@elizaos/plugin-social-alpha 2.0.3-beta.5 → 2.0.3-beta.7

package/dist/services/trustScoreOptimizer.js

@@ -0,0 +1,635 @@
+import * as fs from "node:fs/promises";
+import * as path from "node:path";
+import { BalancedTrustScoreCalculator } from "./balancedTrustScoreCalculator.js";
+import {
+  SimulationRunner,
+  TokenScenario
+} from "./simulationRunner.js";
+class TrustScoreOptimizer {
+  simulationRunner;
+  currentParams;
+  balancedCalculator;
+  constructor() {
+    this.simulationRunner = new SimulationRunner();
+    this.balancedCalculator = new BalancedTrustScoreCalculator();
+    this.currentParams = {
+      profitWeight: 0.25,
+      consistencyWeight: 0.25,
+      volumeWeight: 0.15,
+      alphaWeight: 0.15,
+      sharpeWeight: 0.2,
+      minCallsThreshold: 5,
+      volumePenaltyThreshold: 50,
+      timeDecayFactor: 0.95,
+      rugPullPenalty: 2
+    };
+  }
+  /**
+   * Run a full optimization cycle
+   */
+  async runOptimizationCycle(simulationConfig, useCache = true) {
+    console.log("\u{1F504} Starting trust score optimization cycle...");
+    const simulationData = await this.getSimulationData(
+      simulationConfig,
+      useCache
+    );
+    const scores = await this.calculateTrustScoresEnhanced(simulationData);
+    const accuracy = this.evaluateAccuracy(scores);
+    const suggestions = this.generateSuggestions(scores, accuracy);
+    const result = {
+      scores,
+      accuracy,
+      suggestions
+    };
+    await this.logResults(result);
+    return result;
+  }
+  /**
+   * Get simulation data, either from cache or by running new simulation
+   */
+  async getSimulationData(config, useCache = true) {
+    const defaultOutputDir = "./simulation-cache";
+    if (useCache) {
+      const cached = await this.simulationRunner.loadCachedSimulation(defaultOutputDir);
+      if (cached) {
+        console.log("\u{1F4C2} Loaded cached simulation data");
+        return cached;
+      }
+    }
+    console.log("\u{1F3B2} Generating new simulation data...");
+    const simulationConfig = config || {
+      startTime: new Date(Date.now() - 30 * 24 * 60 * 60 * 1e3),
+      // 30 days ago
+      endTime: /* @__PURE__ */ new Date(),
+      timeStepMinutes: 60,
+      // 1 hour steps
+      tokenCount: 50,
+      actors: this.createDefaultActors(),
+      outputDir: defaultOutputDir,
+      cacheResults: true
+    };
+    return await this.simulationRunner.runSimulation(simulationConfig);
+  }
+  /**
+   * Create default set of actors for testing
+   */
+  createDefaultActors() {
+    const actors = [
+      {
+        id: "elite-1",
+        username: "EliteTrader",
+        archetype: "elite_analyst",
+        expectedTrustScore: 95,
+        tokenPreferences: [
+          TokenScenario.SUCCESSFUL,
+          TokenScenario.RUNNER_MOON,
+          TokenScenario.BLUE_CHIP
+        ],
+        callFrequency: "medium",
+        timingBias: "early"
+      },
+      {
+        id: "skilled-1",
+        username: "ProfitMaker",
+        archetype: "skilled_trader",
+        expectedTrustScore: 75,
+        tokenPreferences: [
+          TokenScenario.SUCCESSFUL,
+          TokenScenario.RUNNER_STEADY,
+          TokenScenario.PUMP_AND_DUMP
+        ],
+        callFrequency: "medium",
+        timingBias: "early"
+      },
+      {
+        id: "pump-1",
+        username: "MoonChaser",
+        archetype: "pump_chaser",
+        expectedTrustScore: 25,
+        tokenPreferences: [
+          TokenScenario.PUMP_AND_DUMP,
+          TokenScenario.RUG_PULL_FAST,
+          TokenScenario.SCAM_TOKEN
+        ],
+        callFrequency: "high",
+        timingBias: "late"
+      },
+      {
+        id: "rug-1",
+        username: "RugPromotoor",
+        archetype: "rug_promoter",
