Doug Garber

From a Zero-Listener Podcast to the Data Layer Behind Google Finance & Perplexity. Fiscal AI's Braden Dennis Braden Dennis @BradoCapital – CEO & co-founder of @fiscal_ai, and a fellow stock nerd. Started as a college podcaster with zero listeners. Today his company aggregates first-party filings agentically and powers @googlefinance , @PPLXfinance 's product, and @Kalshi's KPI prediction markets. It delivers standardized financials in two minutes, not two days. We cover: The three-iteration journey: Stratosphere (YahooFinance on steroids for retail) → FinChat (the first commercial LLM app in finance — 60,000 users in 48 hours) → Fiscal AI (the data-infrastructure layer) Why he abandoned a profitable retail SaaS to own the underlying content and why the data-feed business, launched only ~9 months ago, is already ~70% of revenue The disruption thesis: a $46B financial-data market where five incumbents do ~$13B of feed revenue, most of it still manually aggregated attacked on speed, accuracy, and price (2 minutes vs T+2, even for large caps) The economics nobody expects: Fiscal is a CapEx-heavy token business, not a gross-margin one multiple seven figures a quarter on compute, reportedly the most token usage of any Canadian startup, spent building the dataset, not serving it 120%+ net dollar retention (targeting 150%), and why post-sale activation, the human implementation layer still decides who renews, even in an agentic world The founder playbook: "strong opinions, very loosely held," deciding on gut with no perfect information the same phrase a @TRowePrice co-PM used to describe a great equity analyst Fundraising, plainly: SAFEs over priced rounds at seed, ~$10M caps, why raising too high can trap you, and the ~$100K ARR bar that made them investable Why "what's your moat?" is the wrong question to ask a seed founder — and where the real moat ("the Saudi Arabia of data") actually gets built The news product built to kill "AI slop" 19 event types, ~280 junk sources screened out, signal over the "Citi cut its $NVDA target" noise How the data now runs through an MCP straight into @claudeai and @ChatGPTapp comp-set, earnings-quality, and driver-based model skills that give an analyst "a developer at their disposal" *Not investment advice. A founder/business-building episode. Highlights: (0:48) Intro: Braden Dennis & Fiscal AI (1:14) From a zero-listener college podcast to an investing obsession (5:57) What Fiscal AI actually is, and who it's for (7:59) Three iterations: Stratosphere → FinChat → Fiscal AI (11:39) Getting to a seed round: what you actually have to show (14:38) SAFEs, caps, and the fundraising mechanics founders get wrong (16:37) The pivot to owning the data + powering KPI prediction markets (20:30) Gut decisions and "strong opinions, loosely held" (24:39) The KPIs of the business: 120%+ net dollar retention (26:22) Token economics: CapEx-heavy, not gross-margin (34:52) The $46B data market and the incumbent oligopoly (44:48) Live demo: MCP into Claude & Codex

The Oil-Cycle Analyst T. Rowe Brings In to Consult — Robert Connors on the Marginal-Cost Floor, Refiners, and Offshore Robert Connors (@crudechronicle ) — author of Crude Chronicles, former sell-sider, now doing some of the deepest oil-cycle work anywhere: 100+ years of history, Library of Congress data. His list runs 10-15k CEOs, CFOs, analysts, and PMs managing trillions. @TRowePrice brings him in to consult on the oil cycle. "What actually drives these cycles - with a 0.9-plus R² – is the well productivity cycle." Everyone else is forecasting supply and demand. He's not. We cover: The framework: why supply/demand forecasting is the wrong game. Marginal cost, set by the well-productivity cycle, is the floor for oil (0.94 R² since 1979) The three-wave cycle: doubt → optimism → parabolic euphoria and why the Iran war just moved us from doubt into the optimism wave The productivity turn nobody's watching: well-productivity growth peaked at 18% in 2017, ran just 4% in 2025 - slowest outside COVID. Rising marginal cost = rising floor The tell in the well count: producing wells across the majors peaked ~325k in 2012-13, more than halved to under 150k and it's now bottoming (latest Texas Railroad Commission data) Why he's still bullish refiners and the metric that actually drives refining profit: secondary-unit capacity (hydrocrackers, cokers, FCC), not nameplate utilization ~1.7M b/d of upgrading capacity pulled out of the market, and no major refineries coming online the rest of the decade The incentive edge: why $PSX , $MPC and $VLO have the best comp design in oil & gas (relative TSR benchmarked to the S&P 100/500, not just each other) and why he likes $CHEV.NE over $XOM The offshore setup: lowest newbuild orderbook since ~1990, a consolidated field, European majors coming back, and 10%-to-40% operating leverage. Top pick: Transocean The AI-capex parallel: why hyperscaler capex looks exactly like the shale and refining supercycles, why "not one metric" ties mega-tech pay to a capital budget, and why that means capex only goes higher "The rig is the liability. The contract is the asset." Stocks: $PSX $MPC $VLO $CVX $XOM $RIG *Not investment advice Highlights: (0:00) Intro — Robert Connors & Crude Chronicles (2:10) The Iran war and the three-wave oil cycle (4:05) Where we are in the US shale productivity cycle (9:07) What the 4% productivity growth signal means (11:29) Pegging the marginal cost of production (14:21) Where the next 10M barrels come from — Guyana, Canada, offshore (18:15) The marginal-cost model and the 0.94 R² with oil (24:14) The bull case for refiners and crack spreads (29:33) How much refining capacity is permanently gone (34:31) AI capex vs the oil & gas capex supercycle (39:54) Favorite oil stocks — refiners, Chevron, and why incentives matter (43:24) The offshore driller setup and Transocean

