← All work
LunarCrush

LunarCrush

Real-time social and market intelligence for crypto.

CDO
Co-Founder · led a team of 2 designers
'17–'22
Tenure
2
Galaxy Score™ · AltRank™
Overview

In late 2016, crypto news and analyst notes arrived after the price had already moved, but social media reacted in real time. We asked whether machine learning could read the market from what people were saying. I built the prototype that showed the correlation, then designed the brand, the web app and the Galaxy Score™ as co-founder, and later led design as Chief Design Officer.

Result

Built from zero: the brand, the web app and the Galaxy Score™ algorithm. After the v2 launch, users went from 150,000 to 2.5 million in sixty days. LunarCrush is still growing today under the original co-founders.

It started as a question, not a company: could machine learning read a crypto market from what people were saying about it?

In late 2016 the traditional signals were slow. News, analyst notes and media coverage all arrived after the fact. Social was different. People reacted in real time, and they moved in herds. The premise was that if you could measure the herd you might predict the price, and you might also find out whether attention was driving the market rather than following it.

I built a prototype to find out. It ingested Twitter, Reddit and a few other sources and plotted social engagement on a chart against price, so the answer would be visible before any model was. It was. Engagement moved, and price moved with it.

That chart is what turned an experiment into LunarCrush. We co-founded the company in 2017 and spent a year building the first version of lunarcrush.com, which I designed and built the front end for. Seed funding followed, from prominent blockchain investors with Tim Draper among them, and adoption took off. We built the platform out from there, adding sources until it was ingesting hundreds of millions of data points a day and correlating them with price across every coin, and later traditional stocks. LUNR came next, and I joined full-time in 2021.

The premise

Social reacts faster than news, and it moves in herds. Measure the herd and you may be ahead of the price.

The prototype

Twitter, Reddit and a few other feeds on one chart against price. The correlation showed up by eye.

The platform

Many more sources, hundreds of millions of data points a day, every coin, and later stocks.

Galaxy Score™

Correlating those metrics needed a new instrument: price performance, social activity over time and a few other signals combined into one number that reads as the health of a coin. Named to fit the world we had built.

A market that moves on attention needed an instrument that measures it.

The platform ingested hundreds of millions of social and market data points a day, and nobody can read that. The design problem was turning it into something a trader could act on at a glance: Galaxy Score™, one number for a coin’s health built from price performance and social activity over time, and AltRank™, which ranks a coin against the rest of the market.

Before

  1. Read the news after the price has moved
  2. Check analyst notes, one coin at a time
  3. Guess at what the crowd is doing

With LunarCrush

  1. Social and market data arrive in real time
  2. Models correlate engagement with price
  3. One score per coin: Galaxy Score™
  4. See the herd move, then decide

The trader

  • Choose the coins to watch
  • Read the score and the signals behind it
  • Decide whether to act

The models

  • Ingest social and market data from many sources
  • Correlate engagement with price
  • Score each coin’s health and rank it against the market

Draw the chart before building the model

Before any machine learning, I plotted social engagement against price on one chart. The correlation showed up by eye, and that picture, not a model, is what turned an experiment into a company.

Open-source the widgets

We released an open-source widget framework so anyone could embed LunarCrush’s live data on their own site. It became the company’s first revenue source, in 2018.

Did a chart of social data become a product people used? These say it did.

150K → 2.5MUsers in sixty days after the v2 launch
50M+Pageviews a month
$2.5MSeed round the first brand and product helped raise
15Full-time staff by 2021

LunarCrush turns billions of social data points into market intelligence for crypto. Galaxy Score™ and AltRank™ help investors and traders read a market that moves on attention.

Co-founded with three partners in 2017, part-time while at Google, to design and build the first brand, website and product. Those helped raise a $2.5M seed round and grow the company to fifteen full-time staff by 2021. An open-source widget framework for embedding live data became the first revenue source in 2018.

I joined full-time as Chief Design Officer in June 2021, leading UX and UI for web and mobile with a team of two designers. Within six months we shipped a new brand, website and product, iOS and Android apps from scratch, and the LUNR utility token and wallet. After the v2 launch in October 2021, users went from 150,000 to 2.5 million in sixty days and the site passed 50 million pageviews a month. I stepped back from day-to-day in January 2022. LunarCrush keeps growing under the original co-founders.

Brand & web app

Designed the brand and the web app from scratch, then the rebrand, lunarcrush.com and a new design system in 2021.

Galaxy Score™

Developed the first Galaxy Score™ algorithm, a measure of a coin’s health over time.

React front-end

Built the first web app front end in React and supported the engineering team on UI.

Mobile and the token

iOS and Android apps from scratch, and the LUNR token and wallet, inside six months of joining full-time.

Without a community, there is no crypto.

LunarCrush brand line
The Lunr mark, from the opening of the explainer film
What is Lunr? Fifty-four seconds on the token that pays for the data.
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • LunarCrush
  • Brand & visual identity
  • Product design (web + mobile)
  • React front-end engineering
  • Data visualization
  • Algorithm design (Galaxy Score™)
  • Design leadership
  • Product direction
  • 0 → 1 building

Make the answer visible before you model it.

LunarCrush started as one chart: social engagement plotted against price. The correlation showed up by eye before any model did, and a chart anyone can read is what turned an experiment into a company. When you are testing an idea, the fastest proof is usually a picture, not a pipeline.

Still a leading source of social-market intelligence for crypto.

Visit lunarcrush.com ↗