The launch of the cryptoAI platform Alphacat caught our attention. The Chinese startup describes itself as the “world’s first robot adviser marketplace focused on cryptocurrencies” and expects to be the world’s first platform providing various AI trading robots for users to offer investment advisory services.
If that sounds like a tall order, the team that’s filling it is packed with both Wall Street and academic experience and credentials. Heading that team is founder and CEO Bin Li, who spent 20+ years in finance after an academic career in theoretical physics that started in China in 1979.
Li was part of the first wave of scientists to join Wall Street, starting at Merrill Lynch where he and his colleagues built the model and systems for the fixed income market for derivatives. He went on to become an entrepreneur, running a few hedge funds with partners, one of which is now one of Hong Kong’s most successful financial web portals, AAStocks.com.
We talked to Li to find out more about what he hopes to achieve with Alphacat.
MarketBrains: What led you to launch Alphacat?
Bin Li: I came back to China four years ago, and made some angel investments in some companies, and found out that the hedge fund industry was booming.
My training, experience and also my expertise is in using artificial intelligence and computer systems and quantitative trading strategies to manage hedge funds, so I thought about a way to do this, focus on building an AI-based system to replace human traders, managers and fund managers, with one hedge fund.
I called the system Alphacat, cat stands for computer aided trading. That was about three years ago, and we got about $30 million R&D private equity investment and hired some of the best people in the field to develop the system.
A year later, we did a backtest and real money test, and the system can really automatically create algorithms to trade the Chinese stock market and futures market to make profits.
MB: I’ve seen some of your results from the whitepaper, what are the most updated figures?
BL: The digital currency hedge fund, after fees and profit share, for every dollar invested, you received 2.69 dollars. And in the past few months bitcoin dropped 60% but our fund made a little bit gain, so it protected capital well.
Source: Alphacat – since August, the fund value has increased three times its value. The hedge fund is a combination of BTC, ETH, BCH, NEO.
MB: What are your trading strategies right now?
BL: We have three models.
The first is based on artificial neural networks, it is a self-learning algorithm. We feed in the model, historical data, and market sentiments and time the system so that you can make a prediction of the cryptocurrencies going up or down for the next 24 hours.
And we create 500 such neural nets, such robots, and actively make the prediction. It’s a multi-dimensional highly non-linear system.
MB: And that’s specifically for digital currency?
BL: We succeeded in doing that for traditional financial instruments, but last year we applied it to digital currency and it worked even better.
Digital currencies are more predictable than stocks and fixed income products in the traditional financial world.
MB: And the digital currency strategy is going to be completely separated into its own fund?
BL: Yes, that’s right it’s not mixed with any other things so we have purely digital currency funds that only trade most liquid large-cap digital currencies, like bitcoin, ether, EOS, NEO, the top 20 most liquid.
MB: What are the other models?
BL: The second model we use is called a pattern recognition.
Years ago, I was awarded a US invention patent because I created a computer system that can look at financial data time series to figure out whether it’s showing textbook technical chart pattern.
And when certain patterns appear under certain market alignments we may have an advantage to buy it or to sell it.
The system is examining all of the cryptocurrencies trading data, if it recognizes one of our own patterns that represents opportunity, it will then make an alert and generate buy or sell signal. The results are weekly.
And the third model is market sentiment analysis, which combines big data and money flow to make a prediction of the next move of the contract.
One of the most important things is looking at the money flow data.
When I come in to buy the contract, digital currencies at the ask price or higher, there is money flowing into the digital currencies. If someone sells the coin at a bid price or lower that means money is flowing out.
So, if we record all the details, all the client sales details of the trade, we can find out whether money is flowing into or out of the contract.
And by looking at that we can make an educated guess of whether it is more likely to go up or down.
MB: And that’s what you’re making available on the Alphacat platform?
BL: It’s similar to the app store. Alphacat is a marketplace for apps. And then our trading algorithms, those robots we already have, are applications on that platform.
MB: So these three strategies are the robot advisors?
BL: Yes, or tools.
If a client opens a trading account and attaches our robot to it, eventually our robot will generate buy or sell signals for that account and make the trades, and there will be other tools and applications for optimal allocation risk analysis.
So, if you want to sell one cryptocurrency and buy another one, if you do it by hand it’s cumbersome. Our robots will be able to automatically do that for you.
In the beginning, we will put a few of such robots on the platform, but the platform is designed more like an app store.
Later on, other members, institutions or individuals, will be able to also offer their robots on Alphacat for other members to use, and will use the ACAT token as the currency on the platform, so membership fees, for users, for buying or selling of product.
That is the large picture: Alphacat itself is more like the app store. In order to make it work we have to provide the seed applications otherwise it is empty.
MB: I noticed you launched on the Kucoin exchange without an ICO – is that because of regulations?
BL: We first got all our investments from private investors, from qualified client investors and then we listed on exchange. If we do an ICO in China, it is against the rules, that is the main reason.
But also, we had very little exposure.
MB: Do you have users now?
BL: We do have users, for example, our hedge fund [digital currencies], and our traditional hedge fund [China A-shares, net value from December 5, 2016 launch to January 2018 is 1.22 or a 22% gain. By comparison net value of the two main indexes for the same time period is: SSE 500 at 0.96, CSI 300 at 1.17]
There are institutions here who are using our system to manage assets. And our robots now make a daily forecast for the major cryptocurrencies and major indexes.
Recently, we invited a few thousand people from our community to test the users’ product, so they receive an email early in the morning for forecast for bitcoin, EOS, and NEO. Soon we will open that up for everyone, our plan is to let everyone to get our forecasts from our robot.
The bitcoin forecast has already been opened up for part of ACAT holders and some institutions. The forecast for other cryptocurrencies will also be opened up soon, but we don’t have an exact time yet. It depends on our marketing plan.
We already have the model working for making real time predictions for most digital currencies, but in order to have better results the digital currency should have at least a few months of data of trading history, the longer the more accurate.
You just type in the name of your favourite digital currency, press a button, the system will show you the five-day forecast: most probable line, upper band, lower band.
And it’s a scientific forecast with a percentage of probability for what direction the asset will go in. That’s the other thing in the pipeline.
This interview is edited and condensed.