---
ticker: "RDDT"
company: "Reddit, Inc."
date: "2026-10-26"
view: "Strong Buy"
fair_value: 311.31
price: 138.47
price_source: "last trade 15:59:59 ET 2026-10-26 (regular session; official close 2026-10-23 $144.25)"
price_time: "2026-10-26T19:59:59Z"
price_vs_fair_value: "-55.5%"
bear: 115.04
bull: 722.72
exit_price: 413.29
p_price_justified: "over 50%"
restated: "2026-10-26"
method: "case-spread-2026-10-24"
original_fair_value: 311
original_view: "Buy"
page: "https://alpha.dming.si/research/2026-10-26-rddt"
---

# RDDT · Reddit, Inc. · Strong Buy

Research report as of 2026-10-26, partly unblinded. Page: https://alpha.dming.si/research/2026-10-26-rddt

We value RDDT at $311 a share. The report's $138.47 price (Oct 26, 2026) was 56% below fair value: Strong Buy.

_Restated Oct 26, 2026 (method case-spread-2026-10-24); the report's original value was $311, Buy._

| Lens | Range | Marks |
| --- | --- | --- |
| Our report | $115 to $723 | fair $311; original $311 |
| Our simulation | $111 to $687 | median $287 |
| Street targets (15) | $110 to $265 | average $203, median $200 |
| Price | | $138.47 |

Bands run from our bear to bull (the hollow mark is the original $311), our simulation's 10th to 90th percentile and the Street's low to high. Street targets forecast future prices, not value today. Market data, captured Oct 26, 2026.

| Figure | Value |
| --- | --- |
| Fair value | $311 |
| Price | $138.47 (market data, Oct 26, 12:59 PM PT) |
| Five-year return | 31.7% (a year, at the report's price) |
| Bear / bull | $115 / $723 |
| Exit price | $413 (on Dec 31, 2031) |
| Confidence | 22% |
| Price against our scenarios | The price is justified in over 50% of our scenarios (4,000 simulated outcomes, each drawing the uncertain drivers from ranges we chose: our judgement, not a measured chance) |

_This text argues the original call. The current figures are at the top._

**Original call, Oct 26, 2026 (before the Oct 26, 2026 restatement)**

```
RDDT — Reddit, Inc. — research view (market: Buy)
As of 2026-10-26 · partly unblinded: before the model was committed (04d7e26) the analyst
saw a sell-side 2026 free-cash-flow figure quoted at a September conference, and
directions of the share price (down year to date, down sharply after Q1, an 11% fall on
2026-07-30) in transcripts and headlines. No quote, multiple, target or rating was read.

Fair value today      $311 per diluted share, discounting at a 9.47% cost of equity (blind
                      $339; two review findings, -$28). At Reddit's own beta (1.70
                      Blume-adjusted) it is $239. Weighting the bear, central and bull
                      cases 25/50/25 (our weights, not calibrated) gives $365.
Price                 $138.47, last regular-session trade 2026-10-26 15:59:59 ET
                      (official close 2026-10-23 $144.25)  →  125% upside to fair value
Five-year exit price  $413 a share on 2031-12-31: the mean of a long fade of Reddit's
                      returns toward its cost of capital ($386) and what peers have been
                      valued at, 24.9 times operating profit ($441). Fading over 10 or 20
                      years instead gives $384 or $444.
Five-year return      31.7% a year from $138.47 to the central exit, payouts included, against
                      the 9.47% cost of equity (bear 4.6%, bull 57.1%; market.txt)
Bear / bull today     $115 / $723. The bear is the world where AI answers in search keep
                      the click; the bull is Reddit becoming a daily destination.
Confidence            22%: the share of our 4,000 simulated fair values within ±20% of the
                      central $311 ($249-373; confidence.txt); subjective, not calibrated
Moat                  21 years of human conversation, organised into communities that people
                      and AI models both seek out by name; no one else can rebuild it.
Market expects        About our bear case's business: revenue near $5.5-6.1 billion in 2031
                      (10-12% a year from 2026) at a 29-31% GAAP operating margin. Our bear's
                      operations at our default exit multiple are worth $134 (market.txt)
Price needs           No single headline driver reaches the price within its range (the
                      nearest: costs of 0.70 per revenue dollar added gives $174, U.S. revenue
                      per user 15 points a year below ours $184); jointly, every driver 80% of
                      the way from our central to our bear, exit at 18.7x (odds.json, market.txt)
Odds in our scenarios P(price justified): 82%, in the over-50% band (subjective, not
                      calibrated); simulated median $287, 10th to 90th $111-687. The blind
                      odds said 92%; the simulated bear world was milder than our bear case
                      (an auditor finding, now fixed)
Key driver            U.S. advertising revenue per user: 5 points a year less growth is -$56
                      a share; the bear's pricing path alone takes $58 off the central.
Street                Average 12-month target $202.52 (median $200, range $110-265); none set
                      in the last month, 15 over the quarter average $198.53, 51 over the year
                      $222.06: falling (market data via research-check, 2026-10-26). An expectations
                      reference, not validation
```

**Fair value $311 a share.** Reddit sells advertising against
conversations that people trust because other people wrote them, and it is
converting that into money very fast: revenue grew more than 60% for eight
straight quarters, to $805 million in Q2 2026, at a 29% GAAP operating
margin, with almost no capital spending. Our central case slows that growth
to 36% in 2027 and 13% in 2031, with revenue of $9.2 billion and a 39%
operating margin in 2031, and values it at the method's default exit.

