A few years after graduation, former college roommates Greg Ugwi and Justin Zhen found themselves on Wall Street, building financial models for an investment bank and a hedge fund, respectively. They grew certain there had to be a better application than the Excel spreadsheet, and a better approach to collaboration and sharing than emailing massive files among analysts.
Turning to the web, they coded a platform for people who create valuations models for publicly traded companies, and named it Thinknum (a combination of the words “think” and “numbers”). Thinknum offers free, immediate access to thousands of data sets and enables users to share and collaborate on financial models that are stored in the cloud.
These models update automatically when company earnings reports hit EDGAR. Multiple analysts can work on the same document simultaneously with no issues of version control. No coding is required, though you still need to know your way around a spreadsheet and an income statement, a balance sheet, and a statement of cash flows.
Thinknum recently announced $1 million in seed funding led by Pejman Mar Ventures.
Greg and I recently sat down for a conversation via Skype.
Q. Greg, congratulations on closing your seed round. How long did it take to pull together?
A. When we pitched potential investors, we noticed a bifurcation. Angel investors with a background in finance understood what we were building and were quite enthusiastic about investing in Thinknum. Angels from the tech world did not necessarily understand the problem that we were solving. Fortunately, Pejman Mar Ventures was the first institutional fund that we pitched. They brought in their friends from hedge funds who understood our vision and ultimately Pejman Mar Ventures became our lead investor.
Q. Did you use Thinknum to share your financial model with potential investors?
A. Yes we did. We eat our own dog food!
Q. How are you planning to use the new capital?
A. We plan to use the new capital to hire engineers. The most difficult aspect about building our company is getting the right talent. The biggest banks and hedge funds are fighting over the same people that we are after. Fortunately, some engineers are excited about getting into the tech startup space and working on the problems we’re trying to solve. We plan on bringing on three engineers over the next couple of weeks.
Q. You recently took part in 500 Startups. Are you happy with what you got out of it? Would you recommend it to other FinTech startups?
A. Absolutely. Justin and I worked on Wall Street so we know our way around financial models and we have been writing code daily for years. Where we needed help was how to start and grow a business. At 500 Startups, we learned a lot about building a company. For example, how to raise money, attract the best talent and market our product. We also learned a great deal from other founders in our batch who are building companies.
Q. Did you consider any other accelerator programs with greater FinTech focus?
A. Dave McClure, the founder of 500 Startups, was part of the Paypal mafia so 500 definitely has FinTech in its blood. I would advise other FinTech entrepreneurs to go for any accelerator that can help you grow the quickest – that is certainly what we optimized for. I don’t think you can learn to write code, price bonds or read a balance sheet at an accelerator. As an entrepreneur in FinTech, you should already know your way around finance and tech. However, you can learn how to build and grow a startup at an accelerator and this is where 500 startups excels.
A. The core question that we are attempting to answer is what computation should look like in a networked world for non-programmers. In the desktop age, spreadsheets won that war. Users are quite familiar with the cells and charts and spreadsheets. We have no interest in recreating the wheel, so our web software uses the same spreadsheet and charts which are familiar interfaces for our users. Because our application is based in the browser, our users can take advantage of distributed computing to run analysis on thousands of computing nodes concurrently. They can access our database of over 5 million time series using the plotter and Cashflow Model. This is not possible with a desktop application.
Q. What capabilities do you plan to add next?
A. Users will soon be able to add annotations and comments. We are also about to roll out our portfolio tool where a user can run multiple models, screen the universe of securities and connect their brokerage accounts to analyze their current holdings. Finally we are rolling out additional version control features that would enable non-programmers to interact with and benefit from GIT.
Q. Access to financial data is often a costly hurdle for FinTech startups, but you’ve managed to pull together thousands of datasets for financial modelers to use. Company data is coming directly from EDGAR filings. Economic data comes from a very wide array of sources. How did you do it? How do you ensure that all of the data is clean?
A. Thinknum wouldn’t have been possible about five years ago. In terms of technology, we have seen semantic technology such as Freebase and XBRL appear over the past couple of years. We are now at an inflection point where this technology is even more robust than what I had access to working at one of the biggest banks on Wall Street. We are incredibly excited to bring that technology to tackle the problem of having a robust updated database of the world’s economic data. It certainly won’t be easy, but it will definitely be worth it.
Q. Thinknum makes life easier for financial analysts by facilitating collaboration without requiring them to email complex spreadsheets back and forth. It eases version control headaches, it speeds simulation processing, and allows analysts to jumpstart their next models without having to begin from scratch.
