Liam Walsh

𝗔𝗿𝗲 𝗹𝗼𝘄𝗲𝗿-𝗶𝗻𝗰𝗼𝗺𝗲 𝗰𝗼𝗻𝘀𝘂𝗺𝗲𝗿𝘀 “𝗿𝘂𝗻𝗻𝗶𝗻𝗴 𝗼𝘂𝘁 𝗼𝗳 𝗺𝗼𝗻𝗲𝘆” 𝗮𝘁 𝘁𝗵𝗲 𝗲𝗻𝗱 𝗼𝗳 𝘁𝗵𝗲 𝗺𝗼𝗻𝘁𝗵? We used Carbon Arc’s ZIP-level credit card data to show March+April Y/Y changes in late-month spend ratios for low-income vs. high-income geographies. The results? Lower-income ZIPs showed stronger late-month gas-station spending and the sharpest relative deterioration in discount merchandise, restaurants & QSR, and a pullback in traditional grocers. This is consistent with lower-income budget stress in a high gas price environment. Try Lenses free for 30 days using code 𝗟𝗔𝗨𝗡𝗖𝗛𝟯𝟬: carbonarc.co/lenses-welcome

Two days. Main stage. Workshop floor. Carbon Arc was everywhere at AI Hot 100 Summit this week. Our co-founder @MckeownKirk held it down alongside LSEG and Morgan Stanley on why data infrastructure is the real bottleneck in AI. @yuvkumar and @liam_walsh19 showed a room full of practitioners what it looks like when that problem is solved. Try Lenses free, code 𝗣𝗥𝗢𝗗𝗨𝗖𝗧𝟯𝟬 at carbonarc.co/lenses-welcome #CarbonArc #AIHot100

Most people wait for confirmation in macro prints or company commentary. By then stories tactically move and if you’re trying to find an edge it’s already degrading. Following physical flows like shipping and transit movement can give you a cleaner read on where pressure is building. Then if you know your history you know how this macro reality can translate into a market reality and move accordingly. All of this data is available to everyday folks for the first time @CarbonArcAI

What stands out to me: - Strong build in arrivals through Feb 27 - Clear slowdown immediately after - Concentrated across Hassyan and Sharjah/Khalid anchorage areas We can detect hesitation and deceleration in the aggregate before markets call a freeze in traffic.

How do you actually track geopolitical risk before it shows up in markets? At CA we can watch where physical flows start to break. Tanker traffic into key Gulf anchorages built steadily into late Feb, then shifted quickly as tensions escalated. 🧵👇

Elections with real disruption potential often show up in the data before they show up in polling. DM if interested in more data im looking at across 26 elections. Like the below congressional map 👀 by @CarbonArcAI

Alaska voters tend to highly sensitive to energy prices. With supply shocks from the Middle East pushing energy futures higher, pressure on households could rise quickly. Instead of only watching polls, we can track signals like: - POS spend at gas stations - Convenience store spend - Vehicle registrations of heavy autos

How do you predict election upsets before the polls show it? Here is one example from the AK Senate race. Mary Peltola is expected to face two-term incumbent Dan Sullivan, who currently holds one of the lowest approval ratings in the Senate (-8 net). Peltola already flipped Alaska once in the 2022 special election. 🧵 below 👇

@SowingAlphaSeed @ethanrkho We’re building that source @CarbonArcAI (Kirk’s company). Wall Streets data stack, available to anyone.

@ethanrkho > The only personal edge that doesn't decay: a corpus of knowledge you built yourself. Somewhat analogous, I have a theory that Renaissance's biggest source of alpha is having the largest, cleanest set of historic market data.

The only personal edge that doesn't decay: a corpus of knowledge you built yourself. Got this from Kirk McKeown, ex-Point72 Head of Prop Research who worked alongside Steve Cohen. "You want to get it right, not be right." "Alpha rewards those who value assets in a cold way." "History rhymes because people are animals. They do the same thing over and over expecting a different result." "Reading a book. Having these conversations. That's the competitive advantage. Not asking a chatbot." The irony: in an AI world, the edge goes to whoever puts in the most analog reps.

