RDDT (Reddit, Inc.) — Framework and Thesis #

Status: OPEN, 5 sh @ USD 156.88 (2026-08-27, EUR 676.82). Bought without a thesis; this document is the thesis written after the fact on 2026-09-02 at USD 157.25. Data: stock-data-analyst pass 2026-09-02 (SEC 8-K/10-Q, yfinance, Finnhub), fact-check pass same day. Plan: RDDT_plan.json. Execution log: RDDT_execution.md.

1. How to judge this business #

Reddit is an advertising platform with a small data-licensing line. A platform is three multiplications, and each one has a number to watch:

Then the repo's two questions: is it good (rule-003 checklist) and is it cheap (rule-004: Graham sensitivity table, implied growth, SBC-adjusted P/FCF, scenario value, Kelly go/no-go).

2. What SBC is, and what it does to Reddit's numbers #

Stock-based compensation: paying employees with shares or options instead of cash. It costs the owner in two ways. New shares are created, so each existing share owns less of the company. And because no cash leaves the building, "free cash flow" and "adjusted EBITDA" look bigger than what the owner actually keeps. GAAP net income does subtract it, which is why the repo prefers GAAP earnings and FCF-minus-SBC over adjusted figures.

The trend is the good news: SBC was 21.5% of revenue in Q3 2024 and is ~12% now; basic share count grew 2.8% in the last year and a USD 1B buyback (USD 240M executed) now offsets new issuance. The rule-003 example ("P/FCF doubled from 42x to 84x") was early-2026 data at a lower revenue base; the doubling still happens, from 30x to 44x. Check SBC as a share of revenue every quarter.

3. Good? (rule-003) #

A genuinely profitable, cash-rich business with one structural dependence it does not control, and an insider pattern that fails the owner's own rule.

4. Cheap? (rule-004) #

Price USD 157.25. Diluted shares 202M. Market cap ~USD 30B.

The gap between trailing (14%) and forward (4–7%) implied growth is the rule-004 signal that the market is pricing an inflection: earnings roughly doubling in FY2026. Consensus has revenue +54% in FY2026 and +31% in FY2027.

Graham sensitivity, V = EPS × (8.5 + 2g):

Break-even growth at 157: 10.4% on FY26E, 6.9% on FY27E, against ~30% expected. On the formula the margin of safety is real, but it is the difference between what Google keeps sending and what the market assumes; it is not a cushion of assets or of a low multiple (10.9x sales, 44x SBC-adjusted FCF; peers: Pinterest 46x, Meta 95x on the same measure).

Scenarios on FY2027 GAAP EPS (assumptions, stated so they can be argued with):

Verdict: fair, not cheap. Cheap was USD 119 in March, where the director bought. At 157 the price is the base case; Kelly as a go/no-go (rule-006) says no edge, so no adds here. Holding is a judgement about the Google dependence, not about price.

5. The dip that was bought #

All-time high close 270.71 (2025-09-18) → 119.27 low (2026-03-30) → 2026 peak close 203.27 (2026-07-14) → 156.88 on 2026-08-27, which is −23% from the July peak and −42% from the 2025 high. The sequence, with dates: Jul 22 −8% on a WSJ report that Reddit may restrict Google's AI access as the ~USD 60M/yr licensing deal nears expiry; Jul 30 Q2 print beat on revenue, EPS and guide, but US DAUq slipped quarter-on-quarter and the letter called search referrals "choppy"; Jul 31 −21% on 30M shares; Aug 14 +13% on S&P 500 inclusion, then a sell-the-news fade; Aug 17–20 third-party data showing Reddit's share of ChatGPT search citations collapsing after an OpenAI change; Aug 31 EU "very large online platform" designation. Not macro: the index was up that week.

Two things to know about that fill. It sits almost exactly at the company's own buyback average (157.27 for the USD 240M repurchased in the first half). And a −23% drawdown in a name still at 11x sales is a lower price, not a margin of safety (rule-001, rule-004). The reddit threads recommending Reddit stock were the attention signal rule-001 warns about; none of them contained the SBC, logged-out-user or licensing-backlog numbers in this document.

6. Kill conditions (fundamental only, rule-008) #

1. Global DAUq down quarter-on-quarter for two consecutive quarters, or US DAUq down year-on-year, or total DAUq growth below 10% year-on-year.
2. Management attributes a revenue or user miss to search-engine or AI-answer changes (logged-out users are no longer reported from Q3 2026, so the commentary is the metric), or guidance is cut.
3. SBC above 15% of revenue for two quarters, or fully diluted share count up more than 3% year-on-year (the buyback must keep offsetting issuance).
4. Insiders: any discretionary (non-10b5-1) sale by the CEO, COO or CFO; the CFO resuming sales; or a further enlargement of the CEO or COO selling plans beyond the May 2026 sizes. Owner's rule.
5. Advertising revenue growth below 25% year-on-year for two consecutive quarters (Q3 guide already implies ~47–49% total growth, decelerating from 61%).
6. Google licensing deal ends without a replacement, or "other" revenue falls year-on-year. The licensing backlog (RPO) is USD 92M at June 2026 and runs off by end-2027 unless renewed; substantially all of it comes from two customers.

Any one fires: re-score within five trading days; exit unless the thesis survives on the new numbers.

7. Take-profit ladder (rule-008, mechanical: no margin of safety cushion) #

8. Sizing and monitoring #

• EUR 677 = 21% of the single-name book: inside the rule-006 band. No adds above 133.
• Next print ~2026-10-29: US DAUq, ad growth vs the 860–870M guide, SBC %, buyback pace, licensing RPO, any Google or OpenAI renewal (a renewal well above the ~USD 140M/yr run-rate is the bull-case swing; sell-side talk of USD 550M/yr combined is speculation, not guidance).
• Goldman Sachs Communacopia 2026-09-08 (management commentary on Google and licensing).
• Form 4: 10b5-1 sales by the CEO and COO continue on schedule and are expected; anything outside the plans, or the CFO resuming, is kill condition 4.
• Litigation: Perplexity/SerpApi scraping case survived the motion to dismiss (Jul 31, 2026), which supports the licensing line; securities class action pending; EU DSA compliance due ~Dec 2026.