**The answer on AI search.** On the evidence through Q2 2026, Google's AI
Overviews and other AI answers are taking Reddit's search visitors, not
sending more: U.S. logged-out daily users stopped growing and fell late in
the quarter, and machine-translated pages lost search rank abroad. Because
a visitor from search sees roughly a third of the ads a logged-in user
sees, that loss by itself is worth about $4 a share per point of yearly
decline; the risk that matters is the world where the same AI answers also
starve the funnel to logged-in users, slow ad pricing and end the Google
licensing deal, which is our bear case at $115.

Terms used throughout: the **central case** is our most likely outcome; the
**bear** and **bull** cases are coherent worlds below and above it, not the
central plus or minus a percentage. **DAUq** is Reddit's daily active
uniques, users it can identify on a given day; **logged-in** users have an
account and are signed in, **logged-out** users are not and mostly arrive
from search engines. The **long fade** extends our forecast 15 years past
2031, with growth and returns on new capital fading toward the economy's,
and values it at 2031. The **outside view** values 2031 operating profit at
the median multiple peers traded at over the last ten year-ends. The
**exit price** is the mean of the two.

## 1. What Reddit sells, who pays and what limits it

Reddit is about 100,000 active communities (subreddits), each run by
volunteer moderators who set its rules, holding 26 billion posts and
comments. Users come to ask, compare and argue: the CEO says about 40% of
conversations are commercial, people deciding what to buy, watch or where
to go (management claim, Q1 2026 call). Revenue is 95% advertising
(advertising $761.6M of $804.9M in Q2 2026, 10-Q note 3); the rest, "other
revenue", is mostly licensing the corpus to AI companies.

**Who pays.** Advertisers, buying through an auction on the home feed and
conversation pages. The ten largest were 21% of 2025 revenue (25% in 2024,
10-K). Active advertisers grew more than 70% in the year to Q2 2026, the
mid-market and small-business channel doubled, 11 of 15 verticals grew over
50%, and more than half of revenue comes from credit-line relationships
(COO, Q2 2026 call; management claims). Advertisers pay for both reach and
intent: conversion volume grew over 100% and click volume over 30% in Q2
2026 (management claim).

**Users.** 130.3 million daily uniques in Q2 2026 (U.S. 53.2M,
international 77.1M) and 514.6 million weekly. They split very unevenly:

| Q2 | 2024 | 2025 | 2026 | 2026 vs 2025 |
| --- | ---: | ---: | ---: | ---: |
| U.S. logged-in (M) | 20.4 | 22.9 | 23.1 | +1% |
| U.S. logged-out (M) | 25.1 | 27.4 | 30.1 | +10% |
| International logged-in (M) | 21.6 | 26.4 | 29.5 | +12% |
| International logged-out (M) | 24.1 | 33.7 | 47.6 | +41% |
| U.S. revenue per daily user, quarter ($) | 4.94 | 7.87 | 11.85 | +51% |
| International revenue per daily user ($) | 1.24 | 1.73 | 2.26 | +31% |

Source: earnings releases (reported). Q2 2026 is the last quarter Reddit
reports the logged-in and logged-out split (CFO, Q2 2026 call).

Two things stand out. Nearly all the growth in users for two years has
been logged-out, from search engines and, abroad, from machine-translated
pages; U.S. logged-in users have barely moved. And money per user has
compounded at 30-50% a year, which is where the growth in revenue came
from: ad revenue grew on both more impressions and higher prices (Q2 2026
letter).

**What limits it.** First, traffic it does not control: Reddit relies "in
part" on search engines, mostly Google, and says a change in algorithms,
interface or "AI-generated search or answer features" could cut traffic
(10-Q risk factors). Second, users who come once: the CEO describes a
histogram of days per week whose two tallest bars are one day and seven
days (Q1 2026 call), and new-user retention is "lower than others in the
space" (CFO, Q2 2026 call). Third, its own product: the feed is "a year to
18 months behind the state of the art" (CEO, Goldman Sachs conference,
September 2026), and ad load is "quite low compared to peers", with no ads
yet in search (COO, Q1 2026 call). The second and third are also the
upside.

