Finding interesting investment goals before others is one of the biggest challenges that venture capital companies face. The good news is that machine learning and predictive analytics are gradually changing the way investors build portfolios.
“My job is to fly once a week to different European cities to find people who are engaged in interesting things.†Roberto Bonazinga, co-founder of IReachVentures, said that he had previously worked as a partner in British venture capital firm Baldeerton Capital, mainly investing in early Europe. Technology company.
â–³Roberto Bonanzinga
This practice is inefficient. "I see 50 companies every day, about 1,500 per month, and 100 of them may go to the next stage. We will make a deal every month."
Bonanzinga said that, under normal circumstances, prospective companies will not seek high-profile funds. Unless the founder's founder has a large network of contacts or is located in a technology center such as London or Silicon Valley, investors have little chance of discovering them.
If we combine Internet data and machine learning, can we better find a good company? InReachVentures spent two years and £5 million to develop this software, which can grab 95,000 European startups and select 2,000 companies that may be interested.
The decision-making of this software is based on factors such as the person being recruited by these startup companies, the products under development, and the website traffic. For example, InReach sees the Lithuanian startup Oberlo as an investment goal because it is advertising to engineers to solve certain e-commerce problems. Bonanzinga said: "We completed a deal before other venture capital companies in Europe knew about them."
“The work that was originally required to be done by hand can already be scaled up significantly,†says Bonazinga. “The efficiency has increased 10 times.†So far, InReach Ventures has invested in 7 companies, although it is too early to say that the performance of the portfolio is too early. Bonanzinga has already withdrawn from the transaction. Just 12 months after investing, he sold Oberlo to the Canadian e-commerce company Shoipify.
△ North America leads the AI ​​industry
San Francisco-based SignalFire was one of the first venture capital companies to turn to this data-driven model. Its founder, Chris Farmer, began using data patterns in venture capital in about 2007. He used basic algorithms in previous effective companies to track various factors such as product performance in Apple AppStore.
He hopes to create a more complex system to track the company in a more comprehensive way, so in the year founded SignalFire in 2013. He spent 8 years and spent tens of millions of dollars building a "mini Google." The software currently tracks 8 million startups globally. Data sources include sales data, academic publications, and financial statements. If a company performs well or engages in interesting projects, it will be marked on the control panel. Then SignalFire can deploy the $375 million it manages.
â–³ChrisFarmer
Ten years ago, Farmer said that this project was impossible to achieve. "At the time there was no such data and processing power. We needed only large consumer internet companies for computing power and storage capacity." But now, thanks to the advent of database tools such as Hadoop and Apache Spark, plus the possibility of renting cheap from AWS Servers, so even smaller companies can process data on a large scale.
Like Bonangzinga, Farmer also said that the system is helping him to dig out companies that he could not mine.
“Our portfolio comes from a wider geographic scope. We supported a company from Romania. If it was in the past, we wouldn’t find the company at all. We’ve been more than just a connected entrepreneur. Some people who started their own business deal with it. Although it does not completely eliminate prejudice, it does make it more like an elite rule, letting you take another look,†he said. However, it is still too early to judge the performance of the portfolio.
Aaron Joyce, a co-founder of Stockholm-based startup AiblTech, benefited from the data-driven approach. Aibl helped the company analyze customer data and they were discovered by Bonazinga just a few months after their establishment.
“We have already begun to finance some local investors, but we are not very lucky and can only chat with juniors. After that, Bonazinga sent us an e-mail. I was initially very skeptical that there was a venture capital partner who contacted me. I never thought we could have the opportunity to raise money through people like him,†Joyce said. “This is a more elitist form of investment. The key is no longer to whomever you know, nor to where you graduated. â€
Andreas Thorstensson, a partner at EQTVentures in Stockholm, said that he is currently investing about 30% of his investment decisions through a data analysis platform called Motherbrain. The platform monitors about 2 million companies daily. Mr. Thorstensson stated that he often invests in entrepreneurs who may have had funds before. "The data won't lie," he said.
In addition to providing opportunities for different startups, machine learning may also change the structure of the venture capital industry. SignalFire's structure is like a technology company - data scientists and engineers are critical to the company's business and hold company shares. "We don't treat them like IT. They are the core of the company," Farmer said.
At the same time, Bonazinga is the only investment partner at InReach, which has a team of software and data scientists who are responsible for replacing traditional venture capital company employees.
The cost of running and maintaining a data platform is quite high. Farmer said that he spends more than $10 million a year, and Bonangzinga plans to spend at least £1 million a year. "This will change the way venture capital companies spend. Most of the expenses of traditional venture capital are spent on wages. The future will be used to hire data and computer scientists, as well as to purchase excellent data sources." Bonangzinga said.
However, EQT's Thorstensson said he does not believe that machines will significantly reduce job opportunities across the industry. He said: "This will allow us to spend more time with the companies we invest in, rather than do ordinary things and do tedious work. Artificial intelligence is a good way to filter noise, but the investment still depends on intuition."
JUUL VAPE PEN, Wholesale Juul , Juul manufacture
Shenzhen Xcool Vapor Technology Co.,Ltd , https://www.szxcoolvape.com