exposure separately. Gross Exposure: Sum of absolute notional positions. Net Exposure: Long Notional - Short Notional It does not assume that a low net exposure means low portfolio risk. A portfolio can have: Low Net Exposure High Gross Exposure High Leverage High Correlation at the same time. The guardian can calculate gross leverage using a supplied policy definition such as: Gross Leverage = Gross Notional Exposure ÷ Account Equity Alternative leverage definitions may also be used when supplied by the operator. It never compares leverage values created under incompatible methodologies. Margin risk can be evaluated when margin information is supplied. Possible metrics include: Used Margin Free Margin Post-Trade Margin Margin Utilization Minimum Free-Margin Buffer If margin requirements are dynamic or unknown, the agent does not invent them. The guardian evaluates portfolio correlation. It can work with: Operator-Defined Correlation Buckets Correlation Matrices Rolling Correlations Sector Mappings Asset-Class Mappings Factor Groups Examples of practical exposure buckets include: US Equity Index Futures Technology Equities USD Directional FX Precious Metals Energy Rates Crypto Beta For example: Long ES Long NQ Long MNQ may represent materially concentrated US equity-index exposure even though three different symbols are used. The guardian can compare: Existing Correlated Risk Candidate Correlated Risk Post-Trade Correlated Risk Correlation-Bucket Limit If a smaller quantity fits: REDUCE SIZE If even the minimum tradable unit breaches the limit: BLOCK TRADE The guardian evaluates maximum open positions. It distinguishes between: New Position Scale-In Partial Exit Full Exit Hedge Reversal Stop Modification Target Modification A risk-reducing exit should not automatically be treated like a prohibited new trade. The guardian can evaluate scale-in orders. For a proposed addition, it recomputes: Incremental Risk Total Position Risk Average Entry Stop Risk Portfolio Open Risk Gross Exposure Leverage Margin Correlation It can also evaluate stop modifications. Moving a stop farther from price may materially increase account risk without changing position quantity. The guardian can calculate: Risk Before Stop Change Risk After Stop Change Incremental Risk and block the modification when it breaches policy. Removing a stop can also be treated as a risk-increasing action when the operator requires all positions to have predefined risk. Drawdown control is supported. Possible definitions include: Peak-to-Current Equity Drawdown Balance Drawdown Session Drawdown Static Drawdown Trailing Drawdown Prop-Firm Trailing Threshold The guardian uses the definition supplied by the policy. For peak equity: Drawdown Amount = Peak Equity - Current Equity Drawdown % = (Peak Equity - Current Equity) ÷ Peak Equity × 100 If the maximum drawdown stop is reached: STOP TRADING Risk-tier policies can also be enforced. Example: Drawdown below 2%: Normal risk. Drawdown 2%–3%: Reduced risk. Drawdown 3% or greater: Stop trading. The agent does not invent those thresholds. The guardian evaluates consecutive-loss rules. A supplied policy can specify: After 3 losses: Reduce size. After 5 losses: Stop trading. The exact definition of a loss can be supplied as: Negative Closed Trade Negative Strategy Cycle Negative Trading Session The guardian does not assume one definition. Cooldown rules can also be enforced. Example: No new risk for 30 minutes after 3 consecutive losses. The agent can identify whether the candidate trade occurs while that cooldown remains active. Daily profit-protection rules can also be enforced when explicitly supplied. Example: After reaching +$1,000 realized profit, do not allow account P&L to fall below +$600. The guardian does not create such a rule automatically. The skill can calculate portfolio heat. A common definition is: Portfolio Heat = Total Open Risk ÷ Account Equity Post-trade portfolio heat can also be calculated. Risk utilization can be reported for: Per-Trade Risk Daily Loss Total Open Risk Gross Exposure Leverage Margin Correlation Position Count It can also report remaining headroom: Remaining Daily Loss Capacity Remaining Open-Risk Capacity Remaining Exposure Capacity Remaining Leverage Capacity Remaining Correlation Capacity Remaining Position Slots The guardian identifies the binding constraint. For example: Per-Trade Limit Allows: 5 contracts Daily-Loss Headroom Allows: 3 contracts Portfolio Open-Risk Limit Allows: 4 contracts Leverage Allows: 2 contracts Correlation Allows: 2 contracts Maximum Permitted Size: 2 contracts Binding Constraints: Leverage Correlation If the user requested 5 contracts: REDUCE SIZE Maximum Permitted: 2 contracts The skill supports fail-closed risk architecture. For safety-critical agentic trading systems: If a required risk input is: Missing Stale Contradictory Unverifiable the guardian should not authorize additional risk. Examples include: Broker Reports Flat but Position List Contains an Open Position Used Margin Is Positive While Position State Reports No Exposure Daily P&L Timestamp Is Stale Instrument Tick Value Is Unknown Currency Conversion Is Missing Correlation State Is Required but Unavailable Depending on policy scope, the result can be: BLOCK TRADE or: STOP TRADING The guardian can detect account-state conflicts and explicitly label them. Example: ACCOUNT STATE CONFLICT Broker Position: 0 Portfolio Position List: 1 Open Contract Policy: Fail Closed Decision: STOP TRADING The guardian can use structured reason codes such as: RG-001 MISSING_STOP RG-002 INVALID_STOP_GEOMETRY RG-003 MISSING_INSTRUMENT_ECONOMICS RG-004 RISK_PER_TRADE_EXCEEDED RG-005 DAILY_LOSS_STOP RG-006 TOTAL_OPEN_RISK_EXCEEDED RG-007 LEVERAGE_LIMIT RG-008 MARGIN_BUFFER RG-009 MAX_OPEN_POSITIONS RG-010 CORRELATION_LIMIT RG-011 DRAWDOWN_STOP RG-012 LOSS_STREAK_STOP RG-013 MIN_REWARD_RISK RG-014 SIZE_INCREMENT RG-015 STALE_ACCOUNT_STATE RG-016 ACCOUNT_STATE_CONFLICT RG-017 INSTRUMENT_LIMIT RG-018 DIRECTIONAL_EXPOSURE RG-019 STRATEGY_EXPOSURE RG-020 CURRENCY_CONVERSION_MISSING Severity can be classified as: INFO LOW MEDIUM HIGH CRITICAL The guardian supports deterministic execution-adjacent decision logic. Given identical: Account State Candidate Trade Instrument Specification Risk Policy the numeric calculations and final decision should remain identical. This makes the agent suitable for auditable trading workflows. The skill can maintain a decision audit trail containing: Decision Timestamp Account-State Timestamp Policy Name Policy Version Candidate Trade Existing Positions Unit Risk Requested Risk Maximum Risk Budget Maximum Permitted Size Binding Constraints Post-Trade Open Risk Post-Trade Exposure Post-Trade Leverage Correlation Result Daily P&L Result Drawdown Result Loss-Streak Result Reward/Risk Result Final Decision Reason Codes Assumptions Unknowns The skill supports policy versioning. Recommended fields include: Policy Name Policy Version Effective Date Owner Change Log The guardian should never silently modify risk limits between decisions. For multi-agent systems, it recognizes concurrency risk. Example: Two trading agents independently observe: Remaining Open-Risk Capacity: $500 Agent A proposes: $400 risk Agent B proposes: $400 risk Each proposal appears acceptable independently. Together: $800 new risk Result: Portfolio limit breached. A robust architecture may therefore require: Centralized Portfolio State Atomic Risk Reservation Decision IDs Risk Reservation Expiry Pre-Execution Revalidation The skill can identify this race condition and design the risk-control workflow accordingly. A recommended architecture is: Strategy Agent → Candidate Trade → AI Trading Risk & Position Sizing Guardian → Authorization Decision → Human or Execution Layer → Broker Supporting services: Broker State Portfolio State Risk Policy Audit Log The guardian should be logically independent from the signal-generation agent. A strategy agent must not silently approve its own risk exception. The guardian can generate an authorization object such as: Decision ID Policy Version Instrument Direction Maximum Quantity Account-State Version Expiration Execution systems should reject orders that exceed the authorized size. Risk authorization can expire because: Positions Change Another Trade Executes P&L Changes Margin Changes Daily Stop Is Reached Correlation Exposure Changes The guardian can therefore require revalidation immediately before execution. It can also identify martingale or recovery-sizing behavior. If proposed size increases solely because previous trades lost and the supplied policy does not authorize such behavior: RECOVERY-SIZING / MARTINGALE RISK The guardian must not relax policy because a strategy claims that the next trade has greater confidence. Signal confidence is deliberately separated from risk permission. A 95% confidence signal does not automatically deserve more account risk than a 55% confidence signal. The same supplied risk policy applies unless the policy itself explicitly defines confidence-based sizing. This makes the skill particularly suitable for AI trading systems where an autonomous or semi-autonomous strategy agent can come close to order execution. The risk guardian acts as the independent control layer between opportunity generation and capital exposure. Its core objective is not to maximize trades. Its objective is to prevent unauthorized risk.