**Ad load and pricing against Meta and Google.** Reddit does not disclose
ad load, impressions or price per ad, so the comparison is per user. In Q2
2026 Reddit made $11.85 per U.S. daily user and $6.18 worldwide; Meta made
$16.89 per daily person worldwide ($60.80 billion over 3.60 billion), with
impressions up 14% and price per ad up 12% (Meta release, 2026-07-29);
Pinterest made $8.30 per monthly U.S. and Canada user (Pinterest release,
2026-08-04). Google gives no comparable figure. So a Reddit user in the
richest ad market still yields less than Meta's average person worldwide.
That gap is the room the central case uses: U.S. revenue per
logged-in-equivalent user rises about 2.3 times from 2026 to 2031.

**Data licensing.** Other revenue was $140.7M in 2025 and $82.0M in H1 2026
(+15% and +24% in Q1 and Q2). The 2024 prospectus listed $203.0M of
aggregate licensing contract value on two-to-three-year terms (secondhand,
TechCrunch 2024-02-22); Google pays about $60M a year (press, CNBC
2026-07-22 relaying the Wall Street Journal) and OpenAI is the other large
buyer (CEO, July 2026, press). Remaining obligations on long-term licensing
contracts fell from $143.7M at 2025 year-end ($24.8M of it in 2027) to
$92.1M at 2026-06-30 ($30.1M in 2027) (10-K and 10-Q note 3), so the
contracts run out early in 2027. How durable it is: the content is not
replaceable and is "used for training, post-training, grounding, and as a
search index" (CEO, Q2 2026 call), and Reddit has sued Perplexity over
scraping (press, CNBC 2025-10-23); against that, the CEO would not rule out
not renewing Google and said "the range of outcomes is wide", and the Wall
Street Journal reported Reddit discussing cutting off Google's AI access
(press). We hold licensing at the 2026 level in the central (a renewal at
1.0x), half in the bear and double in the bull. It is a small part of the
value: renewal at 0.3x costs $5 a share and 2x adds $7, because licensing
is about 5% of revenue.

**Stock pay and dilution.** Stock pay and related taxes were 17.7% of
revenue in 2025 ($389.4M) and 12.6% in H1 2026, and Q3 2026 is guided at
$140-155M as older grants are replaced (CFO). Fully diluted shares were
207.0M at 2026-06-30, up 0.2% in a year after buybacks; gross of buybacks
dilution was 1% in the quarter, and the target is 1-3% a year before
buybacks (CFO, Q2 2026 call). We charge stock pay as a cost (15% of revenue
in 2027 falling to 12% from 2030, after the review; section 6) and hold the count at 202.3M shares (shares
outstanding plus all 10.0M options; the 4.7M unvested restricted units are
left out because their cost is in the stock-pay line). Three points of
revenue more stock pay is worth -$22 a share.

## 2. Recent results beat the guide every quarter

Reddit gives one quarter of guidance, revenue and adjusted EBITDA (earnings
before interest, tax, depreciation and stock pay). In the eight quarters
from Q3 2024 to Q2 2026 revenue beat the guide's midpoint every time, by a
median 10.3% (range 7.5% to 19.0%), and adjusted EBITDA beat by a median
0.86 of each dollar of revenue beat (`output.txt`). Q2 2026: revenue $804.9M
against a $715-725M guide (+11.8%), adjusted EBITDA $342.8M against
$285-295M, GAAP operating income $231.7M, net income $252.8M. Operating
cash flow crossed $1 billion over the last twelve months; capital spending
was $1.1M in the quarter (Q2 2026 release).

The guide for Q3 2026 is revenue $860-870M (47-49% growth) and adjusted
EBITDA $385-395M. The CFO cut the full-year stock-pay guide from "high
teens" to "low to mid teens" of revenue and set a 50% long-term adjusted
EBITDA margin (management claim). The same call disclosed that search
referrals "were choppy" and fell late in the quarter, U.S. DAUq fell
300,000 from Q1, and machine-translated traffic declined (CEO).

## 3. The forecast and the value

The full model is `model.py`, its printout
`output.txt` and the inputs with sources
`assumptions.md`.

**How it is built.** Advertising revenue in each geography is revenue per
logged-in user times logged-in users plus 0.35 of logged-out users. The
0.35 is our assumption: the COO says an impression is worth the same for
both and logged-in users simply see more of them (Q1 2026 call); the CEO
says direct and app users are "worth multiples more" than search visitors
(Q2 2026 call). Costs before stock pay rise by 0.45 of each revenue dollar
added (0.43 in the year to H1 2026). 2026 is half reported, Q3 at the guide
plus the median beat and Q4 at +55%.

