{
  "schemaVersion": "deck-intelligence.v1",
  "id": "019dd923-622c-750b-8b97-e6ccc43e808f",
  "slug": "9deb5f96d1a5",
  "title": "GSX Techedu (GSX)",
  "subtitle": "Andrew Left · 2020-04",
  "author": "Andrew Left",
  "date": "2020-04-14 00:00:00",
  "pageCount": 0,
  "kind": "research-note",
  "orientation": "portrait",
  "aspectRatio": 0.77272725,
  "targetCompany": "GSX Techedu",
  "targetTicker": "GSX",
  "document": {
    "id": "019dd923-622c-750b-8b97-e6ccc43e808f",
    "slug": "9deb5f96d1a5",
    "title": "GSX Techedu (GSX)",
    "rawTitle": "GSX Conclusive evidence final",
    "authorId": "52_Citron_Research",
    "authorName": "Andrew Left",
    "authorDisplay": "Andrew Left",
    "kind": {
      "slug": "research-note",
      "label": "Research note"
    },
    "sourceType": {
      "slug": "short_seller",
      "label": "Short seller"
    },
    "fileName": "2020-04-GSX-Conclusive-evidence-final.pdf",
    "sourceUrl": "https://cdn.gmtresearch.com/public-ckfinder/Short-sellers/Citron%20Research/Citron_GSX-Conclusive-evidence_Apr30-2020.pdf",
    "sourceDate": null,
    "presentationDate": "2020-04-14 00:00:00",
    "ingestedAt": "2026-04-22 20:20:12+00",
    "status": "ready",
    "pageCount": 0,
    "orientation": "portrait",
    "aspectRatio": 0.77272725,
    "targetCompany": "GSX Techedu",
    "targetTicker": "GSX",
    "metadata": {
      "activist_firm": "Citron Research",
      "origin_corpus": "contrarian",
      "redownloaded_at": "2026-04-29T12:45:48.743194+00:00",
      "extraction_seconds": 176.45884585380554
    }
  },
  "coverage": {
    "pageCount": 0,
    "pagesIndexed": 0,
    "pagesWithImages": 0,
    "pagesWithMetadata": 0,
    "imageCoveragePct": 0,
    "metadataCoveragePct": 0,
    "componentCount": 0,
    "chartCount": 0,
    "metricSlides": 0,
    "toolMatches": 2,
    "documentToolMatches": 2,
    "frameworkMatches": 0,
    "arcMatches": 0,
    "beatMatches": 0,
    "loopMatches": 0,
    "patternMatches": 0,
    "storymakersMatch": false,
    "scoreAvailable": false,
    "reviewAvailable": false,
    "activistThesisAvailable": true,
    "pitchdeckMetadataAvailable": false,
    "pitchdeckProfileAvailable": false
  },
  "blocks": [],
  "matches": {
    "arcs": [],
    "beats": [],
    "loops": [],
    "tools": [
      {
        "tool": {
          "name": "CEO quote contradiction",
          "slug": "ceo-quote-contradiction",
          "status": "active",
          "bestFor": "When management credibility is the lever",
          "canonId": "019dd9e1-52e9-717f-8454-93e627563ed6",
          "version": 1,
          "bodyDocId": "019dd9a1-8296-720d-8f51-da76937f9415",
          "description": "You reproduce management's own words — verbatim, with source and date — and\nthen place them next to the outcome that contradicts them. The CEO built\nthe rope; you put it around their own neck.",
          "familyLabel": null,
          "categoryName": null,
          "categorySlug": null
        },
        "agent": null,
        "layer": "document",
        "agents": null,
        "matchId": "ee13540b-7ff9-46f8-b4a2-8a943d49a1e8",
        "evidence": null,
        "pageRefs": null,
        "priority": null,
        "whenToUse": "When management credibility is the lever",
        "confidence": null,
        "extraction": {
          "at": "2026-04-29 12:07:50.005477+00",
          "model": "unknown-legacy",
          "runId": null,
          "seconds": null,
          "promptVersion": null
        },
        "pageNumber": null,
        "whyItWorks": "- **Maximum credibility**: no adjective you could write is as damning as the\n  target's own prior statement.\n- **No rebuttal path**: management can't argue they didn't say it.\n- **It forces accountability onto a person**, not an institution — even if\n  you aren't explicitly naming a villain elsewhere.\n- **Reporters amplify it**. A CEO-quote-contradiction slide is quotable,\n  which means it lands in WSJ/FT/Bloomberg coverage.",
        "antipattern": null,
        "cardinality": null,
        "narrativePurpose": null
      },
      {
        "tool": {
          "name": "Villain naming",
          "slug": "villain-naming",
          "status": "active",
          "bestFor": "When governance or specific decisions are the core thesis",
          "canonId": "019dd9e1-55a4-74db-9314-0ced4cd46cd2",
          "version": 1,
          "bodyDocId": "019dd9a1-88f1-72ad-8f4e-56ec29d7b369",
          "description": "You identify a **specific person** responsible for the target's underperformance\nor the behaviour you want changed — and name them. Not \"management\" or \"the\nboard\". A named executive or director with a specific failure attributable to\nthem.",
          "familyLabel": null,
          "categoryName": null,
          "categorySlug": null
        },
        "agent": null,
        "layer": "document",
        "agents": null,
        "matchId": "24e7f42f-1c8b-4d67-8a28-9ef3be620920",
        "evidence": "Larry Xiangdong Chen (GSX Founder & CEO)",
        "pageRefs": null,
        "priority": null,
        "whenToUse": "When governance or specific decisions are the core thesis",
        "confidence": null,
        "extraction": {
          "at": "2026-04-29 12:07:50.005477+00",
          "model": "unknown-legacy",
          "runId": null,
          "seconds": null,
          "promptVersion": null
        },
        "pageNumber": null,