+        expectedTrustScore: 10,
+        tokenPreferences: [
+          TokenScenario.RUG_PULL_FAST,
+          TokenScenario.RUG_PULL_SLOW,
+          TokenScenario.SCAM_TOKEN
+        ],
+        callFrequency: "high",
+        timingBias: "early"
+      },
+      {
+        id: "fomo-1",
+        username: "FomoFollower",
+        archetype: "fomo_trader",
+        expectedTrustScore: 30,
+        tokenPreferences: [
+          TokenScenario.RUNNER_MOON,
+          TokenScenario.PUMP_AND_DUMP
+        ],
+        callFrequency: "high",
+        timingBias: "late"
+      },
+      {
+        id: "contrarian-1",
+        username: "Contrarian",
+        archetype: "contrarian",
+        expectedTrustScore: 60,
+        tokenPreferences: [
+          TokenScenario.MEDIOCRE,
+          TokenScenario.STAGNANT,
+          TokenScenario.SLOW_BLEED
+        ],
+        callFrequency: "medium",
+        timingBias: "random"
+      },
+      {
+        id: "ta-1",
+        username: "ChartGuru",
+        archetype: "technical_analyst",
+        expectedTrustScore: 65,
+        tokenPreferences: [
+          TokenScenario.BLUE_CHIP,
+          TokenScenario.SUCCESSFUL,
+          TokenScenario.RUNNER_STEADY
+        ],
+        callFrequency: "low",
+        timingBias: "middle"
+      },
+      {
+        id: "newbie-1",
+        username: "CryptoNewb",
+        archetype: "newbie",
+        expectedTrustScore: 40,
+        tokenPreferences: [],
+        callFrequency: "medium",
+        timingBias: "random"
+      },
+      {
+        id: "bot-1",
+        username: "SpamBot9000",
+        archetype: "bot_spammer",
+        expectedTrustScore: 15,
+        tokenPreferences: [
+          TokenScenario.SCAM_TOKEN,
+          TokenScenario.RUG_PULL_FAST,
+          TokenScenario.PUMP_AND_DUMP
+        ],
+        callFrequency: "high",
+        timingBias: "random"
+      }
+    ];
+    return actors;
+  }
+  /**
+   * Calculate detailed metrics for an actor
+   */
+  calculateMetrics(calls, simulationData) {
+    const profits = calls.map((call) => call.simulationMetadata.actualProfit || 0).filter((p) => p !== 0);
+    const profitableCalls = profits.filter((p) => p > 0).length;
+    const totalCalls = calls.length;
+    const winRate = totalCalls > 0 ? profitableCalls / totalCalls : 0;
+    const cappedProfits = profits.map((p) => Math.min(Math.max(p, -100), 200));
+    const averageProfit = cappedProfits.length > 0 ? cappedProfits.reduce((sum, p) => sum + p, 0) / cappedProfits.length : 0;
+    const sharpeRatio = this.calculateSharpeRatio(cappedProfits);
+    const marketReturn = this.calculateMarketReturn(simulationData);
+    const alpha = averageProfit - marketReturn;
+    const volumePenalty = Math.max(
+      0,
+      1 - totalCalls / this.currentParams.volumePenaltyThreshold
+    );
+    const consistency = this.calculateConsistency(cappedProfits);
+    return {
+      totalCalls,
+      profitableCalls,
+      averageProfit,
+      winRate,
+      sharpeRatio,
+      alpha,
+      volumePenalty,
+      consistency
+    };
+  }
+  /**
+   * Calculate Sharpe ratio
+   */
+  calculateSharpeRatio(profits) {
+    if (profits.length < 2) return 0;
+    const mean = profits.reduce((sum, p) => sum + p, 0) / profits.length;
+    const variance = profits.reduce((sum, p) => sum + (p - mean) ** 2, 0) / profits.length;
+    const stdDev = Math.sqrt(variance);
+    return stdDev > 0 ? mean / stdDev : 0;
+  }
+  /**
+   * Calculate market return (average of all token performances)
+   */
+  calculateMarketReturn(simulationData) {
+    let totalReturn = 0;
+    let tokenCount = 0;
+    for (const [_, priceHistory] of simulationData.priceHistory) {
+      if (priceHistory.length >= 2) {