| Central | 2026 | 2027 | 2028 | 2029 | 2030 | 2031 |
| --- | ---: | ---: | ---: | ---: | ---: | ---: |
| Global DAUq (M, average) | 130.4 | 134.9 | 139.0 | 142.8 | 146.5 | 150.1 |
| Revenue ($M) | 3,547 | 4,810 | 5,938 | 7,025 | 8,106 | 9,186 |
| Growth | 61.0% | 35.6% | 23.4% | 18.3% | 15.4% | 13.3% |
| of which licensing ($M) | 170 | 170 | 184 | 199 | 215 | 232 |
| Adjusted EBITDA margin | 45.7% | 48.2% | 49.5% | 50.3% | 50.9% | 51.4% |
| Stock pay and taxes ($M) | 481 | 722 | 831 | 913 | 973 | 1,102 |
| GAAP operating margin | 31.7% | 32.7% | 35.0% | 36.8% | 38.4% | 38.9% |
| Owner free cash flow ($M) | 865 | 1,035 | 1,426 | 1,812 | 2,210 | 2,553 |

The table is after the review (section 6): U.S. logged-in growth cut to
+1%, +1.5% then +2%, and stock pay raised to 15% of revenue in 2027 falling
to 12% from 2030.

Where this differs: from guidance, we take the median beat for Q3 and do
not credit management's margin target beyond its own path (our 2031 margin,
51.4%, lands slightly above the 50% target only because costs follow
revenue at the recent rate). From history, growth falls from 61% to 36%
in 2027: the comparison doubles, and money per user cannot compound at 50%
for long from a base that is already 70% of Meta's worldwide average. From
peers, our 39% GAAP operating margin in 2031 is a large-platform margin,
which Reddit's 91% gross margin and almost no capital spending allow. It is
far above the 20.1% of 2025, which carried stock pay of 17.7% of revenue;
the 2031 margin is the adjusted EBITDA margin (51%, against 38% in 2025
and 42.6% in Q2 2026) less 12% stock pay.

**Value.** Cost of equity 9.47%: the 5.29% risk-free rate plus a beta of
0.995 (the median Blume-adjusted beta of META, GOOGL, PINS, SNAP, YELP, TRIP
and TTD) times Damodaran's 4.20% premium. Net cash and securities of $2.79
billion, $0.60 billion of tax losses and credits Reddit will use soon, and
owner cash since July are counted today ($3,596M, $17.77 a share); owner
free cash flow is paid out as earned. The exit price is $413: the long fade
gives $386 (25.0 times 2032 after-tax operating profit) and the outside view
$441 (24.9 times 2031 operating profit). They are 14% apart, so the forecast,
not the method, sets the value.

**The exit multiple: the method's default, checked against growth.**
Reddit has positive GAAP operating income in one year of the last ten, so
the method uses the peers' ten-year median alone: META 21.0x, GOOGL 25.2x,
YELP 30.1x and TRIP 24.7x (Pinterest, Snap and The Trade Desk have fewer
than five profitable years in the window), 24.9x. The question the
2026-10-26 ruling asks is whether that anchor was earned at a different
growth rate. It was not, materially: the four peers grew revenue at a median
14.4% in the years their multiples count (META 24.4%, GOOGL 16.7%, YELP
12.0%, TRIP 3.4%), and our central grows 14% in 2031 and 13% in 2032
(`exit_choices.txt`). So we make no change from the default.

The judgment range across defensible exit choices (central forecast fixed;
`output.txt`):

| Outside view, EV / 2031 operating profit | Fair value |
| --- | ---: |
| Fade alone, no outside view | $294 |
| 17.2x, the anchor at the 25th percentile of realised multiples | $269 |
| 22.9x, the anchor times software starts growing 10-16% five years on, 1996-2009 | $300 |
| 23.1x, Meta and Alphabet alone, the scaled ad platforms | $301 |
| **24.9x, anchor (d), the method's default** | **$311** |
| 35.1x, the anchor times the same software starts, all eras | $368 |
| 36.3x, the anchor at the 75th percentile | $374 |

**Does the default rest on two small peers?** No. Half the anchor is Yelp
and Tripadvisor, but dropping any one of the four peers moves the median
between 24.7x and 25.2x ($310-313), and Meta and Alphabet alone give 23.1x
($301) (`exit_choices.txt`). Yelp's own median is inflated by years of thin
profit (74.9x in 2019 and 75.5x in 2021); it does not move the median
because the median of four ignores the extremes.

Our judgment range is **$294-311**: from the fade alone to the default,
with Meta and Alphabet alone in between. The rows above it rest on a
software sample and an era (2010-2021) whose multiples ran 1.76 times
their anchor, which is a level of that decade, not a growth effect.