    AI Trading Risk & Position Sizing Guardian evaluates compliance with supplied risk rules, but no risk model can guarantee realized loss. A stop-loss order does not guarantee execution at the stop price. Actual losses can exceed planned risk because of: Price Gaps Slippage Thin Liquidity Partial Fills Rejected Orders Broker Failures Exchange Interruptions Connectivity Failures Market Halts Limit-Up / Limit-Down Conditions Extreme Volatility Operational Errors Instrument specifications can change. Margin requirements can change. Broker leverage rules can change. Exchange rules can change. Fees can change. Currency conversion rates can change. External specifications should therefore be verified when current accuracy materially affects the decision. Correlation is not stable. A correlation estimate depends on: Time Period Return Frequency Lookback Window Market Regime Data Quality Instrument Selection Assets that historically exhibited modest correlation can become highly correlated during market stress. Simple Pearson correlation does not capture every form of dependency. Portfolio correlation controls should therefore be treated as risk-management approximations rather than guarantees. Gross exposure and stop-defined open risk are different measures. A trade with a tight stop may have: Low Planned Stop Risk High Notional Exposure High Leverage High Gap Risk The guardian therefore applies multiple constraints rather than relying exclusively on stop distance. Position sizing for nonlinear products can require specialized models. Options, inverse derivatives, structured products, multi-leg spreads, and other nonlinear instruments may not be accurately represented by simple entry-to-stop mathematics. For those instruments, additional risk modeling may be required. Forex and international instruments may require currency conversion. If account currency differs from instrument P&L currency and a valid conversion rate is missing, monetary risk may not be calculable accurately. Paper-trading data may not reproduce: Real Slippage Queue Position Liquidity Partial Fills Exchange Latency Broker Rejections Broker state may arrive with latency. In agentic trading systems, a trade approved using one account snapshot may no longer be valid after: Another Order Fills P&L Changes Margin Changes A Position Closes A Position Opens Daily Loss Changes Correlation Exposure Changes Pre-execution revalidation may therefore be required. Multiple trading agents can create race conditions. Two independently valid trades can jointly violate portfolio limits if they are approved simultaneously without centralized state or risk reservation. A centralized portfolio-risk service or atomic reservation mechanism may be necessary in concurrent systems. This skill does not: Predict Market Direction Generate Trading Signals Guarantee Profitability Guarantee Capital Preservation Determine Appropriate Personal Risk Tolerance Choose Suitable Leverage for a Person Choose a Maximum Daily Loss for a Person Choose a Maximum Drawdown for a Person Guarantee Broker Execution Override Broker Controls Override Exchange Controls Override Regulatory Controls Override Prop-Firm Controls All risk limits used for authorization must come from an explicit policy or another authorized source. ALLOW means policy-compliant under the evaluated inputs and assumptions. It does not mean safe, profitable, recommended, or guaranteed.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 2 months ago

