The Alert DSL

Why Build a DSL?

Hardcoding alert conditions in scripts means every change requires code edits, testing, and deployment. A DSL gives you:

• Composability: Combine conditions with AND/OR logic like building blocks
• Readability: rsi(14) crosses_above 30 AND volume > avg * 2 is self-documenting
• Evolution: Add new condition types without rewriting the engine
• Backtesting: Run the same conditions against historical data to validate
• User-editable: Change alert conditions via Discord commands without touching code

The market-alerts MCP server provides 7 tools: create_alert, list_alerts, update_alert, delete_alert, pause_alert, snooze_alert, get_alert_status.

Alert DSL Syntax

Condition Types

Price Conditions

Volume Conditions

Technical Indicators

Fundamental Triggers

Sentiment Conditions

Composite Operators

Time-Based

Full Grammar (EBNF)

alert       := condition (("AND" | "OR") condition)*
condition   := "NOT"? atom | "(" alert ")" | quantifier
atom        := field operator value
quantifier  := ("ALL_OF" | "ANY_OF") "(" alert ("," alert)* ")"
             | "N_OF" "(" number "," alert ("," alert)* ")"
field       := "price" | "volume" | "rsi" "(" number ")"
             | "macd" | "signal" | "macd_histogram"
             | "ema" "(" number ")" | "sma" "(" number ")"
             | "bb_width" | "bb_squeeze" | "atr" "(" number ")"
             | "pe_ratio" | "earnings_in_days" | "short_interest"
             | "stocktwits_bull_ratio" | "reddit_mentions_24h"
             | "news_sentiment" | "insider_buying_90d"
             | "relative_volume" | "volume_surge"
operator    := ">" | "<" | ">=" | "<=" | "==" | "!="
             | "crosses_above" | "crosses_below"
             | "within_range" | "IN"
value       := number | string | field | function_call
function_call := ("sma" | "ema" | "avg" | "max" | "min"
                | "percentile" | "sector_avg") "(" args ")"
number      := [0-9]+ ("." [0-9]+)? ("%")?
string      := '"' [^"]* '"'

Alert Templates

Pre-built alert patterns for common trading setups. Each template is a tested combination of conditions that has been validated through backtesting.

Setup Confirmation State Machine

Alerts don't just fire once. They track a setup through multiple states from initial detection to confirmation to execution.

function evaluateSetupState(conditions, currentData) {
  const total = conditions.length;
  const met = conditions.filter(c => evaluate(c, currentData)).length;
  const ratio = met / total;

  if (ratio === 0) return "EXPIRED";      // No conditions met
  if (ratio < 0.5) return "WATCHING";      // Less than half
  if (ratio < 0.75) return "FORMING";      // More than half
  if (ratio < 1.0) return "NEAR_CONFIRMED"; // Almost all
  if (ratio === 1.0) return "CONFIRMED";    // All conditions met!

  // After CONFIRMED, wait for entry trigger
  if (state === "CONFIRMED" && entryTriggered(currentData)) {
    return "TRIGGERED"; // Send alert + create trade log
  }
}

// Partial match reporting
function getPartialReport(conditions, currentData) {
  return conditions.map(c => ({
    condition: c.toString(),
    met: evaluate(c, currentData),
    currentValue: getCurrentValue(c.field, currentData),
    targetValue: c.value,
    progress: calculateProgress(c, currentData) // 0-100%
  }));
}

// Example output:
// {
//   state: "NEAR_CONFIRMED",
//   met: 5,
//   total: 6,
//   conditions: [
//     { condition: "rsi(14) < 30", met: true, current: 28.5, target: 30, progress: 100 },
//     { condition: "vol > avg*2", met: true, current: 2.3, target: 2.0, progress: 100 },
//     { condition: "macd > signal", met: false, current: -0.3, target: 0, progress: 85 },
//     ...
//   ]
// }

Building with market-alerts MCP

Creating Alerts Programmatically

create_alert({
  name: "AAPL Breakout",
  symbol: "AAPL",
  conditions: "price crosses_above 195 AND volume > avg_volume * 1.5 AND rsi(14) > 55",
  actions: ["discord_notify:#trading-signals", "log_setup"],
  expiry: "2026-03-15",
  priority: "high",
  metadata: {
    strategy: "momentum_breakout",
    entry: 195.00,
    stop: 188.50,
    tp1: 205.00,
    tp2: 215.00,
    rr: 1.54,
    compositeScore: 88.5,
    grade: "A+"
  }
})

// Response:
{
  "id": "alert_7f8a9b2c",
  "status": "active",
  "created": "2026-03-05T23:15:00Z",
  "nextCheck": "2026-03-06T09:30:00Z"
}