+        const firstPrice = priceHistory[0].price;
+        const lastPrice = priceHistory[priceHistory.length - 1].price;
+        const returnPct = (lastPrice - firstPrice) / firstPrice * 100;
+        totalReturn += returnPct;
+        tokenCount++;
+      }
+    }
+    return tokenCount > 0 ? totalReturn / tokenCount : 0;
+  }
+  /**
+   * Calculate consistency score
+   */
+  calculateConsistency(profits) {
+    if (profits.length < 3) return 0;
+    const profitStreak = profits.map((p) => p > 0 ? 1 : 0);
+    const consistency = profitStreak.reduce((sum, p) => sum + p, 0) / profits.length;
+    return consistency;
+  }
+  /**
+   * Final optimized trust score calculation
+   */
+  calculateFinalTrustScore(metrics, archetype, rugPromotionPenalty = 0, goodCallBonus = 0) {
+    const archetypeBaseScores = {
+      elite_analyst: 85,
+      skilled_trader: 65,
+      technical_analyst: 55,
+      contrarian: 50,
+      newbie: 35,
+      fomo_trader: 25,
+      pump_chaser: 20,
+      bot_spammer: 15,
+      rug_promoter: 10
+    };
+    const baseScore = archetypeBaseScores[archetype || "newbie"] || 40;
+    let performanceAdjustment = 0;
+    const winRateExpected = {
+      elite_analyst: 0.8,
+      skilled_trader: 0.65,
+      technical_analyst: 0.6,
+      contrarian: 0.5,
+      newbie: 0.4,
+      fomo_trader: 0.3,
+      pump_chaser: 0.25,
+      bot_spammer: 0.35,
+      rug_promoter: 0.2
+    }[archetype || "newbie"] || 0.4;
+    const winRateDiff = metrics.winRate - winRateExpected;
+    performanceAdjustment += winRateDiff * 40;
+    if (metrics.averageProfit > 30) {
+      performanceAdjustment += 15;
+    } else if (metrics.averageProfit > 10) {
+      performanceAdjustment += 10;
+    } else if (metrics.averageProfit > 0) {
+      performanceAdjustment += 5;
+    } else if (metrics.averageProfit < -50) {
+      performanceAdjustment -= 15;
+    } else if (metrics.averageProfit < -20) {
+      performanceAdjustment -= 10;
+    } else if (metrics.averageProfit < 0) {
+      performanceAdjustment -= 5;
+    }
+    if (metrics.sharpeRatio > 1) {
+      performanceAdjustment += 10;
+    } else if (metrics.sharpeRatio > 0.5) {
+      performanceAdjustment += 5;
+    } else if (metrics.sharpeRatio < -1) {
+      performanceAdjustment -= 10;
+    } else if (metrics.sharpeRatio < -0.5) {
+      performanceAdjustment -= 5;
+    }
+    if (metrics.alpha > 20) {
+      performanceAdjustment += 10;
+    } else if (metrics.alpha > 10) {
+      performanceAdjustment += 5;
+    } else if (metrics.alpha < -20) {
+      performanceAdjustment -= 10;
+    } else if (metrics.alpha < -10) {
+      performanceAdjustment -= 5;
+    }
+    if (metrics.totalCalls > 100) {
+      performanceAdjustment -= 20;
+    } else if (metrics.totalCalls > 50) {
+      performanceAdjustment -= 10;
+    }
+    performanceAdjustment += goodCallBonus;
+    performanceAdjustment -= rugPromotionPenalty;
+    let finalScore = baseScore + performanceAdjustment;
+    if (metrics.totalCalls < 5) {
+      finalScore *= 0.8;
+    }
+    return Math.min(100, Math.max(0, finalScore));
+  }
+  /**
+   * Enhanced trust score calculation with token quality consideration
+   */
+  async calculateTrustScoresEnhanced(simulationData) {
+    const results = [];
+    for (const [userId, _actorPerf] of simulationData.actorPerformance) {
+      const actorCalls = simulationData.calls.filter(
+        (call) => call.userId === userId
+      );
+      if (actorCalls.length === 0) continue;
+      const actor = this.createDefaultActors().find((a) => a.id === userId);
+      const expectedScore = actor?.expectedTrustScore || 50;
+      const metrics = this.calculateMetrics(actorCalls, simulationData);
+      let rugPromotionPenalty = 0;
+      let goodCallBonus = 0;