Wall Street firms are not generally known for their concern for the hardships suffered by junior analysts. What is the pain point for the investment bank or hedge fund? What problems are you solving for the enterprise, as opposed to the financial analyst?
A. We continue to focus on financial analysts, who have been our advocates in driving adoption within the enterprise. Ultimately, a more productive analyst will mean a more profitable firm. It can be hard to convince managers at established firms to adopt new technology, but we are certainly seeing a lot of enthusiasm from the more innovative players. Financial firms need to manage intellectual property in a more systematic way than ad-hoc spreadsheets.
Q. Are you finding hedge funds and investment banks resistant to hosting their valuation models in the cloud?
A. We only get paid by users who need private workspaces, so no one cares more about keeping their models private more than we do. Our incentives are completely aligned with these users. Once they have their tech teams vet the security measures that we have put in place, they are very receptive to us.
Q. All of the resources available are currently free. But because many institutions consider their models proprietary and don’t want to share them, you’ve recently launched “private workspaces,” which let users keep their work private or restrict access to a specific group of analysts. At $200/month, no one has to go very far up the food chain to get a purchase approved. Can you discuss your pricing strategy? Did you test other price points? Did you look at enterprise as opposed to individual sales?
A. Our pricing model came about organically after seeing how our users were engaging with the platform. We got pushed by some of our most active users who were using Thinknum to read models from others but could not save anything for compliance reasons. Our business model is quite effective since it is simple and handles most cases. A few financial institutions have reached out about enterprise plans and we intend to roll that out in the coming months.
Q. What other ways of generating revenue do you have in mind?
A. We don’t have other monetization plans. Justin and I committed to keeping Thinknum free of ads. Our current business model is very much in line with our philosophy of democratizing financial analysis.
Q. What is your marketing strategy for Thinknum? How are you reaching analysts and how are you planning to drive adoption?
A. We have traded with hundreds of hedge funds while on Wall Street, so we tapped our network and pushed it through word of mouth. We do a lot of in-person demos. In January, we got featured by Jason Voss of the CFA Institute and that was really our big break.
Q. How many financial analysts are out there? How big is this market?
A. There are a million people who analyze financial markets professionally and pay for software to support their work. Yahoo finance has about 37 million active users so there is larger universe of non-professionals who are passionate about finance.
Q. After the investment banks and hedge funds, will you target analysts at mutual funds, PE firms and other asset managers? I think they’d be a logical market.
A. A couple of PE firms just joined the platform in the past couple of weeks. We are going after the most innovative/aggressive players first. Ultimately, anybody who analyzes markets for a living will benefit from Thinknum
Q. Beyond the asset management business, who do you expect your next adopters to be? They’d require different types of models, but what about economists?
A. Right now, we are laser focused on building software for people who need to decide whether a security’s price is high or low. That in itself is a huge mandate. As we gain traction, we certainly have an eye on the larger question of what computation would look like in a networked world. So I can see us expanding past security prices, but we don’t have a clear idea of what the next piece of the puzzle would be. We first need to win the battle we are in.
Q. Could Thinknum be used to crowdsource more accurate equity valuation models and more accurate stock valuation projections?
A. Absolutely. A big problem is when analysts do not share their models and methodologies it is impossible to see if their models have bugs or how their models can be improved. We see the problem where analysts keep recreating the wheel and very little progress is made in terms of the methodologies for analysis. We are excited about what we see analysts share their work on our platform, get feedback and iterate towards a better and more robust analysis.
Q. How do you ensure model quality?
A. We don’t. We allow anyone to create models. Remember, 98% of the software that is open sourced is not worth using; we think the same will happen for financial analysis. We are focused on getting more and more people to publish their work and let the best 2% of models bubble to the top naturally.
Q. What is the incentive for the smartest analysts to make their models public?
A. We have noticed the biggest reason why analysts share their models is the powerful inner motivation that they have to get better at their work. Analysts are drawn to get feedback and embrace best practices from the community of analysts out there. Of course, users have the need to keep some models private, but almost every analyst can get better by putting their ideas out there and engaging with the community. We will also play a role in the capital raising process where the users who generate the best content will gain mind share and win the mandates to allocate capital.
Q. Does StockTwits offer a model you might follow? What about Estimize?
A. StockTwits and Estimize were early movers in the social fintech space. We are building the platform that finance geeks or analytical data driven finance professionals would use; we do not expect to attract the casual finance crowd. We will gain scale by focusing on analytical content, and our servers are optimized for number crunching. We think Twitter, Yahoo Finance message boards, or StockTwits would continue to be the best medium for following quick financial news bits and opinions.