How We Decoded Cryptos & Wall Street with AI

3 min read

Approximately one year ago we published the first article about YUCE-8 — an AI service hosted in Google Cloud that aims to forecast the most likely future of selected assets.

It all started with a student who asked his professor if he could forecast the price of Bitcoin with the help of AI. With his long term experience with capital markets the professor said, that these markets are unfair, too volatile and somewhat unpredictable. Especially big money and people with 6+ million followers can ruin your investment strategy.

70% of the market is machine-driven

Today’s trading market is an unfair game for most. I am not talking about the massive gains of some cryptocurrencies that made a few people very, very rich and others, even more so. I am talking about the daily trading market. The day-to-day trading market is driven by approx. 70% algorithmic trading making it a miracle when the average trader is more successful than a machine. With more and more of these algorithms being driven by AI, those odds fall even less in the day trader’s favor as if we have all brought a sword to a gunfight.

That is why we wanted to change those odds for everyone. Our vision is to give everyone access to AI driven insights into the market.

How Does it Work?

Now, let’s imagine it’s the May 31th, 7:59 UTC and #BITCOIN is at approximately $37,000 USD. It comes from $50,000 USD and social media is divided into 2 groups: one that believes it will be back at 50k USD soon. And the other group that believes BITCOIN will dump to 20k USD.

Now comes the big question: will it go up or down next?

YUCE-8 uses AI technology to analyse the situation. In case of Bitcoin its models are trained with 354,000,000 data points. Within seconds YUCE-8 analyses the current situation at 07:59 UTC / May 31st and suggests the next 13 4-hour highs and lows of Bitcoin. The result is the following picture:

The AI model suggests a maximum drop to $34,000 USD and a maximum raise to $40,000 USD within the next 48 hours. If you want to sell your BTC position than you it’s better to wait until June 1st. If you want to buy BTC you should do it before June 1st.

Now let’s see how this forecast matches the real BTC prices:

Here’s an other example with $SPY, the SPDR S&P500 ETF. On the 18th of May 2021 the forecasted scenario for the next 48 hours is a dip to a minimum of 407 USD followed by a bullish recovery to a maximum of 417 USD:

Let’s see how this price really developed over time: As forecasted $SPY dipped and recovered later. While the dip was deeper than forecasted the bullish recovery ended very close to the forecasted price range.

How Do the AI Algorithms Work?

While other AI algorithms focus on the time series the YUCE-8 algorithms follow a different route. YUCE-8 analyses a broad range of economic factors. And it analyses the behavior of different assets in certain situations. Like this one: In the evening of June, 23rd the NASDAQ100 was fighting with the $14,300 points barrier. Again the social community was divided into 2 groups: those who said it will crash and those who said it will break through. Both groups mainly use chart analysis techniques to underpin their forecast.

YUCE-8 chose a different approach to analyse this situation. The following bubbles express the bahavior of the underlying asset NASDAQ. Every specific behavior is one bubble: a yellow bubble is the current situation. A green bubble is the previous situation. And a red bubble is the future situation.

As you can see in the following picture the behavior of NASDAQ on June 23rd was similar to 7 other situations in the past. And in each of these sitations a new all time high came into place shortly afterwards. So the YUCE-8 forecast was very bullish.

In the morning of June 24th the NASDAQ futures signaled a new all time high as forecasted.

How Can You Support This Project?

We need 2 core components to be successful: a GPU powered cloud and a real time data feed with access to historical data. Both components become very, very expensive once they run 24/7.

We talked to investors and pitched that our algorithms are able to forecast the future prices of certain assets with an accuracy of 72% within a timeframe of 13 trading days. Since it’s an unfair market driven mainly by big money we want to make these results available to everyone.

You could devide the investors into 2 groups: Those who said it’s not interesting and does not address a mass market. And those, who asked us not to make it public but rather become big money by creating a hedge fund.

We decided to stick with our plan and provide the service to everyone. We used our own money to built the service from scratch and launch a website and run a beta. The beta was our litmus test. Do people really care about this? And in fact they did! Here are just 2 examples of the feedback we got:

I traded with help of YUCE-8 for the last a couple of days and I’m impressed. This AI prediction is a gem. Wow! It would be awesome if poor retail investors could outsmart, with the help of Yuce-8. What a super project!

Looks like you guys have a winning product here! I’ve pretty high hopes and really like the look of it!

The positive feedback and the financial support of paying members allow us to continue & improve the service. The social feedback and support of those who love the project help us to spread the message @ https://twitter.com/8_yuce

Torsten Langner creator of the YUCE-8 AI algorithms for capital markets

One Reply to “How We Decoded Cryptos & Wall Street with AI”

  1. Very interesting article
    You have done a great job. I think that in the future AI will most likely converge with neuroscience and philosophy: its thinking will become like a human.

Leave a Reply

Your email address will not be published. Required fields are marked *