**What moves the value most** (`output.txt`, central $311):

| Change | Fair value | Effect |
| --- | ---: | ---: |
| Reddit's own beta, 1.70 Blume-adjusted (raw 2.05) | $239 | -$73 |
| U.S. revenue per user -5 points a year | $256 | -$56 |
| U.S. revenue per user +4 points a year | $370 | +$59 |
| Costs 0.55 of each revenue dollar added (not 0.45) | $256 | -$55 |
| Damodaran internet-software beta, 1.37 Blume-adjusted | $267 | -$44 |
| Logged-in users -3 points a year | $269 | -$42 |
| U.S. logged-out users -8 points a year | $284 | -$27 |
| Stock pay +3 points of revenue | $289 | -$22 |
| Licensing renewed at 0.3x | $306 | -$5 |
| Payouts kept as cash at 2.8% after tax to the exit | $307 | -$5 |

Each $1 billion of 2031 revenue from higher U.S. pricing is worth about $53
a share. **The beta is the largest single choice.** The method uses the
peer median (0.995) by rule, and the rule exists because own betas are
noisy: Reddit has traded only since March 2024, and its own beta of 2.05
raw (1.70 Blume-adjusted, a 12.45% cost of equity; the advocate's market data
capture, `review/evidence.txt`) is measured over a span that includes the
listing and a retail-driven stock. Damodaran's internet-software industry
beta, 1.55 unlevered (January 2026), sits between. At Reddit's own beta the
value is $239, at Damodaran's $267; both are well above the price. The two
betas move the value by more than 10%, so the choice is flagged for the
reviewers (flag 7) and we show all three.

**Changed at the market-pass stage (2026-10-26, after the blind commits).**
`alpha research odds` refused two driver ranges under its median rule (the
central must sit 29-71% of the way from low to high), and one alternative
world's licensing range. Central values did not change, and no fair value,
bear, bull or exit changed:

| Driver | Was | Now | Source |
| --- | --- | --- | --- |
| U.S. search visitors vs our path | -10 to +4 points | -10 to +5 points | Reddit's U.S. logged-out DAUq grew 9.2% and 9.9% in the two years to Q2 2026 (releases), 12-13 points above our 2027 path; +5 stays well below that pace |
| Exit multiple vs anchor (d) | 0.919-1.408 | 0.691-1.456 (17.2x-36.3x) | Pooled 25th and 75th percentiles of realised over anchor five years on (survivor study 2026-10-23, `method.CASE_SPREAD`), the spread the method already gives the bear and bull |
| Licensing in the bear-leaning world | 0.3-1.0x | 0.3-0.9x | The bear world has Google lapsing; 0.9 is everyone else renewing |

Running the odds to confirm the file printed the simulation's summary to
the analyst, including the chance of the price in each world: that is a
view of where the price sits relative to our scenarios. It came after both
blind commits and after these bounds were chosen; nothing was changed
because of it. The market pass writes `odds.json`.

## 4. The price pays for our bear case's business

This section and the next were written after the price, targets and
estimates were read (unblinded); nothing in sections 1-3 or the model
changed. The figures are in `market.py` and its printout
`market.txt`.

At $138.47 the market pays for about the business in our bear case. Our
bear's operations, revenue of $5.5 billion in 2031 at a 29% GAAP operating
margin, are worth $134 at our default exit multiple of 24.9 times operating
profit, and $115 at the bear's 17.2 times. No single driver gets our central
case down to the price within its range: the nearest are costs rising by
0.70 of each revenue dollar added ($174) and U.S. revenue per user growing
15 points a year slower than ours ($184). Moving every driver together, the
price needs them 80% of the way from our central to our bear: revenue of
$6.1 billion in 2031, 12% a year from 2026, at a 31% operating margin and
an exit at 18.7 times (figures after the review; `market.txt`).

**How the price got here.** The shares rose 13% on 2026-05-01, the day
after Q1 results, and fell 21% to $140.67 on 2026-07-31, the day after Q2,
on six times their average volume: the quarter in which Reddit said search
referrals fell late and U.S. daily users dropped 300,000 (daily
bars). They are down from $263.50 on 2026-01-09 and above the year's low of
$119.27 on 2026-03-30; over six months they fell 13.6% against 10.3% for
the communication-services fund XLC (research-check).

**The Street.** Analysts' revenue is close to ours for three years and
then lower: $3.4 billion, $4.5 billion and $5.6 billion for 2026-2028
against our $3.55 billion, $4.81 billion and $5.94 billion (we are 4-7%
higher, mostly because we add Reddit's median beat to each guide), then
$6.2 billion and $7.3 billion for 2029-2030 (22 and 14 analysts) against our
$7.03 billion and $8.11 billion (11-13% higher). The Street's revenue path
run through our model, by slowing revenue per user 3.6 points a year in both
regions and leaving everything else central, is worth $252. That one shift
matches only 2030; matching all five Street years and keeping our 2031-32
growth gives $299 on the blind model (the auditor, `review/codex-rddt.md`).
Its average 12-month target, $202.52, discounted a year at our cost of
equity is $185; our value is 1.68 times it. Of the blind gap to the Street
path ($77), $54 is U.S. and $23 international revenue per user (the
auditor); the rest of the gap to the targets is what the market pays for
the Street's path, well under our exit multiple. We have not modelled the
Street's EPS. The
targets are falling, from $222.06 on average over the year to $198.53 over
the quarter.

## 5. Where we differ from the market, and what settles it

The market is pricing the search bear as the likely world; we give it a
quarter of the weight. Section 6 answers what the review found when it
built the world in which $138 is right.