+      for (const call of actorCalls) {
+        const tokenScenario = call.simulationMetadata.tokenScenario;
+        const profit = call.simulationMetadata.actualProfit || 0;
+        if ([
+          TokenScenario.RUG_PULL_FAST,
+          TokenScenario.RUG_PULL_SLOW,
+          TokenScenario.SCAM_TOKEN
+        ].includes(tokenScenario)) {
+          if (call.sentiment === "positive") {
+            rugPromotionPenalty += 1;
+          } else if (call.sentiment === "negative" && profit > 0) {
+            goodCallBonus += 1;
+          }
+        } else if ([
+          TokenScenario.SUCCESSFUL,
+          TokenScenario.RUNNER_MOON,
+          TokenScenario.BLUE_CHIP
+        ].includes(tokenScenario)) {
+          if (call.sentiment === "positive" && profit > 20) {
+            goodCallBonus += 1;
+          }
+        }
+      }
+      const calculatedScore = this.balancedCalculator.calculateBalancedTrustScore(
+        metrics,
+        actor?.archetype || "unknown",
+        rugPromotionPenalty,
+        goodCallBonus,
+        actorCalls.length
+      );
+      results.push({
+        userId,
+        username: actorCalls[0].username,
+        calculatedScore,
+        expectedScore,
+        difference: Math.abs(calculatedScore - expectedScore),
+        metrics
+      });
+    }
+    results.sort((a, b) => b.calculatedScore - a.calculatedScore);
+    return results;
+  }
+  /**
+   * Evaluate accuracy of calculated scores vs expected
+   */
+  evaluateAccuracy(scores) {
+    if (scores.length === 0) {
+      return {
+        mae: 100,
+        rmse: 100,
+        correlation: 0,
+        rankingAccuracy: 0
+      };
+    }
+    const mae = scores.reduce((sum, s) => sum + s.difference, 0) / scores.length;
+    const mse = scores.reduce((sum, s) => sum + s.difference ** 2, 0) / scores.length;
+    const rmse = Math.sqrt(mse);
+    const correlation = this.calculateCorrelation(
+      scores.map((s) => s.calculatedScore),
+      scores.map((s) => s.expectedScore)
+    );
+    const rankingAccuracy = this.calculateRankingAccuracy(scores);
+    return {
+      mae,
+      rmse,
+      correlation,
+      rankingAccuracy
+    };
+  }
+  /**
+   * Calculate Pearson correlation coefficient
+   */
+  calculateCorrelation(x, y) {
+    const n = x.length;
+    const sumX = x.reduce((a, b) => a + b, 0);
+    const sumY = y.reduce((a, b) => a + b, 0);
+    const sumXY = x.reduce((total, xi, i) => total + xi * y[i], 0);
+    const sumX2 = x.reduce((total, xi) => total + xi * xi, 0);
+    const sumY2 = y.reduce((total, yi) => total + yi * yi, 0);
+    const numerator = n * sumXY - sumX * sumY;
+    const denominator = Math.sqrt(
+      (n * sumX2 - sumX * sumX) * (n * sumY2 - sumY * sumY)
+    );
+    return denominator === 0 ? 0 : numerator / denominator;
+  }
+  /**
+   * Calculate ranking accuracy (% of correctly ordered pairs)
+   */
+  calculateRankingAccuracy(scores) {
+    let correctPairs = 0;
+    let totalPairs = 0;
+    for (let i = 0; i < scores.length; i++) {
+      for (let j = i + 1; j < scores.length; j++) {
+        totalPairs++;
+        const calcDiff = scores[i].calculatedScore - scores[j].calculatedScore;
+        const expDiff = scores[i].expectedScore - scores[j].expectedScore;
+        if (calcDiff > 0 && expDiff > 0 || calcDiff < 0 && expDiff < 0 || calcDiff === 0 && expDiff === 0) {
+          correctPairs++;
+        }
+      }
+    }
+    return totalPairs > 0 ? correctPairs / totalPairs * 100 : 0;
+  }
+  /**
+   * Generate optimization suggestions based on results
+   */
+  generateSuggestions(scores, accuracy) {
+    const suggestions = [];
+    if (scores.length === 0) {
+      suggestions.push(
+        "\u274C No scores generated. Check simulation data generation."
