@ak--47/dungeon-master 1.5.4 → 1.6.0

package/dungeons/vertical/marketplace.js

@@ -1,1019 +0,0 @@
-// ── IMPORTS ──
-import dayjs from "dayjs";
-import utc from "dayjs/plugin/utc.js";
-dayjs.extend(utc);
-import "dotenv/config";
-import * as u from "../../lib/utils/utils.js";
-import * as v from "ak-tools";
-/** @typedef  {import("../../types").Dungeon} Config */
-
-// ── OVERVIEW ──
-/*
- * NAME:       TradeNest
- * APP:        Two-sided marketplace connecting sellers who list products with
- *             buyers who search, purchase, and review. Sellers list items;
- *             buyers browse, message, negotiate, and buy. Revenue from
- *             marketplace fees and listing fees.
- * SCALE:      10,000 users, ~1.4M events, 121 days (2026-01-01 → 2026-05-01)
- * CORE LOOP:  search → view → add to cart → purchase → review
- *             (seller side: create listing → receive offers → accept → ship)
- *
- * EVENTS (18):
- *   item searched (8) > app session (8) > item viewed (7) > notification received (6)
- *   > add to cart (5) > message sent (5) > listing created (4) > purchase completed (3)
- *   > listing updated (3) > offer received (3) > shipping updated (3) > offer accepted (2)
- *   > review submitted (2) > seller rated (2) > profile updated (2) > account created (1)
- *   > refund requested (1) > account deactivated (1)
- *
- * FUNNELS (5):
- *   - Buyer Onboarding:      account created → item searched → item viewed → add to cart (40%)
- *   - Browse to Purchase:    item searched → item viewed → add to cart → purchase completed (30%)
- *   - Seller Listing Flow:   listing created → listing updated → offer received → offer accepted (25%)
- *   - Offer Negotiation:     offer received → message sent → offer accepted (35%)
- *   - Review After Purchase: purchase completed → shipping updated → review submitted (20%)
- *
- * USER PROPS:  user_type, segment, seller_rating, total_transactions, response_time_hours,
- *              store_name, Platform, category
- * SUPER PROPS: Platform, category
- * SCD PROPS:   seller_tier (new/verified/power/featured, monthly fuzzy, max 8)
- * GROUPS:      none
- */
-
-// ── HOOK STORIES ──
-/*
- * NOTE: All cohort effects are HIDDEN — no flag stamping. Discoverable via
- * raw-prop breakdowns (segment, day, category) or behavioral cohorts.
- *
- * ───────────────────────────────────────────────────────────────
- * 1. FEE CHANGE IMPACT (everything hook)
- * ───────────────────────────────────────────────────────────────
- *
- * PATTERN: After day 45 (permanent marketplace fee increase), all
- * "listing created" events get listing_fee multiplied by 1.3x.
- * Simulates the revenue impact of a platform fee adjustment. No flag —
- * discover via line chart of avg listing_fee over time.
- *
- * HOW TO FIND IT IN MIXPANEL:
- *
- *   Report 1: Listing Fee Over Time
- *   • Report type: Insights
- *   • Event: "listing created"
- *   • Measure: Average of "listing_fee"
- *   • Line chart by week
- *   • Expected: Clear step-up around day 45 from ~$15 avg to ~$20 avg
- *
- * REAL-WORLD ANALOGUE: Marketplace platforms periodically adjust
- * commission/listing fees, impacting seller economics and behavior.
- *
- * ───────────────────────────────────────────────────────────────
- * 2. WEEKEND SHOPPING SURGE (event hook)
- * ───────────────────────────────────────────────────────────────
- *
- * PATTERN: "purchase completed" events on Saturday or Sunday get
- * total_amount multiplied by 1.2x. Weekend shoppers spend more
- * per transaction due to leisure browsing.
- *
- * HOW TO FIND IT IN MIXPANEL:
- *
- *   Report 1: Weekend vs Weekday Average Spend
- *   • Report type: Insights
- *   • Event: "purchase completed"
- *   • Measure: Average of "total_amount"
- *   • Breakdown: Day of week
- *   • Expected: Sat/Sun avg ≈ $72 vs Mon-Fri avg ≈ $60
- *
- * REAL-WORLD ANALOGUE: E-commerce platforms see higher average
- * order values on weekends when buyers browse leisurely.
- *
- * ───────────────────────────────────────────────────────────────
- * 3. SELLER SUCCESS → BUYER TRUST (everything hook)
- * ───────────────────────────────────────────────────────────────
- *
- * PATTERN: Users with user_type "seller" and segment "power_seller"
- * (from meta.profile) get their "purchase completed" events cloned
- * 2x. Power sellers attract more transactions due to trust/reputation.
- *
- * HOW TO FIND IT IN MIXPANEL:
- *
- *   Report 1: Purchases per User by Segment
- *   • Report type: Insights
- *   • Event: "purchase completed"
- *   • Measure: Total per user
- *   • Breakdown: user property "segment"
- *   • Expected: power_seller ≈ 2x purchases vs casual_seller
- *
- * REAL-WORLD ANALOGUE: High-rated sellers with established reputations
- * receive disproportionately more sales on platforms like eBay/Etsy.