The market's case at its strongest: U.S. daily users fell in Q2 2026 and
Reddit then stopped reporting the logged-in split, so investors cannot tell
whether AI answers are taking only drive-by visitors or also the users who
pay; the Google licensing deal ends in early 2027 under open negotiation;
and a shareholder suit says Reddit misled investors about AI Overviews. Our
cases answer the first on dated evidence only in part: U.S. logged-in users
grew 1% in a year, which is the bear's funnel, not ours. Our central rests on
U.S. revenue per user still growing 36% in 2027, and that is an assumption
from the last two years' 51-54%, not a dated commitment.

**What settles it.** Q3 results on 2026-10-29: U.S. daily users holding 53
million and revenue near our $954 million keep the central; U.S. daily users
below 52 million is the first quarter of the bear's user path. Then U.S.
revenue per user in 2027: below 20% growth in any quarter moves us toward
the bear, and with it toward the price. We concede the central to the
market if U.S. daily users fall for two quarters and the Google renewal
lapses or changes Reddit's search placement.

**Alpha's view, 2026-10-26.** Alpha, the quantitative model, scores Reddit
90 of 100 (tier Strong), with expectations A, growth A, quality A, momentum
C and valuation D; it gives no rank today and flags its data as limited,
because growth comparisons span large changes in the business (`alpha
ticker analyze RDDT`). Its top drivers are the return on capital and the
growth in operating income per share; its top drags are slowing growth in
free cash flow and revenue and the price-to-sales ratio. We agree on the
returns and on the slowdown, which our central builds in (61% growth in
2026 to 36% in 2027). We differ on valuation: Alpha reads a high
price-to-sales ratio against its universe, we read a price that already
assumes our bear's business.

**The chart, as timing only.** A technical quick read on 2026-10-26 grades
the chart D: a declining trend, below the 21-day ($151), 50-day ($153) and
200-day ($162) averages, with the 200-day still falling (it was $170 a month
earlier). Today's $138.47 is below the $140.05 low set after Q2 results
(2026-08-03); the next support is the March low of $119.27. The shares fell
19% in the three months to 2026-10-23 against a 4% rise in QQQ. Results are
three sessions away. It improves to C on a close back above the 50-day
average. The grade says when, never what Reddit is worth, and it moved no
number here.

## 6. How each review finding changed the value

Two outside reviewers read the committed draft at the market pass (commit
0709cc5) with the price in hand: an auditor (Codex) who reran every script,
and an advocate (Claude Opus) who argued for Avoid. Their scripts and
outputs are in `review/`. The value moved from **$339 to $311**
(bear $120 to $115, bull $738 to $723). Both changes rest on Reddit's own
releases and guidance dated before today, not on the price; both lower the
value, toward the price, and we say so.

| # | Reviewer | Finding | Answer | Before → after |
| --- | --- | --- | --- | --- |
| 1 | Auditor | The simulated bear world (median $175) was milder than the $120 bear case: it lacked the bear's stock pay, costs and exit | **Fixed.** The bear-leaning world now draws the bear's cost ratio (median 0.55), +2.5 points of stock pay and the bear's exit (median 0.691 of the anchor, low end 0.515, the 10th percentile over the median in the survivor software rows). Its median is now $124 against the $115 bear | P 92% → 82%; no fair value change |
| 1 | Auditor | 2031 GAAP margins of 30.7%/41.0%/47.9% against 20.1% in 2025; cutting both revenue-per-user paths 5 points is -$85.62 | **Disclosed.** 2025's margin carried 17.7% stock pay; the gap is the adjusted EBITDA margin rising from 38% to 51% (management targets 50%) and stock pay falling to 12%. After the stock-pay fix the central 2031 margin is 38.9% | — |
| 2 | Auditor | The market pass's Street path matched only 2030; matching all five Street years gives $298.87; the premium over the Street path is U.S. (+$53.82) and international (+$22.83) revenue per user | **Disclosed** in section 4. The Street's EPS was not modelled | — |
| 3 | Auditor | The exit at 17.2x is -$46; Meta and Alphabet alone -$11 | **Disclosed**; already in the judgment range (now -$43 and -$10) | — |
| 4 | Auditor | 4.671M unvested restricted units omitted (-$7.66 gross); option proceeds omitted (+$2.01) | **Rejected as a net change, disclosed.** The units' remaining cost ($397.3M, 10-Q) is inside our stock-pay forecast for 2026-27, so adding the shares would charge them twice. The options are all counted with no proceeds, a $2 understatement we keep | — |
| 5 | Auditor | No loss-carryforward use schedule; licensing loss is only -$7.53 | **Disclosed.** Delaying the tax assets one or two years costs $0.26-0.49; total licensing loss is -$7.53 | — |
| 6 | Advocate | Reddit's own beta is 1.70 (Blume), a 12.45% cost of equity, -$81 | **Disclosed, not adopted.** The method uses the peer median by rule; own value printed beside it ($239 after the fixes, -$73; section 3). Flag 7 | — |
| 7 | Advocate | Stock pay is guided at $140-155M for Q3 2026, 15.5% of our Q3 revenue, against our 13% in 2027 falling to 10%; holding 15.5% is -$42 | **Fixed.** Dated evidence: the Q3 guide (2026-07-30) and the CFO's reason, older grants being replaced. Path now 15%, 14%, 13%, then 12%; bear 16% to 15%, bull 14% to 9% | -$15.47 alone |
| 8 | Advocate | U.S. logged-in users grew 0.4% over five quarters (Q1 2025-Q2 2026); flat is -$33 | **Fixed in part.** U.S. logged-in growth cut from +2% then +3% to +1%, +1.5%, then +2%: the five-quarter record, with some credit for the dated retention gain management reports | -$13.03 alone |
| 9 | Advocate | Exit at today's peer multiples (median 16.7x; Yelp 5.8x, Tripadvisor 12.8x) is -$49 to -$84 | **Disclosed as a sensitivity only.** Today's multiples are a market input (flag 20) and never enter the exit. On the ten-year medians, Yelp's is inflated by thin-profit years (75x in 2019 and 2021) but the median of four is not: dropping any peer gives 24.7-25.2x, and Meta with Alphabet 23.1x ($301) | — |
| 10 | Advocate | The $138 world built from dated evidence is worth $151 at its harshest; Avoid needs an exit below about 9.8x | **Accepted as the market world** (below) | — |
| 11 | Advocate | U.S. revenue per user, the margin path and licensing held up | Noted | — |

