We Read 194 CryptoHealth White Papers And Found 6 Variables That Predicted A Successful ICO
Many thanks to Ava Zhang, Jenna Chen, Shaohua Dong, Wenwen Jia, Yi Ren, Jiayi Chen, Yuxuan Lu, Yuqi Peng, Sunny Bae, Derek Kang, Jacky Liu, and Eleanor Haas, Guilherme de Oliveira and Dr. Mark Ritzmann for their work on this project
A recent report from the Institute of Health Economics at Wharton summarized a long list of deceptive, unsubstantiated and foolish statements in healthcare. Buzzwords like ‘scale’, ‘synergy’, and ‘three-letter-acronyms’ (TLAs) have plagued Healthcare with over-promise (think Watson) and outright fraud (think Theranos).
Similarly, the CryptoHealth landscape has seen hyped reports stating that by 2020, 1-in-5 healthcare organizations will adopt blockchain and reach by 2025 a $5.6 billion market. Complicated by the fact that most projects had failed their ICO, we felt a need not only to map, but also evaluate the quality of current white papers and see if there are any reliable predictors to a successful capital raise.
From August to December, we collaborated with interns and Capstone students from Columbia University and analyzed the Center for Biomedical Blockchain Research (CBBR) database (GitHub repository here), ICO rating sites and other public domain repositories and this is what we did.
Step 1- Preparation: Reading 194 White Papers Required Meticulous Planning
Anyone who read a white paper knows it can take time. In a recent analysis, Michael Rosenberg showed that increasing the length of a white paper by 1 page increases the amount raised in an ICO by around 1%, regardless to seasonal trends in fundraising. Therefore we planned ahead our work (below) and developed an evaluation matrix and data dictionary.
The matrix was based on reviewing published ICO evaluation methods (here, here, here and here) looking at: ICO economics (soft and hard cap, distribution schedule, KYC/AML, token design), project fundamentals (idea, impact, team composition, domain expertise, product stage, roadmap, allocation plan, funding) and hype metrics (telegram, twitter, blogs, rating sites, bounty programs).
In total we read 194 white papers, using 39 variables clustered in 7 categories. After cleaning unstructured data we analyzed 126 live projects using 31 variables in 5 weighted categories.
Step 2- Analysis: We Kinda Knew Already Some of the Results, But Others Caught Us Off Guard
Many results were similar to those reported (like Robert Miller at Honeycomb). Blockchain-based solutions were either Electronic Health Records (EHRs) and health information exchanges (HIEs) (63), supply chain (13) and revenue cycle management (12) projects, operational and inter-operational platforms (13) and credentialing, registries and analytic programs. Almost half of the projects were from the US, 8 from Russia and 7 from the UK; 56 projects were in production, mostly on Etheruem and 16 of them were on exchanges.
With a soft capital raise rate of 24% we found 6 variables that predicted a successful raise namely:
(1) Having domain expertise (CTO, legal, previous ICO - yes; MBA, MD - no)
(2) Having a detailed business model (despite the fact that only a 1/3 of the white papers had one)
(3) Increased Twitter vitality rate (we did not measure twitter activity alone, since most companies significantly decrease it after an ICO)
(4) More economic, financial and legal disclosures (we used a sparseness analysis described here). This is similar to the study results of De George and others who showed that poor disclosure rate is typically followed by an ICO crash within three to six months.
(5) Teams that distribute < 15% of tokens to themselves
(6) Teams that distribute <50% tokens at the ICO
Finally, after we did not find differences between companies with similar/dissimilar team structures, token features and twitter followers using cluster analysis, (specifically k means clustering), we text mined the white papers and found that companies using the words ‘Contract’, ‘Technology’, ‘Users’, ‘Ecosystem’, hit their soft cap whereas those that used ‘Neural’, ‘Network’, ‘Patients’, did not.
These results are limited by our sample size, but are interesting since
the idea that machine learning can identify failing ICOs or cryptocurrency scams before they happen can be extremely useful.
(For example between January-July 2018, telegram alone had 3,767 attempts at pump-and-dump schemes).
Step 3- Deliberation and 3 Conclusions
Conclusion #1: CryptoHealth companies are like any other Blockchain company
The predictors of success, vulnerabilities for failures and potential are like with any industry. Furthermore, because tokens are means by which future users can gain access to a platform, their purchase through an ICO is directly indicative of future demand. However, as Catalini and Gans explained, if the majority of token buyers are speculators and not future users, then the pre-launch demand for the token is uncorrelated to the demand for the platform.
Therefore companies must use token sale strategies (low self distribution, longer lockup and sales periods), that will deter speculative behavior.
Conclusion #2: CryptoHealth companies are like any other non-Blockchain digital health company
When comparing team composition, expertise and structure, CryptoHealth start ups are like any other digital health startup (too male, too white and too rich). The fail fast and frequent, but the basic formula for successful funding remains the same:
- Come up with a good, valuable idea
- Work with accredited investors to gain an adequate amount of funding
- Build your community (show, tell and receive feedback)
- Go public to retail investors, get listed (beware fake ones) and be ready to have token liquidity when you are on an exchange
- Use funds to show traction and deliver based on your roadmap
We did however note that the main cause for blockchain-based companies to fail is when they abandon the main reason to use blockchain, decentralization.
In fact over 80% of ICOs have broken their promise and retained centralized ownership of their issued currencies, thereby violating one of the fundamental principles of cryptocurrency.
Conclusion #3: CryptoHealth Ecosystem remains seriously underfunded to its potential for profit and impact
Investments in CryptoHealth throughout 2018 remained small in scope (1.5% of all crypto investments for an industry that represents close to 20% GDP) and in size ($330M vs. $7B in traditional VC funding).
It is important that investors in blockchain start to become seriously interested in CryptoHealth and fuel innovation in a system where doctors use computers for everything, badly, inefficiently and insecurely.
Final thought:
Earlier this year I wrote “Why Crypto needs a Doctor and Medicine needs Blockchain Technology” and since then I have been trying to “blockchainize” healthcare and “healthify” the crypto space. I still strongly believe that Healthcare is the most amenable industry ready for blockchain disruption.
However what Healthcare needs is more than Blockchain technology. It needs a Blockchain mindset. Blockchain itself will not disrupt a thing. Only we, with Blockchain will.
For more, please check out our upcoming book Blockchain and Healthcare (below)
ISBN 9780367031084 — CAT# K405509 and can be found here
If you liked what you read, go ahead and “Clap” below so others will see it too (up to 50 claps allowed!)