- *
- * ───────────────────────────────────────────────────────────────
- * 4. SEARCH-TO-PURCHASE BY CATEGORY (everything hook)
- * ───────────────────────────────────────────────────────────────
- *
- * PATTERN: Users whose events include "item searched" with category
- * "electronics" get cloned "purchase completed" events (electronics
- * buyers have higher conversion). Creates category-specific lift.
- *
- * HOW TO FIND IT IN MIXPANEL:
- *
- *   Report 1: Purchase Volume by Category
- *   • Report type: Insights
- *   • Event: "purchase completed"
- *   • Measure: Total
- *   • Breakdown: "category" (superProp)
- *   • Expected: electronics ≈ 1.5-2x more purchases than clothing/home
- *
- * REAL-WORLD ANALOGUE: Electronics shoppers have higher purchase
- * intent — they research, decide, and buy with less browsing.
- *
- * ───────────────────────────────────────────────────────────────
- * 5. RESPONSE TIME → CONVERSION (everything hook)
- * ───────────────────────────────────────────────────────────────
- *
- * PATTERN: Users whose average response_time_hours on "message sent"
- * events is <= 4 hours get additional "offer accepted" events cloned
- * (2x existing + 60% of offer_received). Slow responders (> 4h avg)
- * lose 60% of offer_accepted events. Fast responders close more deals.
- *
- * HOW TO FIND IT IN MIXPANEL:
- *
- *   Report 1: Offer Accepted Count vs Response Time
- *   • Report type: Insights
- *   • Event: "offer accepted"
- *   • Measure: Total per user
- *   • Breakdown: user property "response_time_hours"
- *   • Expected: Users with < 2hr avg response ≈ 2x offer accepts
- *
- * REAL-WORLD ANALOGUE: Sellers who respond quickly to inquiries
- * close significantly more deals on marketplace platforms.
- *
- * ───────────────────────────────────────────────────────────────
- * 6. NEW SELLER CHURN (everything hook)
- * ───────────────────────────────────────────────────────────────
- *
- * PATTERN: Users with segment "new_seller" and fewer than 10 total
- * events lose 40% of events after day 14. Simulates new sellers
- * who try the platform and abandon it quickly.
- *
- * HOW TO FIND IT IN MIXPANEL:
- *
- *   Report 1: New Seller Retention
- *   • Report type: Insights
- *   • Event: All events
- *   • Measure: Total per user
- *   • Filter: segment = "new_seller"
- *   • Line chart by week
- *   • Expected: Sharp drop-off around week 2-3
- *
- *   Report 2: Active Users by Segment
- *   • Report type: Insights
- *   • Event: Any active event
- *   • Measure: Uniques
- *   • Breakdown: user property "segment"
- *   • Expected: new_seller retention ~60% vs power_seller ~99%
- *
- * REAL-WORLD ANALOGUE: Most new sellers on marketplace platforms
- * churn within the first month if they don't get early traction.
- *
- * ───────────────────────────────────────────────────────────────
- * 7. POWER SELLER PROFILES (user hook)
- * ───────────────────────────────────────────────────────────────
- *
- * PATTERN: Users with segment "power_seller" get total_transactions
- * set to 100-500, seller_rating to 4.5-5.0, and store_name to a
- * realistic business name. Enriches profiles for realistic cohort analysis.
- *
- * HOW TO FIND IT IN MIXPANEL:
- *
- *   Report 1: Transaction Count Distribution by Segment
- *   • Report type: Insights
- *   • Measure: Profiles → Breakdown by "total_transactions" (histogram)
- *   • Filter: segment = "power_seller"
- *   • Expected: power_seller total_transactions 100-500, others ≈ 0
- *
- *   Report 2: Seller Rating by Segment
- *   • Report type: Insights
- *   • Measure: Profiles → Average of "seller_rating"
- *   • Breakdown: "segment"
- *   • Expected: power_seller ≈ 4.7 avg, casual_seller ≈ 3.5
- *
- * REAL-WORLD ANALOGUE: Top sellers have established storefronts,
- * high transaction volumes, and near-perfect ratings.
- *
- * ───────────────────────────────────────────────────────────────
- * 8. FREQUENT BUYER FUNNEL LIFT (everything hook)
- * ───────────────────────────────────────────────────────────────
- *
- * PATTERN: Non-frequent-buyer users lose ~25% of "purchase completed"
- * events (last step of Browse to Purchase funnel), simulating lower
- * conversion. Frequent buyers retain all purchase events.
- *
- * HOW TO FIND IT IN MIXPANEL:
- *
- *   Report 1: Browse-to-Purchase Conversion by Segment
- *   • Report type: Funnels
- *   • Steps: "item searched" → "item viewed" → "add to cart" → "purchase completed"
- *   • Breakdown: user property "segment"
- *   • Expected: frequent_buyer ≈ 39% vs window_shopper ≈ 20%
- *
- * REAL-WORLD ANALOGUE: Returning buyers have established trust
- * and familiarity with the platform, converting at higher rates.