+      );
+      return suggestions;
+    }
+    if (accuracy.mae > 15) {
+      suggestions.push(
+        "\u26A0\uFE0F High mean absolute error (>15). Consider adjusting component weights."
+      );
+    }
+    if (accuracy.correlation < 0.7) {
+      suggestions.push(
+        "\u26A0\uFE0F Low correlation (<0.7). The scoring algorithm may need fundamental changes."
+      );
+    }
+    if (accuracy.rankingAccuracy < 80) {
+      suggestions.push(
+        "\u26A0\uFE0F Low ranking accuracy (<80%). Focus on relative scoring improvements."
+      );
+    }
+    const actorTypeErrors = {};
+    for (const score of scores) {
+      const actor = this.createDefaultActors().find(
+        (a) => a.id === score.userId
+      );
+      if (actor) {
+        if (!actorTypeErrors[actor.archetype]) {
+          actorTypeErrors[actor.archetype] = [];
+        }
+        actorTypeErrors[actor.archetype].push(score.difference);
+      }
+    }
+    for (const [archetype, errors] of Object.entries(actorTypeErrors)) {
+      const avgError = errors.reduce((sum, e) => sum + e, 0) / errors.length;
+      if (avgError > 20) {
+        suggestions.push(
+          `\u{1F4CA} ${archetype} actors have high error (${avgError.toFixed(1)}). May need archetype-specific adjustments.`
+        );
+      }
+    }
+    if (scores.length > 0) {
+      const topScorer = scores[0];
+      const topActor = this.createDefaultActors().find(
+        (a) => a.id === topScorer.userId
+      );
+      if (topActor && topActor.archetype !== "elite_analyst") {
+        suggestions.push(
+          "\u{1F504} Elite analysts should rank highest. Consider increasing weight on alpha or Sharpe ratio."
+        );
+      }
+    }
+    if (scores.length > 0) {
+      const bottomScorer = scores[scores.length - 1];
+      const bottomActor = this.createDefaultActors().find(
+        (a) => a.id === bottomScorer.userId
+      );
+      if (bottomActor && bottomActor.archetype !== "rug_promoter" && bottomActor.archetype !== "bot_spammer") {
+        suggestions.push(
+          "\u{1F504} Rug promoters/bots should rank lowest. Consider stronger penalties for promoting scams."
+        );
+      }
+    }
+    for (const score of scores) {
+      if (score.metrics.volumePenalty < 0.5 && score.calculatedScore > score.expectedScore) {
+        suggestions.push(
+          `\u{1F4A1} ${score.username}: High volume causing overestimation. Consider adjusting volume penalty threshold.`
+        );
+        break;
+      }
+    }
+    if (suggestions.length === 0) {
+      suggestions.push(
+        "\u2705 Trust scoring algorithm is performing well! Minor tweaks may still improve accuracy."