A Top-Ranked Magic: The Gathering Player Just Pitched Me a Short on the Company That Makes It. The Hasbro Variant View Glenn Lee & Vito, two Georgetown students who won both the Fordham and Notre Dame stock-pitch competitions, now headed to the Balyasny Catalyst program and the @Point72Careers Academy. Vito is a top-ranked Magic Arena player. Their winning pitch: a short on Hasbro ($HAS). $56 PT, ~36% downside, ~2x risk/reward, 12-month horizon. "We came at it as a long first, then followed the evidence to the short side." We cover: How the pitch flipped from long to short: the numbers wouldn't support the street's back-half-'25 and '26/'27 expectations, so they followed the evidence instead of their priors Why Hasbro isn't a toy company: ~75% of EBIT is Wizards of the Coast, overwhelmingly Magic: The Gathering physical cards toys and board games are under 10% of the value What actually drove 2025's ~60% Magic growth: set count going 4→7 a year, price up 6-7% annually (doubled since 2019), and "Universes Beyond" and why they think it's a one-off, not a new run-rate The variant view: Universes Beyond backlash. Management cites 9% of players unhappy (a 2020 survey); their own player survey put it closer to 40% The alt-data that sealed it: the last three Universes Beyond sets selling at a discount to MSRP for the first time, while in-universe sets hold premiums; a set-premium series with ~0.85 correlation to revenue @CarbonArcAI weekly app users down ~11% quarter-to-date into Q2, and a Steam player count that spiked on Final Fantasy then halved players aren't sticking Reading the setup: short interest doubling from ~3-4M to ~7M shares YTD, and why Doug pushes back on calling it a "consensus long" (lazy short vs up-and-to-the-right) The CEO tell: Chris Cox sold ~40% of his stake two days after a record Q4 and how to weigh insider selling honestly (evidence, not proof) The margin trap the street's missing: Universes Beyond carries higher partner royalties (~7% of revenue and rising), diluting the mix-shift-to-Magic margin story The number: $56 PT on ~flat-to-down Magic vs the street's low-double-digit growth, ~10x EBITDA held flat (no multiple contraction assumed) A real-time clinic in the 10-step Variant View process: quantifying crowding (13F fast money, ETF ownership, short-interest z-scores) and staying intellectually honest with disconfirming evidence Stocks: $HAS *Not investment advice. Shorts are especially risky and not suitable for most investors. Highlights: (0:00) The short-interest tell and the CarbonArc app-user data (0:20) Intro: Glenn & Vito, competition winners headed to Balyasny & Point72 (2:03) Disclosures and how a Magic superfan ends up bearish (4:17) Flipping the pitch from long to short following the evidence (6:14) The two-pillar thesis: Magic weakening + directionless consumer products (7:31) Reading the setup short interest doubling and "consensus long" (13:05) Why Hasbro is really a Magic: The Gathering company (75% of EBIT) (13:43) How Magic makes money sets, pricing, and the secondary market (17:38) Who actually buys Magic the 35-year-old "kidult" (23:24) What drove the ~60% 2025 growth set count, pricing, Universes Beyond (25:28) The CEO selling 40% and how to weigh insider sales (30:34) Intellectual honesty collecting evidence that cuts against you (36:51) The alt-data set-premium discounts and the backlash