- *
- * ───────────────────────────────────────────────────────────────
- * 9. BROWSE TO PURCHASE TIME-TO-CONVERT (funnel-post)
- * ───────────────────────────────────────────────────────────────
- *
- * PATTERN: Power_seller and frequent_buyer users complete the
- * Browse to Purchase funnel 1.4x faster (factor 0.71); window_shopper
- * 1.4x slower (factor 1.4).
- *
- * HOW TO FIND IT IN MIXPANEL:
- *
- *   Report 1: Browse to Purchase Median Time-to-Convert by Segment
- *   - Funnels > "item searched" -> "item viewed" -> "add to cart" -> "purchase completed"
- *   - Measure: Median time to convert
- *   - Breakdown: segment
- *   - Expected: power/frequent ~ 0.71x; window ~ 1.4x
- *
- *   NOTE (funnel-post measurement): visible only via Mixpanel funnel
- *   median TTC. Cross-event MIN→MIN SQL queries on raw events do NOT
- *   show this — funnel-post adjusts gaps within funnel instances, not
- *   across the user's full event history.
- *
- * ───────────────────────────────────────────────────────────────
- * 10. MESSAGE-COUNT MAGIC NUMBER (in-funnel, everything)
- * ───────────────────────────────────────────────────────────────
- *
- * PATTERN: Sweet 2-5 message-sent events between item-viewed and
- * offer-received → +35% on offer_amount of offer-received events.
- * Over 6+ → drop 25% of purchase-completed events (haggling deadlock).
- * No flag.
- *
- * HOW TO FIND IT IN MIXPANEL:
- *
- *   Report 1: Avg Offer Amount by Message Bucket
- *   - Cohort A: users with 2-5 messages between item-viewed and offer-received
- *   - Cohort B: users with 0-1
- *   - Event: "offer received"
- *   - Measure: Average of "offer_amount"
- *   - Expected: A ~ 1.35x B
- *
- *   Report 2: Purchase Rate on Heavy Messengers
- *   - Cohort C: users with >= 6 messages
- *   - Cohort A: users with 2-5
- *   - Event: "purchase completed"
- *   - Measure: Total per user
- *   - Expected: C ~ 25% fewer purchases per user
- *
- * REAL-WORLD ANALOGUE: A few quick clarifying messages close deals;
- * extended haggling kills conversion.
- *
- * ═══════════════════════════════════════════════════════════════
- * EXPECTED METRICS SUMMARY
- * ═══════════════════════════════════════════════════════════════
- *
- * Hook                        | Metric              | Baseline | Effect    | Ratio
- * ----------------------------|---------------------|----------|-----------|------
- * Fee Change Impact           | listing_fee post-D45| 1x       | 1.3x      | 1.3x
- * Weekend Shopping Surge      | total_amount Sat/Sun| 1x       | 1.2x      | 1.2x
- * Seller Success → Trust      | purchases/user      | 1x       | 2x        | 2x
- * Electronics Category Lift   | electronics purch.  | 1x       | 1.5x      | 1.5x
- * Response Time → Conversion  | offer_accepted/user | 1x       | 2x        | 2x
- * New Seller Churn            | events post-D14     | 1x       | 0.6x      | -40%
- * Power Seller Profiles       | total_transactions  | 0        | 100-500   | --
- * Frequent Buyer Funnel       | funnel conversion   | 30%      | 39%       | 1.3x
- * Browse to Purchase T2C      | median min by segment| 1x      | 0.71/1.4x | ~ 2x range
- * Message Magic Number        | sweet offer_amount  | 1x       | 1.35x     | 1.35x
- * Message Magic Number        | over purchases/user | 1x       | 0.75x     | -25%
- */
-
-// ── SCALE ──
-const SEED = "dm4-marketplace";
-const NUM_USERS = 10_000;
-const DATASET_START = "2026-01-01T00:00:00Z";
-const DATASET_END = "2026-05-01T23:59:59Z";
-const EVENTS_PER_DAY = 1.2;
-const token = process.env.MP_TOKEN || "your-mixpanel-token";
-
-const chance = u.initChance(SEED);
-
-// ── KNOBS (tweak these to reshape stories) ──
-const FEE_CHANGE_DAY = 45;
-const FEE_CHANGE_MULT = 1.3;
-
-const WEEKEND_SPEND_MULT = 1.2;
-
-const POWER_SELLER_CLONE_LIKELIHOOD = 65;
-
-const ELECTRONICS_CLONE_LIKELIHOOD = 40;
-const ELECTRONICS_MAX_CLONES = 3;
-
-const FAST_RESPONDER_HASH_MOD = 5;
-const FAST_RESPONDER_HASH_THRESHOLD = 2;
-const FAST_OFFER_CLONE_COUNT = 2;
-const FAST_OFFER_FROM_RECEIVED_LIKELIHOOD = 60;
-const SLOW_OFFER_DROP_LIKELIHOOD = 60;
-
-const NEW_SELLER_CUTOFF_DAYS = 14;
-const NEW_SELLER_DROP_LIKELIHOOD = 50;
-