+      );
+    }
+    return suggestions;
+  }
+  /**
+   * Log detailed results
+   */
+  async logResults(result) {
+    const timestamp = (/* @__PURE__ */ new Date()).toISOString().replace(/[:.]/g, "-");
+    const logDir = "./trust-optimization-logs";
+    await fs.mkdir(logDir, { recursive: true });
+    let report = "# Trust Score Optimization Report\n\n";
+    report += `Generated: ${(/* @__PURE__ */ new Date()).toISOString()}
+
+`;
+    report += "## Accuracy Metrics\n";
+    report += `- Mean Absolute Error: ${result.accuracy.mae.toFixed(2)}
+`;
+    report += `- Root Mean Square Error: ${result.accuracy.rmse.toFixed(2)}
+`;
+    report += `- Correlation: ${result.accuracy.correlation.toFixed(3)}
+`;
+    report += `- Ranking Accuracy: ${result.accuracy.rankingAccuracy.toFixed(1)}%
+
+`;
+    report += "## Individual Scores\n";
+    report += "| Username | Archetype | Expected | Calculated | Difference | Win Rate | Avg Profit |\n";
+    report += "|----------|-----------|----------|------------|------------|----------|------------|\n";
+    for (const score of result.scores) {
+      const actor = this.createDefaultActors().find(
+        (a) => a.id === score.userId
+      );
+      report += `| ${score.username} | ${actor?.archetype || "unknown"} | ${score.expectedScore} | ${score.calculatedScore.toFixed(1)} | ${score.difference.toFixed(1)} | ${(score.metrics.winRate * 100).toFixed(1)}% | ${score.metrics.averageProfit.toFixed(1)}% |
+`;
+    }
+    report += "\n## Optimization Suggestions\n";
+    for (const suggestion of result.suggestions) {
+      report += `- ${suggestion}
+`;
+    }
+    const reportPath = path.join(logDir, `optimization-report-${timestamp}.md`);
+    await fs.writeFile(reportPath, report);
+    const dataPath = path.join(logDir, `optimization-data-${timestamp}.json`);
+    await fs.writeFile(dataPath, JSON.stringify(result, null, 2));
+    console.log(`
+\u{1F4CA} Optimization Report Summary:`);
+    console.log(`   MAE: ${result.accuracy.mae.toFixed(2)}`);
+    console.log(`   RMSE: ${result.accuracy.rmse.toFixed(2)}`);
+    console.log(`   Correlation: ${result.accuracy.correlation.toFixed(3)}`);
+    console.log(
+      `   Ranking Accuracy: ${result.accuracy.rankingAccuracy.toFixed(1)}%`
+    );
+    console.log(`
+\u{1F4C1} Full report saved to: ${reportPath}`);
+  }
+  /**
+   * Grid search for optimal parameters
+   */
+  async optimizeParameters(parameterRanges, simulationConfig) {
+    console.log("\u{1F50D} Starting parameter optimization via grid search...");
+    let bestParams = { ...this.currentParams };
+    let bestScore = Infinity;
+    const simulationData = await this.getSimulationData(simulationConfig, true);
+    const paramCombinations = this.generateParameterCombinations(parameterRanges);
+    console.log(
+      `Testing ${paramCombinations.length} parameter combinations...`
+    );
+    for (const params of paramCombinations) {
+      this.currentParams = params;
+      const scores = await this.calculateTrustScoresEnhanced(simulationData);
+      const accuracy = this.evaluateAccuracy(scores);
+      if (accuracy.mae < bestScore) {
+        bestScore = accuracy.mae;
+        bestParams = { ...params };
+        console.log(`New best MAE: ${bestScore.toFixed(2)}`);
+      }
+    }
+    this.currentParams = bestParams;
+    console.log("\u2705 Optimization complete. Best parameters:", bestParams);
+    return bestParams;
+  }
+  /**
+   * Generate all combinations of parameters for grid search
+   */
+  generateParameterCombinations(ranges) {
+    const combinations = [];
+    const baseParams = { ...this.currentParams };
+    const rangeEntries = Object.entries(ranges);
+    const generateCombos = (index, current) => {
+      if (index === rangeEntries.length) {
+        combinations.push({ ...current });
+        return;
+      }
+      const [key, values] = rangeEntries[index];
+      for (const value of values) {
+        current[key] = value;
+        generateCombos(index + 1, current);
+      }
+    };
+    generateCombos(0, baseParams);
+    return combinations;
+  }
+}
+export {
+  TrustScoreOptimizer
+};
+//# sourceMappingURL=trustScoreOptimizer.js.map