Artificial Intelligence became almost an El Dorado for experts trying to make the predictive analysis. A decade ago, there was a huge excitement around this potential. But it was cut off when the results turned out to be inaccurate. 

A group of researchers from MIT released a study concerning the AI capability to forecast the prices of different assets like shares on futures markets and hedge funds. The study is called Prediction Markets, by David Shrier, Dhaval Adjodah, Weige Wu, Alex Pentland. 
The confidence in prediction markets dramatically decreased after the 2008 financial crisis. At 10 years after this moment, prediction models offer a new promise to beat the skepticism. 

What changed since 2008

Technology is the new guest at the table of prediction markets. Secondly, a better way of reading the innovation signals from AI technologies. Now we can appreciate and select better data and the model of machine learning and create patterns is far way better from what we had back in 2018. 
The neural network that actually functions similarly to the human brains is one of the most important attributes of prediction evolution that we acquired into the last years. The results of these improvements are that droves of data can now be filtered and selected without being processed by humans. Also, the speed of the process is incredibly reduced. Most of these operations are made in real-time. The newest strategies resulted from the AI methods apply immediately to market changes allowing an instant adaptation to new rules. That’s quite a change, isn’t it?

How AI is used by hedge funds

The hedge funds were some of the innovators in the AI field of prediction markets. They adopted the technology for experimenting the market forecasting. 
Two years ago, an AI hedge fund was launched meaning that all the trades are implemented by machines. The system does the trades using neural networks and probabilistic logic, with no human influence. The decisions of Aidyia are made on its own forecasting. There are also other similar examples as Sentient – AI platform and Numerai – hedge fund which distributes machines learning patterns for the masses. 

The setback

Creating patterns may lead to an outdated data content. Because we apply causalities and models that happened in the past. But now the markets evolve at the speed of light, so a more organic process is highly necessary.
Even though the routine correlations are made very fast, there is still a huge gap between a model and a cause. The only way to solve is to add some context attributed to those patterns.