𝗧𝗼𝗱𝗮𝘆 𝘄𝗲’𝗿𝗲 𝗹𝗮𝘂𝗻𝗰𝗵𝗶𝗻𝗴 𝗟𝗲𝗻𝘀𝗲𝘀. For the past five years, we’ve been building Carbon Arc — the infrastructure that makes real-world transaction data accessible to institutions around the world. Today, on our anniversary, we’re expanding that foundation with a new product. Lenses is an AI-powered research interface on top of our full external data catalog — consumer spend, foot traffic, app downloads, pharmacy claims, payroll, logistics, and more. Pick a lens that fits how you work: 𝗜𝗻𝘃𝗲𝘀𝘁𝗶𝗻𝗴? Build conviction before the Street catches up. 𝗠𝗮𝗿𝗸𝗲𝘁𝗶𝗻𝗴? Track competitive share as it shifts. 𝗖𝗼𝗻𝘀𝘂𝗹𝘁𝗶𝗻𝗴? Scope a vertical in minutes, not weeks. No pipelines to build. No queries to write. Just ask a question and get structured answers — tables, charts, narrative context, and follow-up depth — all in one conversation. See it in action in the video below ↓ 𝗙𝗶𝘃𝗲 𝘆𝗲𝗮𝗿𝘀 𝗼𝗳 𝗶𝗻𝗳𝗿𝗮𝘀𝘁𝗿𝘂𝗰𝘁𝘂𝗿𝗲. 𝗢𝗻𝗲 𝗶𝗻𝘁𝗲𝗿𝗳𝗮𝗰𝗲. To celebrate the launch, use code LAUNCH30 for 30 days free. 𝗗𝗼𝗻’𝘁 𝗯𝗲𝘁 𝗼𝗻 𝗼𝘂𝘁𝗰𝗼𝗺𝗲𝘀. 𝗖𝗿𝗲𝗮𝘁𝗲 𝘁𝗵𝗲𝗺.

High-income job declines don’t work well for the housing market. This is a trend (namely at local level) we’ve been hammering on the last few years as we maintain a weak outlook for housing in general. wsj.com/economy/jobs/t…

If you have the Carbon Arc MCP Server connected in @claudeai, you have it in Excel! Seamlessly ask for and bring data from Carbon Arc right into your spreadsheet. Demo below 👇

He’s due

@the_P_God TouchTunes has a “wrapped” btw.

Not delta wrapped

@AdamOnFinance What data are you looking at indicating 95% of BF txns are non-cash or debit?

Notable that Black Friday sales data shows: -A 9.1% increase spend from last year. But: -A -1% in total item volume from last year. -Prices +7% higher. -Consumers bought on average 4.1% fewer items. And: -An 11% increase on buy-now-pay-later use. -Klarna specific use up 45% by volume since last year Meaning: -Roughly 11% of ALL Black Friday spending was financed through BNPL. -And 84% of all purchases were financed by credit cards, where 67% of those consumers expect to not pay the full balance in the first month. So overall: -A total of 95% of Black Friday shopping ($11.2B) was financed. -And, 67% ($7.9B) was financed on debt that consumers do not expect to be able to pay in the next 30 days. This is the sign of a weakening and stretched consumer.

@JacobAShell Based.

Why pay a consulting fees to analyze Enterprise cases when any small data org can wire an enterprise data stack to an MCP sever and query their own data in natural language? This crushes the narrative around consulting firms as decision insulation for CEOs.

@patrick_oshag thedublinreview.com/article/even-i…

Anyone read a great essay or article lately?

@AJBrownOrtiz They don’t tell you the deli has no minimum weight on cold cuts, have a $0.30 slice

Need a doctors note that lets me buy a single slice of cheese from the grocery store