传统的量化工具都是通过编程来实现，人工智能的出现，大大加强了编程代码的效率。
神经网络模型，深度学习模型，机器学习模型，都是对机器学习算法的改进。
<<<<<<< HEAD
我觉得没有那么神奇。比如商品，数据量并不大（量价，产业供需，宏观，新闻），因子也不多，传统的量化模型如果效果不好，机器学习也不一定好使。对股票可能效果会好点，个股多，因子也多，机器处理强一些。 ——高联军
You have no idea how many clients we get that describe an idea and then follow up with “can you put some AI in there” ? Like it’s an ingredient! Anyways…
I want to play devils advocate here and say that the use of AI in algorithmic trading is massively overblown. Especially, the cited statistic of “80% of trades are made by computers”.
The role of algorithmic trading began as a way to accelerate the rate at which trades can be placed, and eliminate a lot of the redundant procedures that are associated with making a trade - especially at the institutional level.
AI is not used on Wall Street as much as you think - not in the traditional sense. While there are elements of machine learning, and neural networks, supportive of risk management systems, those don’t fly on autopilot, and the real AI isn’t artificial at all.
You know what that is? Thats a plastic tube that would contain a trade order. It would be shot threw a series of pneumatic pipe systems to and from the trading floor. That’s what computers helped solve.
We are very far from true Artificial Intelligence, and if we come by it then it’s applications will be far greater than trading and finance. A real AI algorithm would have to be able to independently process, reason and create in the same capacity as the human brain. Is replicating human intelligence the most impactful application, and therefore the key goal, of intelligence technology for businesses?[1]
Drop the “artificial” from AI and replace it with “augmented”, that’s right, augmented intelligence is the new AI.
Augmented intelligence is the use of technology to supplement the human decision-making process. Most algorithms and trading systems that brand themselves as artificial intelligence, are probably just doing a more sophisticated and automated form of intelligence augmentation.
At Remi Technologies we build trading algorithms for clients, and what happens most of the time is that the client supplies us with the idea, otherwise known as “logic”, and we realise it into a working algorithmic strategy. I would say that about once every two months like clockwork we get a client that asks us to “make AI”. In fact this happens often enough that we added it as an issue in our FAQ.
Disclosure: I am the company’s public representative and although my opinions expressed here are based on my personal and professional experience, my business could benefit from you clicking on one of the links posted in the answer.
46c81940dea9782c6531115b9a4711313d86e89a