    Permissions

    Read Files
    Browser
    Write Files

    File Scopes

    **/*.md
    **/*.txt
    **/*.csv
    **/*.tsv
    **/*.json
    **/*.jsonl
    **/*.yaml
    **/*.yml
    **/*.xml
    **/*.html
    **/*.pdf
    **/*.doc
    **/*.docx
    **/*.xls
    **/*.xlsx
    **/*.ods
    **/*.png
    **/*.jpg
    **/*.jpeg
    **/*.webp
    **/trading/**
    **/risk/**
    **/risk-management/**
    **/risk-policy/**
    **/risk-policies/**
    **/position-sizing/**
    **/portfolio/**
    **/portfolios/**
    **/positions/**
    **/open-positions/**
    **/account/**
    **/account-state/**
    **/broker/**
    **/broker-state/**
    **/execution/**
    **/orders/**
    **/trades/**
    **/trade-proposals/**
    **/strategy/**
    **/strategies/**
    **/tradingview/**
    **/futures/**
    **/equities/**
    **/stocks/**
    **/forex/**
    **/fx/**
    **/crypto/**
    **/derivatives/**
    **/margin/**
    **/leverage/**
    **/exposure/**
    **/correlation/**
    **/drawdown/**
    **/daily-pnl/**
    **/pnl/**
    **/loss-streak/**
    **/sessions/**
    **/limits/**
    **/policy/**
    **/governance/**
    **/guardrails/**
    **/kill-switch/**
    **/audit/**
    **/audits/**
    **/logs/**
    **/reports/**
    **/calculations/**
    **/authorizations/**
    **/approvals/**
    **/risk-decisions/**
    **/instrument-specs/**
    **/contract-specs/**
    **/market-data/**
    **/paper-trading/**
    **/live-trading/**
    **/agentic-trading/**
    **/automation/**
    **/templates/**
    **/docs/**
    **/output/**