-const FREQUENT_BUYER_DROP_LIKELIHOOD = 25;
-
-const TTC_FAST_FACTOR = 0.71;
-const TTC_SLOW_FACTOR = 1.4;
-
-const MSG_SWEET_MIN = 2;
-const MSG_SWEET_MAX = 5;
-const MSG_OVER_THRESHOLD = 6;
-const MSG_OFFER_BOOST = 1.35;
-const MSG_PURCHASE_DROP_LIKELIHOOD = 25;
-
-// ── DATA ARRAYS ──
-// Generate consistent seller store and listing IDs at module level
-const storeIds = v.range(1, 150).map(() => `STORE_${v.uid(6)}`);
-const listingIds = v.range(1, 500).map(() => `LST_${v.uid(8)}`);
-
-// ── HELPER FUNCTIONS ──
-function handleUserHooks(record) {
-	// H7: Power seller profiles — enrich segment-tagged profiles
-	if (record.segment === "power_seller") {
-		record.total_transactions = chance.integer({ min: 100, max: 500 });
-		record.seller_rating = chance.floating({ min: 4.5, max: 5.0, fixed: 1 });
-		record.store_name = chance.pickone([
-			"TechVault Pro", "StyleHaven", "GearFactory", "PrimeFinds",
-			"TopShelf Goods", "EliteDeals", "QualityFirst Store", "BestBuy Resale",
-		]);
-	} else if (record.segment === "casual_seller") {
-		record.total_transactions = chance.integer({ min: 5, max: 50 });
-		record.seller_rating = chance.floating({ min: 3.0, max: 4.5, fixed: 1 });
-		record.store_name = chance.pickone([
-			"My Garage Sale", "Closet Cleanout", "Random Finds", "Weekend Seller",
-		]);
-	} else if (record.segment === "new_seller") {
-		record.total_transactions = chance.integer({ min: 0, max: 3 });
-		record.seller_rating = 0;
-		record.store_name = "New Store";
-	}
-	return record;
-}
-
-function handleFunnelPostHooks(record, meta) {
-	// H9: Browse-to-purchase TTC scaled by segment
-	const segment = meta?.profile?.segment;
-	if (Array.isArray(record) && record.length > 1) {
-		const factor = (
-			segment === "power_seller" || segment === "frequent_buyer" ? TTC_FAST_FACTOR :
-			segment === "window_shopper" ? TTC_SLOW_FACTOR :
-			1.0
-		);
-		if (factor !== 1.0) {
-			for (let i = 1; i < record.length; i++) {
-				const prev = dayjs(record[i - 1].time);
-				const newGap = Math.round(dayjs(record[i].time).diff(prev) * factor);
-				record[i].time = prev.add(newGap, "milliseconds").toISOString();
-			}
-		}
-	}
-	return record;
-}
-
-function handleEverythingHooks(record, meta) {
-	const datasetStart = dayjs.unix(meta.datasetStart);
-	const FEE_CHANGE_CUTOFF = datasetStart.add(FEE_CHANGE_DAY, "days");
-	let events = record;
-	if (!events.length) return record;
-
-	const profile = meta.profile;
-
-	// Stamp superProps from profile so they are consistent per user.
-	if (profile) {
-		events.forEach(e => {
-			if (profile.Platform) e.Platform = profile.Platform;
-			if (profile.category) e.category = profile.category;
-		});
-	}
-
-	// H1: Fee change impact — listings after d45 get listing_fee 1.3x.
-	// In everything hook so timestamp comparison sees post-bunchIntoSessions times.
-	events.forEach(e => {
-		if (e.event === "listing created" && dayjs(e.time).isAfter(FEE_CHANGE_CUTOFF)) {
-			e.listing_fee = Math.floor((e.listing_fee || 15) * FEE_CHANGE_MULT);
-		}
-	});
-
-	// H8: Frequent buyer conversion filter — non-frequent-buyer users
-	// drop ~25% of "purchase completed" events.
-	if (profile && profile.segment !== "frequent_buyer") {
-		record = record.filter(e => {
-			if (e.event === "purchase completed" && chance.bool({ likelihood: FREQUENT_BUYER_DROP_LIKELIHOOD })) return false;
-			return true;
-		});
-		events = record;
-	}
-
-	// H3: Seller success → buyer trust — power sellers get 2x purchase events cloned.
-	if (profile && profile.segment === "power_seller") {
-		const purchases = events.filter(e => e.event === "purchase completed");
-		const templatePurchase = purchases[0];
-		if (templatePurchase && purchases.length > 0) {
-			purchases.forEach(p => {
-				if (chance.bool({ likelihood: POWER_SELLER_CLONE_LIKELIHOOD })) {
-					events.push({
-						...templatePurchase,
-						time: dayjs(p.time).add(chance.integer({ min: 1, max: 48 }), "hours").toISOString(),
-						user_id: p.user_id,
-						total_amount: chance.integer({ min: 15, max: 400 }),
-						item_count: chance.integer({ min: 1, max: 4 }),
-					});
-				}
-			});
-		}
-	}
-
-	// H4: Search-to-purchase by category — electronics gets cloned purchase events.