Together: $339.24 to $311.31 (-$27.93; the two alone sum to -$28.50, the
interaction +$0.57; `output.txt` and the isolating runs in `process.md`).
The view is unchanged: Buy.

**Flag 16: a heavily traded price outside nearly all our scenarios.** The
blind odds put 92% of our simulated values above $138.47. Part of that was
our error: the bear world in the simulation was milder than our own bear
case, and fixing it brings P to 82%, with the bear world's median at $124,
below the price. The rest we take as the warning the flag intends. The
advocate's evidence-built world, with revenue of $6.8 billion in 2031 at a
36% margin, is worth $195 at our exit and $151 with both today's peer
multiples and Reddit's own beta; it reaches the price only at an exit of
about 12.5 times operating profit, and Avoid below 9.8 times
(`review/worlds.txt`). Our joint reverse solve needs every driver 80% of
the way to the bear with an 18.7 times exit (`market.txt`). So the market
is paying for something close to our bear: a search-dependent business
valued like Yelp and Tripadvisor today rather than like Meta and Alphabet
over ten years. Where our ranges may be too narrow is the exit: the
simulation's lowest outside-view multiple is 12.8 times (the bear world's
low end), and today Yelp trades below that and Tripadvisor at about it. We do not widen it toward the price; we name
it. We concede the central if U.S. daily users fall for two quarters
and U.S. revenue per user grows below 20% in a 2027 quarter: that is the
bear's path, and the market's.

**Flag 23: the odds' 90th percentile ($687) is 5% below the bull ($723).**
The simulation does reach our optimistic case: 348 of its 4,000 draws (8.7%)
value Reddit at or above $723 (`odds.json`), so the bull is about a
one-in-eleven outcome rather than one-in-ten. The gap comes from the shared
case spread, as on the Atlassian and Salesforce reports: the method sets
the bull's outside view at the pooled 75th percentile of realised over
anchor (1.456 times, 36.3x), which is also the top of our exit-ratio range,
so few draws sit there, while the simulation's own exit-price draw is
centred on 1. The bull's operations at the default exit are worth $612
(`output.txt`: the case spread adds $111), below the 90th percentile. The
flag changes no value: the central $311, bear $115 and bull $723 stand, and
the ranges stay as registered.

## 7. The bear case: search traffic falls, and the rest follows

**The world.** Google's AI Overviews and AI Mode, and the AI assistants,
answer the question on their own page. Reddit's search visitors fall 10% a
year in the U.S. in 2027 and 2028 and keep falling more slowly; machine
translated pages lose rank abroad. Fewer visitors means fewer conversions
to logged-in and app users, so U.S. logged-in users stop growing. The
commercial-intent questions, the impressions advertisers pay most for, are
answered by AI first, so revenue per user slows from 50% growth to 26% in
2027 and 5-8% later. Google and OpenAI renew licensing at half. Reddit
spends to replace the traffic (0.55 of each revenue dollar added) and keeps
stock pay at 15-16% of revenue. Global DAUq falls from 130 million to 118 million;
revenue grows 20% in 2027 and 6-9% a year after, to $5.5 billion in 2031 at
a 29% operating margin. At the bear's exit multiple (17.2x, the pooled 25th
percentile) the value is **$115**.