    Use Read Files to inspect authorized risk policies, account snapshots, broker-state exports, position files, trade proposals, instrument specifications, correlation matrices, portfolio reports, daily P&L reports, drawdown records, margin reports, strategy exposure data, and previous risk decisions. Use Write Files only when the user requests creation or updating of derived artifacts such as: Risk Decision Reports Position-Sizing Calculations Portfolio Risk Tables Pre-Trade Approval Records Risk Authorization Objects Reason-Code Logs Risk-Policy Documentation Exposure Summaries Correlation Summaries Risk Audit Trails Daily Risk Reports Use Browser only when current authoritative public information is necessary and authorized, such as: Official Exchange Contract Specifications Tick Size Tick Value Contract Multiplier Official Trading Sessions Official Margin Documentation Broker Commission Schedules Broker Instrument Specifications Prefer official exchange, broker, regulator, or platform documentation when external verification is required. Do not use public web information as a substitute for account-specific broker state. Always distinguish: VERIFIED ACCOUNT STATE Data supplied by an authorized account or broker source. USER-SUPPLIED POLICY Risk limits explicitly provided by the user or organization. VERIFIED INSTRUMENT SPECIFICATION Contract or instrument economics supported by an authoritative source. CALCULATED VALUE A value derived directly from verified or supplied inputs. ASSUMPTION A clearly disclosed modeling assumption. UNKNOWN A value that cannot currently be established. REQUIRES EXTERNAL VERIFICATION A public specification that should be confirmed before risk authorization. The guardian must not invent: Maximum Risk per Trade Maximum Daily Loss Maximum Drawdown Maximum Leverage Maximum Open Positions Maximum Correlation Risk Minimum Reward/Risk Suitable Position Size Margin Requirements Tick Value Pip Value Contract Multiplier Broker Fees Currency Conversion Risk Tolerance The operator defines policy. The guardian enforces policy. Apply decision precedence exactly: STOP TRADING BLOCK TRADE REDUCE SIZE ALLOW Do not allow a lower-priority result to override a higher-priority risk condition. Examples: An excellent reward/risk ratio cannot override a daily-loss stop. A high-confidence trading signal cannot override a drawdown stop. A tight stop cannot override leverage limits. A low net exposure cannot override high gross exposure. A different symbol name cannot override correlation concentration. A profitable strategy history cannot override account-level risk rules. Risk calculations should fail closed when a required safety-critical value is missing, stale, contradictory, or unverifiable. If a smaller valid quantity exists: REDUCE SIZE If no valid tradable quantity exists: BLOCK TRADE If an account-level halt condition exists: STOP TRADING Do not round a permissible position upward. For whole-contract instruments, round down to the nearest valid number of contracts. For lot- or share-based markets, respect the supplied minimum quantity and quantity step. Do not assume a stop-loss guarantees maximum realized loss. Explicitly recognize: Gap Risk Slippage Liquidity Risk Execution Failure Market Dislocation Broker Failure Exchange Interruption When existing positions are evaluated, distinguish: Risk to Stop Notional Exposure Gross Exposure Net Exposure Margin Correlation These are different concepts. When evaluating correlation, do not assume that opposite positions are perfect hedges. When evaluating leverage, use the policy's exact leverage definition. When daily P&L is evaluated, use the policy's exact realized/unrealized basis and session boundary. When drawdown is evaluated, use the policy's exact static, trailing, balance, or equity definition. When loss streaks are evaluated, use the policy's definition of a losing event. The guardian must remain independent from market direction. It must not convert: ALLOW into: BUY or: SELL ALLOW means only: The proposed trade complies with the evaluated risk policy at the evaluated size and under the supplied assumptions. This skill provides risk-control calculations and policy enforcement, not personalized investment advice or guarantees of capital protection.

    Listed2 months ago

    What's inside

    Frequently Asked Questions

    Unlike signal generators, this agent acts as an independent compliance layer that evaluates trade proposals against your specific risk policy to prevent over-leveraging, correlation spikes, and account-level blowups.

    The Guardian works with any systematic or discretionary workflow by acting as a pre-trade gatekeeper, accepting account and trade data via standard inputs to return an ALLOW, REDUCE, BLOCK, or STOP decision.

    The agent supports diverse instrument economics including linear equities, futures (tick/point value), forex (lots/pips), and crypto derivatives (inverse/linear contracts), ensuring sizing is calculated correctly for each asset class.

    The Guardian follows a 'fail-closed' architecture; if critical data like margin, tick value, or account equity is missing or stale, it will default to BLOCK TRADE or STOP TRADING to protect your capital.

    Yes, the agent can be configured to enforce strict drawdown limits and trailing thresholds, automatically halting new risk if your specific loss-streak or peak-to-equity targets are breached.

    You receive the fully configured Risk Guardian agent, a comprehensive template for defining your Risk Policy, and a standardized schema for the decision audit trail to ensure every trade is recorded for future review.

    New in Business & Operations