-	const hasElectronicsSearch = events.some(e =>
-		e.event === "item searched" && e.category === "electronics"
-	);
-	if (hasElectronicsSearch) {
-		const templatePurchase = events.find(e => e.event === "purchase completed");
-		if (templatePurchase) {
-			const electronicsSearches = events.filter(e =>
-				e.event === "item searched" && e.category === "electronics"
-			);
-			electronicsSearches.slice(0, ELECTRONICS_MAX_CLONES).forEach(search => {
-				if (chance.bool({ likelihood: ELECTRONICS_CLONE_LIKELIHOOD })) {
-					events.push({
-						...templatePurchase,
-						time: dayjs(search.time).add(chance.integer({ min: 1, max: 12 }), "hours").toISOString(),
-						user_id: search.user_id,
-						total_amount: chance.integer({ min: 50, max: 300 }),
-						category: "electronics",
-					});
-				}
-			});
-		}
-	}
-
-	// H5: Response time → conversion — deterministic fast/slow cohorts via user hash.
-	// Fast (~40%): set response_time_hours 1-4, clone offer_accepted.
-	// Slow (~60%): set 8-36, drop 60% of offer_accepted events.
-	const msgEvents = events.filter(e => e.event === "message sent");
-	if (msgEvents.length > 0) {
-		const uid = msgEvents[0].user_id || "";
-		const isFast = (uid.charCodeAt(0) + uid.charCodeAt(uid.length - 1)) % FAST_RESPONDER_HASH_MOD < FAST_RESPONDER_HASH_THRESHOLD;
-		msgEvents.forEach(m => {
-			m.response_time_hours = isFast
-				? chance.floating({ min: 0.5, max: 4, fixed: 1 })
-				: chance.floating({ min: 8, max: 36, fixed: 1 });
-		});
-		const templateOffer = events.find(e => e.event === "offer accepted");
-		if (isFast && templateOffer) {
-			// Fast responders: clone existing offer_accepted 2x each
-			const existingAccepts = events.filter(e => e.event === "offer accepted");
-			existingAccepts.forEach(accept => {
-				for (let c = 0; c < FAST_OFFER_CLONE_COUNT; c++) {
-					events.push({
-						...accept,
-						time: dayjs(accept.time).add(chance.integer({ min: 1, max: 8 }), "hours").toISOString(),
-						user_id: accept.user_id,
-						response_time_hours: chance.floating({ min: 0.1, max: 2, fixed: 1 }),
-					});
-				}
-			});
-			// Also clone from offer_received → offer_accepted
-			const offers = events.filter(e => e.event === "offer received");
-			offers.forEach(offer => {
-				if (chance.bool({ likelihood: FAST_OFFER_FROM_RECEIVED_LIKELIHOOD })) {
-					events.push({
-						...templateOffer,
-						time: dayjs(offer.time).add(chance.integer({ min: 1, max: 6 }), "hours").toISOString(),
-						user_id: offer.user_id,
-						response_time_hours: chance.floating({ min: 0.1, max: 3, fixed: 1 }),
-					});
-				}
-			});
-		} else if (!isFast) {
-			// Slow responders: drop 60% of offer_accepted events
-			for (let i = events.length - 1; i >= 0; i--) {
-				if (events[i].event === "offer accepted" && chance.bool({ likelihood: SLOW_OFFER_DROP_LIKELIHOOD })) {
-					events.splice(i, 1);
-				}
-			}
-		}
-	}
-
-	// H6: New seller churn — drop 50% of events after user's first 14 days.
-	if (profile && profile.segment === "new_seller") {
-		const firstEventTime = events.length > 0 ? dayjs(events[0].time) : null;
-		if (firstEventTime) {
-			const userCutoff = firstEventTime.add(NEW_SELLER_CUTOFF_DAYS, "days");
-			for (let i = events.length - 1; i >= 0; i--) {
-				if (dayjs(events[i].time).isAfter(userCutoff) && chance.bool({ likelihood: NEW_SELLER_DROP_LIKELIHOOD })) {
-					events.splice(i, 1);
-				}
-			}
-		}
-	}
-
-	// H2: Weekend shopping surge — Sat/Sun purchases get total_amount 1.2x.
-	events.forEach(e => {
-		if (e.event === "purchase completed") {
-			const dow = new Date(e.time).getUTCDay();
-			if (dow === 0 || dow === 6) {
-				e.total_amount = Math.floor((e.total_amount || 60) * WEEKEND_SPEND_MULT);
-			}
-		}
-	});
-
-	// H10: Message-count magic number (in-funnel, no flags)
-	// Sweet 2-5 messages between item-viewed and offer-received →
-	// +35% on offer-received offer_amount. Over 6+ → drop 25% of purchases.