Batch Alert Creation from Scanner

// After scanner identifies 10 A+ setups:
for (const setup of scannerResults) {
  create_alert({
    name: `${setup.ticker} ${setup.strategy}`,
    symbol: setup.ticker,
    conditions: generateConditions(setup),
    actions: [
      setup.grade === "A++" ? "discord_notify:#priority-alerts" : "discord_notify:#trading-signals",
      "log_setup"
    ],
    expiry: addDays(today, setup.horizon === "swing" ? 10 : 30),
    priority: setup.grade === "A++" ? "critical" : "high",
    metadata: {
      entry: setup.entry,
      stop: setup.stop,
      tp1: setup.tp1,
      tp2: setup.tp2,
      rr: setup.rr,
      compositeScore: setup.composite,
      grade: setup.grade,
      scanDate: today
    }
  });
}

function generateConditions(setup) {
  switch (setup.strategy) {
    case "oversold_bounce":
      return `rsi(14) crosses_above 30 AND price > ${setup.entry * 0.98}
              AND volume > avg_volume * 1.3`;
    case "momentum_breakout":
      return `price crosses_above ${setup.entry}
              AND volume > avg_volume * 1.5 AND rsi(14) > 55`;
    case "safe_haven":
      return `price > ema(20) AND vix > 25`;
    default:
      return `price within_range(${setup.entry * 0.98}, ${setup.entry * 1.02})
              AND volume > avg_volume`;
  }
}

Managing Alert Lifecycle

// List all active alerts
list_alerts({ status: "active" })

// Pause during earnings blackout
pause_alert({ id: "alert_7f8a9b2c", reason: "AAPL earnings tomorrow" })

// Snooze for 24 hours (e.g., FOMC day)
snooze_alert({ id: "alert_7f8a9b2c", duration: "24h" })

// Check specific alert status
get_alert_status({ id: "alert_7f8a9b2c" })
// → { state: "NEAR_CONFIRMED", met: 4, total: 5, lastCheck: "..." }

// Update conditions (e.g., tighten after partial move)
update_alert({
  id: "alert_7f8a9b2c",
  conditions: "price crosses_above 193 AND volume > avg_volume * 1.3"
})

// Delete expired or irrelevant alerts
delete_alert({ id: "alert_7f8a9b2c" })

Auto-Detection of Setup Confirmations

The killer feature: after the scanner identifies setups, the system automatically monitors for entry conditions and fires a CONFIRMED alert when all conditions align.

// 1. Scanner runs at 23:00 → identifies CF as A+ oversold bounce
// 2. Auto-creates monitoring alert:
create_alert({
  name: "CF Oversold Bounce — Scanner Mar 5",
  symbol: "CF",
  conditions: `
    rsi(14) crosses_above 30
    AND price > 76.50
    AND volume > avg_volume * 1.5
    AND macd_histogram > 0
  `,
  actions: ["discord_notify:#confirmed-setups", "create_trade_log"],
  expiry: "2026-03-15",
  priority: "critical",
  metadata: {
    entry: [76.50, 78.50],
    stop: 73.80,
    tp1: 83.00,
    tp2: 88.50,
    rr: 1.67,
    thesis: "Fertilizer stock oversold on broad market selloff. Strong fundamentals, insider buying, sector tailwinds from energy rally.",
    grade: "A+"
  }
})

// 3. System checks every 15 min during market hours
// 4. Day 2, 10:45 AM: RSI crosses above 30 ✓, volume 1.8x ✓
//    → State: NEAR_CONFIRMED (3/4 met, waiting for MACD)
// 5. Day 2, 14:30: MACD histogram turns positive ✓
//    → State: CONFIRMED! All 4 conditions met!
// 6. Discord notification fires:

// ═══════════════════════════════════════
// 🟢 CONFIRMED: CF Oversold Bounce
// ═══════════════════════════════════════
// Score: 86.2 | Grade: A+
// Strategy: Oversold Bounce | Regime: Early Risk-Off
//
// Entry Zone: $76.50 – $78.50
// Stop Loss:  $73.80 (-4.1%)
// Target 1:   $83.00 (+6.4%)
// Target 2:   $88.50 (+14.3%)
// R/R:        1.67
//
// Thesis: Fertilizer stock oversold on broad market selloff.
// Strong fundamentals, insider buying, sector tailwinds.
//
// Conditions Met:
// ✅ RSI(14) crossed above 30 (current: 32.1)
// ✅ Price > $76.50 (current: $77.85)
// ✅ Volume 1.8x average
// ✅ MACD histogram positive (0.12)
// ═══════════════════════════════════════

The Evolving DSL

The DSL grows over time based on feedback from retrospectives. Each version adds new condition types or refines existing ones.

Version	New Conditions	Trigger
v1.0	price, volume, rsi, macd, ema/sma	Initial release
v1.1	bb_squeeze, bb_width, atr	Squeeze setups requested
v2.0	earnings_in_days, pe_ratio, insider_buying	Fundamental triggers needed
v2.1	stocktwits_bull_ratio, reddit_mentions	Social sentiment layer added
v2.2	crosses_above, crosses_below operators	State change detection
v3.0	N_OF quantifier, sector_avg, vix conditions	After retro showed 15% hit rate improvement with sector filter
v3.1	gap_up, gap_down, volume_profile_poc	Gap trading setups
v4.0	composite_score, ai_conviction, regime	Full multi-layer integration

// v2.0 alert (before sector filter):
"rsi(14) < 30 AND volume > avg * 2 AND price > sma(200)"

// v3.0 alert (after retro feedback — add sector + regime):
"rsi(14) < 30 AND volume > avg * 2 AND price > sma(200)
 AND sector_momentum > 0 AND regime IN (EARLY_RISK_OFF, RECOVERY)"

// Hit rate improvement: 42% → 57% (+15 percentage points)
// This is what the feedback loop (Part 6) enables.