**How much of the bear is search** (`output.txt`, one driver group at a
time from the central $311):

| Step | Effect |
| --- | ---: |
| U.S. logged-out users fall | -$19 |
| International logged-out users fall | -$10 |
| Logged-in growth stalls (the funnel) | -$25 |
| U.S. revenue per user slows | -$58 |
| International revenue per user slows | -$26 |
| Licensing at half | -$5 |
| Q4 2026 at +45% | -$4 |
| Higher costs to replace traffic | -$19 |
| Stock pay stays at 15-16% | -$12 |
| Exit at the bear's multiple | -$19 |

The search visitors themselves cost $28. The bear is mostly what follows
from them: the funnel and the pricing.

**Why the other side could be right.** Reddit's own disclosures support
this bear more than its words do. Its logged-in base in the U.S. grew 1% in
a year, after two years of product and marketing work; the CFO called paid
marketing "mixed" and retention the "key unlocker"; the company stopped
reporting the logged-in split; and a shareholder suit says it misled
investors about the effect of AI Overviews (10-Q note 9). If the commercial
questions move into AI answers, Reddit's pricing growth has no floor we can
date.

**The de-rating path.** If our central forecast holds but the market keeps
paying the bear's all-in 15.8 times 2031 operating profit, the exit is $279
and the value today $227, 27% below the central. What would keep the
multiple there: U.S. DAUq falling for several quarters with the split no
longer reported, so the market cannot tell drive-by losses from core
losses; a lapsed Google deal read as a broken relationship with Reddit's
biggest traffic source; and growth slowing toward the 13% our own central
reaches by 2031.

## 8. Next results preview: Q3 2026, reported 2026-10-29 after the close

Date from the company's release on Business Wire (2026-10-15): results after
the close on Thursday 2026-10-29, call at 1:30 p.m. Pacific. Forecasts from
the guide plus the eight-quarter beat record (`output.txt`):

| Metric | Ours | Range | Reason |
| --- | ---: | --- | --- |
| Revenue ($M) | 954 | 930-1,029 | Guide midpoint $865M plus the median 10.3% beat (63% growth); range is the smallest and largest beat |
| Adjusted EBITDA ($M) | 466 | 446-531 | Guide midpoint $390M plus 0.86 of the revenue beat |
| Diluted EPS ($) | 1.64 | 1.51-1.99 | EBITDA less stock pay at the guide's midpoint ($147.5M) and D&A, plus about $25M interest, 2% tax, 202M diluted shares |
| U.S. DAUq (M) | 53.0 | 52.0-54.2 | Q2 53.2 and falling search referrals late in Q2 |
| International DAUq (M) | 78.0 | 75.5-80.5 | Q2 77.1; machine-translated pages losing rank |
| Other revenue ($M) | 44 | 42-48 | Licensing contracts unchanged until early 2027 |
| Q4 2026 revenue guide midpoint ($M) | 1,020 | 945-1,046 | Our Q4 $1,125M (+55%) less the median beat |

The one that moves value most is revenue: each $10M of Q3 revenue above
ours, at the median flow-through to EBITDA and carried as a higher level of
revenue per user, is worth about $1.12 a share (`output.txt`).
A U.S. DAUq below 52 million would be the first quarter of the bear's user
path.

## 9. Catalysts and falsifiers

- **2026-10-29, Q3 2026 results** (above). The first quarter without the
  logged-in split; watch whether U.S. DAUq holds 53 million.
- **Early 2027, Google and OpenAI licensing renewals** (contracts run out,
  10-Q note 3). Renewal near or above $170M a year supports the central; a
  lapse with Google that also changes Reddit's search placement moves toward
  the bear.
- **Q4 2026 results, about early February 2027** (not announced): the 2027
  shape. Our central needs 2027 revenue growth near 36% (35.6%).
- **Falsifier for the central:** U.S. revenue per user growing below 20% in
  any 2027 quarter (our path is 36% for the year), or two quarters of U.S.
  DAUq declines.
- **Falsifier for the bear:** U.S. DAUq above 55 million by Q4 2027 with
  revenue per user still growing 25% or more.

## 10. Sources

Primary: Reddit 10-K FY2025 (filed 2026-02-06), 10-Q Q1 and Q2 2026 (filed
2026-05-01 and 2026-07-31), earnings releases and shareholder letters Q2 2024
to Q2 2026, 8-Ks to 2026-08-12 (SEC EDGAR, fetched 2026-10-26, masked); Q1
and Q2 2026 calls and the Goldman Sachs Communacopia conference (September
2026) transcripts (market data/Quartr, secondhand transcripts of primary
speech); Meta release 2026-07-29 and Pinterest release 2026-08-04 (EDGAR);
Market data statements and peer histories (fetched 2026-10-26); Damodaran industry
betas, January 2026 data. Results date: Reddit release on Business Wire,
2026-10-15. Press: CNBC 2026-07-22 and 2026-07-30, TechCrunch 2024-02-22,
CNBC 2025-10-23.