-	const itemViewed = events.find(e => e.event === "item viewed");
-	const offerReceived = events.find(e => e.event === "offer received");
-	if (itemViewed && offerReceived) {
-		const aTime = dayjs(itemViewed.time);
-		const bTime = dayjs(offerReceived.time);
-		const msgBetween = events.filter(e =>
-			e.event === "message sent" &&
-			dayjs(e.time).isAfter(aTime) &&
-			dayjs(e.time).isBefore(bTime)
-		).length;
-		if (msgBetween >= MSG_SWEET_MIN && msgBetween <= MSG_SWEET_MAX) {
-			events.forEach(e => {
-				if (e.event === "offer received" && typeof e.offer_amount === "number") {
-					e.offer_amount = Math.round(e.offer_amount * MSG_OFFER_BOOST);
-				}
-			});
-		} else if (msgBetween >= MSG_OVER_THRESHOLD) {
-			for (let i = events.length - 1; i >= 0; i--) {
-				if (events[i].event === "purchase completed" && chance.bool({ likelihood: MSG_PURCHASE_DROP_LIKELIHOOD })) {
-					events.splice(i, 1);
-				}
-			}
-		}
-	}
-
-	return record;
-}
-
-// ── CONFIG ──
-/** @type {Config} */
-const config = {
-	version: 2,
-	seed: SEED,
-	datasetStart: DATASET_START,
-	datasetEnd: DATASET_END,
-	avgEventsPerUserPerDay: EVENTS_PER_DAY,
-	numUsers: NUM_USERS,
-	format: "json",
-	gzip: true,
-	credentials: {
-		token,
-	},
-	switches: {
-		hasSessionIds: true,
-		alsoInferFunnels: false,
-		hasLocation: true,
-		hasAndroidDevices: true,
-		hasIOSDevices: true,
-		hasDesktopDevices: true,
-		hasBrowser: false,
-		hasCampaigns: false,
-		isAnonymous: false,
-		hasAdSpend: false,
-		hasAvatar: true,
-	},
-	identity: {
-		avgDevicePerUser: 2,
-	},
-	concurrency: 1,
-	writeToDisk: false,
-	scdProps: {
-		seller_tier: {
-			values: ["new", "verified", "power", "featured"],
-			frequency: "month",
-			timing: "fuzzy",
-			max: 8
-		}
-	},
-	mirrorProps: {},
-	lookupTables: [],
-
-	events: [
-		{
-			event: "account created",
-			weight: 1,
-			isFirstEvent: true,
-			isAuthEvent: true,
-			properties: {
-				signup_source: ["organic", "google_shopping", "tiktok", "seller_referral", "email_campaign", "word_of_mouth"],
-			},
-		},
-		{
-			event: "item searched",
-			weight: 8,
-			isStrictEvent: false,
-			properties: {
-				search_query: ["laptop", "vintage jacket", "sneakers", "headphones", "phone case", "gaming console", "watch", "sunglasses", "backpack", "camera"],
-				results_count: u.weighNumRange(0, 50, 0.5),
-				sort_by: ["relevance", "relevance", "price_low", "price_high", "newest", "best_selling"],
-			},
-		},
-		{
-			event: "item viewed",
-			weight: 7,
-			isStrictEvent: false,
-			properties: {
-				listing_id: chance.pickone.bind(chance, listingIds),
-				item_price: u.weighNumRange(5, 500, 0.3, 45),
-				condition: ["new", "new", "like_new", "good", "fair"],
-				seller_rating: u.weighNumRange(1, 5, 0.7, 4),
-				photos_count: u.weighNumRange(1, 10, 0.5, 3),
-			},
-		},
-		{
-			event: "add to cart",
-			weight: 5,
-			properties: {
-				listing_id: chance.pickone.bind(chance, listingIds),
-				item_price: u.weighNumRange(5, 500, 0.3, 45),
-				quantity: u.weighNumRange(1, 5, 0.2),
-			},
-		},
-		{
-			event: "purchase completed",
-			weight: 3,
-			isStrictEvent: false,
-			properties: {
-				listing_id: chance.pickone.bind(chance, listingIds),
-				total_amount: u.weighNumRange(10, 500, 0.3, 60),
-				item_count: u.weighNumRange(1, 5, 0.3),
-				payment_method: ["credit_card", "credit_card", "paypal", "apple_pay", "debit"],
-				shipping_method: ["standard", "standard", "express", "pickup"],
-			},
-		},
-		{
-			event: "listing created",
-			weight: 4,
-			isStrictEvent: false,
-			properties: {
-				listing_id: chance.pickone.bind(chance, listingIds),
-				store_id: chance.pickone.bind(chance, storeIds),
-				asking_price: u.weighNumRange(5, 500, 0.3, 50),
-				condition: ["new", "new", "like_new", "good", "fair"],
-				listing_fee: u.weighNumRange(5, 30, 0.5, 15),
-			},
-		},
-		{
-			event: "listing updated",
-			weight: 3,
-			properties: {
-				listing_id: chance.pickone.bind(chance, listingIds),
-				field_updated: ["price", "price", "description", "photos", "condition", "shipping"],
-				price_change_pct: u.weighNumRange(-30, 30, 0.5),
-			},
-		},
-		{