        "whyItWorks": "- **Accountability needs a face.** Voters (shareholders, regulators,\n  journalists) act faster against people than institutions.\n- **Credibility rises with specificity.** \"The CFO has repeatedly missed\n  guidance\" > \"Management has missed guidance\".\n- It **pre-empts evasion**. An individual can be removed; an abstraction\n  cannot.\n- It **forces the target to respond personally** — and their defensive\n  response usually generates more attention for your thesis.",
        "antipattern": null,
        "cardinality": null,
        "narrativePurpose": null
      }
    ],
    "frameworks": [],
    "patterns": []
  },
  "storymakers": null,
  "score": null,
  "review": null,
  "activistThesis": {
    "tone": "adversarial",
    "notes": "SOURCE FILE CORRUPTED/TRUNCATED: the on-disk PDF is exactly 1,048,576 bytes (1MB) but the linearization header declares /L 6125125 (~6MB original) with N/29 pages, so the file is truncated and unreadable — pdftotext, pdfinfo, ghostscript, and pypdf all fail with 'Couldn't find trailer dictionary' / 'Invalid XRef entry'. No text could be extracted. All content fields are inferred from the filename ('2020-04-GSX-Conclusive-evidence-final.pdf'), the firm folder (Citron Research), and external knowledge of Citron's well-documented April 2020 'GSX: Conclusive Evidence' short report on GSX Techedu (NYSE: GSX, now Gaotu/GOTU), which followed Citron's February 2020 initial report and parallel work by Grizzly and Muddy Waters. presentation_date 2020-04-14 reflects the historically documented release date of this report but was NOT verified from the PDF itself. page_count left at pre-filled 0 because the file cannot be opened; the linearization header suggests ~29 pages. Visual / KPI / notable-pages fields are best-guess defaults for a typical Citron short report (dense text-heavy research-note format, screenshots of evidence, Andrew Left byline) and should be re-extracted once a clean copy of the PDF is available.",
    "kpisCited": [
      {
        "detail": "Citron alleges a large majority of registered 'students' are bots or non-paying ghost accounts",
        "metric": "Share of bot / fabricated student accounts"
      },
      {
        "detail": "Claimed revenue growth materially exceeds independently observable Chinese app-usage and tax-filing signals",
        "metric": "Reported revenue vs. MIIT/tax data"
      }
    ],
    "closingAsk": "Sell / short GSX Techedu shares; Citron views the equity as uninvestable pending regulatory action.",
    "extraction": {
      "at": "2026-04-22 20:20:12+00",
      "model": "unknown-legacy",
      "runId": null,
      "seconds": 176.45885,
      "promptVersion": null
    },
    "scqaReward": "Citron implies fair value is near zero — the stock is presented as a total fraud with no defensible earnings, so downside to a short position is effectively -100%.",
    "thesisTypes": [
      "fraud_exposure"
    ],
    "top3Reasons": [
      "GSX fabricates student enrollment — registrations dominated by bots and ghost accounts",
      "Reported revenue growth inconsistent with China's MIIT app-usage and tax filings",
      "Insider actions and related-party transactions point to orchestrated stock promotion"
    ],
    "scqaSituation": "GSX Techedu is a US-listed Chinese online K-12 after-school tutoring platform whose reported paying-student and revenue growth has made it one of the best-performing China ADRs of 2019-2020.",
    "thesisSummary": "Citron Research's April 2020 'Conclusive Evidence' report on GSX Techedu (NYSE: GSX), a Chinese online K-12 tutoring company, argues that the company has fabricated a majority of its reported paying students and revenue. Following an earlier February 2020 Citron report and parallel work by Grizzly and Muddy Waters, this follow-up claims to present 'conclusive' evidence: bot-driven enrollment counts, divergence between reported user activity and independent Chinese regulatory/app-tracking data, suspicious related-party transactions, and patterns consistent with pump-and-dump stock promotion. Citron positions GSX as the most blatant US-listed Chinese fraud since the 2011 wave and argues the equity is ultimately worthless, urging investors to short the stock ahead of what it frames as inevitable regulatory action and collapse.",
    "primaryDemands": [],
    "scqaResolution": "Short GSX ahead of regulatory action; Citron calls on the SEC, auditors, and index providers to scrutinize the filings and delist what it frames as a fraudulent issuer.",
    "thesisOneLiner": "Citron argues GSX Techedu is a near-total fraud — the vast majority of claimed K-12 students are bots, revenues are fabricated, and the stock is a zero.",
    "precedentsCited": [
      "2011-era US-listed Chinese RTO fraud wave",
      "Luckin Coffee fraud disclosure (April 2020)"
    ],
    "scqaComplication": "Citron argues the growth is manufactured: bots inflate class rosters, reported revenue diverges from Chinese app-usage and tax data, and insiders and related parties orchestrate the promotion.",
    "narrativeInterest": 4,
    "notableSlidePages": [
      1
    ],
    "peerCompaniesCited": [
      "TAL Education",
      "New Oriental Education",
      "Luckin Coffee"
    ],
    "valuationFrameworks": [
      "not_shown"
    ],
    "visualCraftInterest": 2
  },
  "pitchdeck": {
    "metadata": null,
    "profile": null
  },
  "slides": [],
  "arcBeats": [],
  "loops": []
}