Yesterday, a company claiming to be the first of its kind in the financial industry was launched. Gstock is "a technical analysis-based website for private and institutional investors." It harnesses volunteer computing power, similar to the way SETI does, to become, according to its claim, "the world's first-ever virtual supercomputer dedicated to stock picking."
I talked to Oren Rossen, co-founder of Gstock along with Nir Ben Levy, and found a fascinating entrepreneurial spirit, much in the manner of early Google days.
Melly: Oren, what is your background?
Oren: After working as an analyst and being the treasurer of the MIT Israeli chapter, I became a consultant to companies that could see through the technology but had no business perspective.
Melly: Tell me how the idea for Gstock came about and how you developed it?
Oren: About two years ago, I received a call from someone in the business department of the Technion, the #1 technical university in Israel, with an interesting question: "We have a donor who is extremely wealthy and who believes that the next big thing is supercomputing based on grid or distributed computing. He is willing to pore large sums of cash on a project that would both have a technological edge and be commercially viable. However, we don't have a clue what can be done commercially with so much computing power."
I was familiar with non-profit ventures on the Internet that use distributed computing networks such as SETI@home or protein and DNA mapping, but there was no valid business based on distributed computing.
I discussed it long with Nir and one day it just popped out. It was very ambitious, but the idea was to gather huge computing power and give the private investor an edge on any fund manager to basically win the market. If until now private investors had to compete against giants and professionals (who have better resources), then we can now equip this regular private investor with a supercomputer, and not only that, he would also enjoy the flexibility that professionals don't have building his portfolio in a very dynamic manner.
Melly: How is it you claim Gstock can help the individual investor?
Oren: Until now, the tools available for private investors have been a joke. Financial websites give investors technical tools, but they don't give them any indication if those indicators worked well with a certain stock in the past. For example, they give them ratios, but they don't tell them how that ratio performed in accordance with the stock price over time. In other words - private investors get all the information they could ask for, but no knowledge what so ever. So with loads of information on any stock, a normal user still can't handle the basic question - should I buy or sell this stock? Gstock, being the investor's personal supercomputer in this chess game will give the guidance to the bottom line - profits.
Melly: You're not the first to claim that, how does Gstock propose to do that?
Oren: The project started as an "out-of-the-box" thinking. The financial world is built on stereotypes of how things should be done, what works and what doesn't. When we started, we had 3 objectives:
1. We will build a flexible system that will be able to test different angles and in different ways, which are sometimes beyond the consensus.
2. We will gain enormous computing power that will enable us to test the wildest things.
3. The system must be systematically very profitable, not some lucky trade or on a specific segment, but across the board.
To explain the first point further, most common practices are about finding the one "divine" strategy that fits a large array of assets. We said, "Why only one? Let's find what best fits each asset. Why limit ourselves when we have infinite computing power?" So instead of using a 14-day average simply because its inventor, Wilder, said (almost 40 years ago) that he thinks 14-day average is the best one to use, why not test other parameters also?
This was the first leg, the second leg which we still have to add is econometrics, such as which stock moves in correlation with which stock and at which time lag. The third leg, I prefer not to speak about yet because it is completely innovative and is part of the patent we filed.
Melly: How did you build the enormous computing power you speak of?
Oren: That took a while. We started by gathering loads of old PCs in my flat and connected all of them. Then in a second flat... basically, just like Google did in their first days when they needed computing power. It was great for the start, but soon after the "monster" began to ask for food. If we wanted to reach 1 billion backtests on thousands of stocks, each test running on 10 years of data, we needed much more. We turned to all our friends, and managed to gather several hundreds PCs, which was really great but still not enough. So we turned to the Technion. It turned out that they had a system that wasn't deployed because they never had any application for it. They let us use it. This is basically how it was built.
We asked a few tech guys previously involved with ICQ and Vocaltech to help us code the system and build the unique web 2.0 GUI, which has moving windows and on-the-fly quote refresh within the stock pages. We asked Danny Peretz to guide and consult with the trading algorithms, Tal Shwartz to guide and consult on the testing of the system, and Gil Tiberger to define patents of the innovative things we found.
Melly: Do you trade on your own system?
Oren: No we don't. All our friends and relatives do, but no-one that has direct access to the algorithms and strategies. Either we give service or we trade, but not both, since we don't want to be in the slightest suspicion of front-running or any other personal gain fraud.
Melly: Right now most of your services are free, what's your economic model?
Oren: We have a subscription package and it costs #39.95 per month. What it does is send a daily email to the subscriber and also enables the subscriber to use the Gstock stock picker which is open to subscribers only.
Beyond that, we are going also into the "institutional trading" area by providing hedge-funds co-management services of their portfolio based on our alerts. In return, we will get 50% of the "success fee" which is normally about 20% of the gains.
Melly: Where do you go from here?
Oren: On the business side, we plan to develop now towards services for hedge-funds by providing them co-management services on their portfolios. On the technological side, Gstock is already the most ambitious tool ever built for the use of private investors but it will not stop here. Our plans go much beyond towards adding econometric analysis per stock, and other high computational elements that will increase the system's success ratio from its current already very high 70% to new heights. This would mean that users will have the confidence that nearly each of their trades is bound to produce profits.
Melly: What do you see in Gstock future?
Oren: Gstock will grow in coverage to all stock markets and financial instruments. In time, it will become an immense supercomputer moving the markets and constantly adjusting its strategies in order to fit to the changing markets of its creation. As a result, the duration of the strategies will become shorter and they will change more often. The stock market is considered the largest casino in the world, and having such a supercomputer to guide investors and increase profits from this huge casino will have a big impact on financial markets as we know them today and the way we handle our money.