-			event: "offer received",
-			weight: 3,
-			isStrictEvent: false,
-			properties: {
-				listing_id: chance.pickone.bind(chance, listingIds),
-				offer_amount: u.weighNumRange(5, 400, 0.3, 40),
-				offer_pct_of_asking: u.weighNumRange(50, 100, 0.5, 80),
-			},
-		},
-		{
-			event: "offer accepted",
-			weight: 2,
-			isStrictEvent: false,
-			properties: {
-				listing_id: chance.pickone.bind(chance, listingIds),
-				final_price: u.weighNumRange(10, 500, 0.3, 55),
-				negotiation_rounds: u.weighNumRange(1, 5, 0.3),
-				response_time_hours: u.weighNumRange(0.1, 48, 0.3, 4),
-			},
-		},
-		{
-			event: "review submitted",
-			weight: 2,
-			properties: {
-				listing_id: chance.pickone.bind(chance, listingIds),
-				rating: u.weighNumRange(1, 5, 0.7, 4),
-				review_length: u.weighNumRange(10, 300, 0.5, 80),
-				is_verified_purchase: [true, true, true, false],
-			},
-		},
-		{
-			event: "seller rated",
-			weight: 2,
-			properties: {
-				store_id: chance.pickone.bind(chance, storeIds),
-				seller_rating: u.weighNumRange(1, 5, 0.7, 4),
-				would_buy_again: [true, true, true, true, false],
-			},
-		},
-		{
-			event: "message sent",
-			weight: 5,
-			isStrictEvent: false,
-			properties: {
-				message_type: ["inquiry", "inquiry", "negotiation", "shipping_question", "complaint", "thank_you"],
-				recipient_type: ["seller", "seller", "buyer", "support"],
-				response_time_hours: u.weighNumRange(0.1, 48, 0.3, 6),
-			},
-		},
-		{
-			event: "shipping updated",
-			weight: 3,
-			properties: {
-				status: ["label_created", "in_transit", "in_transit", "out_for_delivery", "delivered", "delivered"],
-				carrier: ["usps", "ups", "fedex", "dhl"],
-				tracking_viewed: [true, true, false],
-			},
-		},
-		{
-			event: "refund requested",
-			weight: 1,
-			properties: {
-				reason: ["not_as_described", "damaged", "wrong_item", "changed_mind", "late_delivery"],
-				refund_amount: u.weighNumRange(5, 300, 0.4, 40),
-				resolution_status: ["pending", "approved", "approved", "denied"],
-			},
-		},
-		{
-			event: "notification received",
-			weight: 6,
-			properties: {
-				notification_type: ["price_drop", "price_drop", "new_offer", "shipping_update", "message_received", "sale_reminder"],
-				channel: ["push", "push", "email", "sms"],
-				opened: [true, true, true, false],
-			},
-		},
-		{
-			event: "profile updated",
-			weight: 2,
-			properties: {
-				field_updated: ["address", "payment_method", "profile_photo", "bio", "phone", "store_description"],
-			},
-		},
-		{
-			event: "app session",
-			weight: 8,
-			properties: {
-				session_duration_sec: u.weighNumRange(10, 1800, 0.4, 120),
-				pages_viewed: u.weighNumRange(1, 20, 0.5, 4),
-			},
-		},
-		{
-			event: "account deactivated",
-			weight: 1,
-			isChurnEvent: true,
-			returnLikelihood: 0.15,
-			isStrictEvent: true,
-			properties: {
-				reason: ["low_sales", "high_fees", "found_alternative", "bad_experience", "no_longer_needed"],
-			},
-		},
-	],
-
-	funnels: [
-		{
-			name: "Buyer Onboarding",
-			sequence: ["account created", "item searched", "item viewed", "add to cart"],
-			conversionRate: 40,
-			order: "sequential",
-			isFirstFunnel: true,
-			timeToConvert: 72,
-			weight: 3,
-		},
-		{
-			name: "Browse to Purchase",
-			sequence: ["item searched", "item viewed", "add to cart", "purchase completed"],
-			conversionRate: 30,
-			order: "sequential",
-			timeToConvert: 48,
-			weight: 5,
-		},
-		{
-			name: "Seller Listing Flow",
-			sequence: ["listing created", "listing updated", "offer received", "offer accepted"],
-			conversionRate: 25,
-			order: "sequential",
-			timeToConvert: 168,
-			weight: 3,
-		},
-		{
-			name: "Offer Negotiation",
-			sequence: ["offer received", "message sent", "offer accepted"],
-			conversionRate: 35,
-			order: "sequential",
-			timeToConvert: 72,
-			weight: 2,
-		},
-		{
-			name: "Review After Purchase",
-			sequence: ["purchase completed", "shipping updated", "review submitted"],
-			conversionRate: 20,
-			order: "sequential",
-			timeToConvert: 336,
-			weight: 2,
-		},
-	],
-
-	superProps: {
-		Platform: ["ios", "android", "web"],
-		category: ["electronics", "clothing", "home_garden", "collectibles", "sports", "toys", "automotive"],
-	},
-
-	userProps: {
-		user_type: ["buyer"],
-		segment: ["window_shopper"],
-		seller_rating: u.weighNumRange(0, 5, 0.5),
-		total_transactions: [0],
-		response_time_hours: u.weighNumRange(0, 48, 0.3),
-		store_name: ["none"],
-		Platform: ["ios", "android", "web"],
-		category: ["electronics", "clothing", "home_garden", "collectibles", "sports", "toys", "automotive"],
-	},
-
-	personas: [
-		{
-			name: "power_seller",
-			weight: 8,
-			eventMultiplier: 5.0,
-			conversionModifier: 1.8,
-			churnRate: 0.01,
-			properties: {
-				user_type: "seller",
-				segment: "power_seller",
-			},
-		},
-		{
-			name: "casual_seller",
-			weight: 20,
-			eventMultiplier: 1.5,
-			conversionModifier: 1.0,
-			churnRate: 0.08,
-			properties: {
-				user_type: "seller",
-				segment: "casual_seller",
-			},
-		},
-		{
-			name: "frequent_buyer",
-			weight: 25,
-			eventMultiplier: 2.0,
-			conversionModifier: 1.3,
-			churnRate: 0.03,
-			properties: {
-				user_type: "buyer",
-				segment: "frequent_buyer",
-			},
-		},
-		{
-			name: "window_shopper",
-			weight: 35,
-			eventMultiplier: 0.5,
-			conversionModifier: 0.4,
-			churnRate: 0.15,
-			properties: {
-				user_type: "buyer",
-				segment: "window_shopper",
-			},
-		},
-		{
-			name: "new_seller",
-			weight: 12,
-			eventMultiplier: 0.8,
-			conversionModifier: 0.5,
-			churnRate: 0.3,
-			properties: {
-				user_type: "seller",
-				segment: "new_seller",
-			},
-			activeWindow: { maxDays: 28 },
-		},
-	],
-
-	worldEvents: [
-		{
-			name: "marketplace_fee_change",
-			type: "product_launch",
-			startDay: 45,
-			duration: null,
-			injectProps: { fee_change: "increased" },
-			affectsEvents: ["listing created"],
-		},
-		{
-			name: "viral_tiktok_moment",
-			type: "campaign",
-			startDay: 55,
-			duration: 3,
-			volumeMultiplier: 3.0,
-			affectsEvents: ["item searched", "item viewed", "purchase completed"],
-		},
-	],
-
-	engagementDecay: {
-		model: "exponential",
-		halfLife: 75,
-		floor: 0.1,
-	},
-
-	attribution: {
-		model: "last_touch",
-		window: 7,
-		campaigns: [
-			{
-				name: "google_shopping",
-				source: "google",
-				medium: "cpc",
-				activeDays: [0, 100],
-				dailyBudget: [200, 800],
-				acquisitionRate: 0.03,
-				userPersonaBias: { frequent_buyer: 0.5 },
-			},
-			{
-				name: "tiktok",
-				source: "tiktok",
-				medium: "social",
-				activeDays: [10, 90],
-				dailyBudget: [100, 400],
-				acquisitionRate: 0.02,
-				userPersonaBias: { window_shopper: 0.5 },
-			},
-			{
-				name: "seller_referral",
-				source: "referral",
-				medium: "referral",
-				activeDays: [0, 100],
-				dailyBudget: [50, 200],
-				acquisitionRate: 0.02,
-				userPersonaBias: { power_seller: 0.4 },
-			},
-		],
-		organicRate: 0.30,
-	},
-
-	geo: {
-		sticky: true,
-		regions: [
-			{
-				name: "us",
-				countries: ["US"],
-				weight: 50,
-				timezoneOffset: -5,
-				properties: { currency: "USD", locale: "en-US" },
-			},
-			{
-				name: "uk",
-				countries: ["GB"],
-				weight: 20,
-				timezoneOffset: 0,
-				properties: { currency: "GBP", locale: "en-GB" },
-			},
-			{
-				name: "australia",
-				countries: ["AU"],
-				weight: 15,
-				timezoneOffset: 10,
-				properties: { currency: "AUD", locale: "en-AU" },
-			},
-			{
-				name: "canada",
-				countries: ["CA"],
-				weight: 15,
-				timezoneOffset: -5,
-				properties: { currency: "CAD", locale: "en-CA" },
-			},
-		],
-	},
-
-	anomalies: [
-		{
-			type: "extreme_value",
-			event: "purchase completed",
-			property: "total_amount",
-			frequency: 0.002,
-			multiplier: 30,
-			tag: "whale_purchase",
-		},
-		{
-			type: "coordinated",
-			event: "account created",
-			day: 68,
-			window: 0.02,
-			count: 200,
-			tag: "viral_signup",
-		},
-	],
-
-	hook(record, type, meta) {
-		if (type === "user") return handleUserHooks(record);
-		if (type === "funnel-post") return handleFunnelPostHooks(record, meta);
-		if (type === "everything") return handleEverythingHooks(record, meta);
-		return record;
-	},
-};
-
-export default config;