How are neural networks written for financial exchanges?

Question

Fintech

There was an idea to feed the neural network with high-quality historical csv data. I know that there is a metatrader 4 program for forex. But I don't like Forex because of the lack of reality. Any slippage where they should not be.
Therefore, I want to teach the network to analyze different cryptocurrencies.
1. Which (real) exchanges offer trading bots?
2. On what libraries is it all written?
3. Where can I find good historical data?
4. *How to write a network?
P.S. I know the basics, but I don't want to reinvent the wheel. Maybe some tensorflow to use?

Answer 1

successful solutions of this kind, no one publishes. Well, or only after they stop working;)

Answer 2

1) for everyone, adjusted for exchange rate (a sharp drop in the exchange rate - like everything hung up)
2) https://www.quantlib.org/
3) for the crypt? 3-4 sources, the same exchanges, some give, well, coinmarketcap
4) on Python

Answer 3

https://numer.ai/learn
1. bots are not available everywhere,
but no one will forbid you to use MT4 for analytics, but open trades with a broker.
if you are not doing it for yourself, then creating an adviser that will give signals about entry points,
such a solution will be more flexible and convenient for further sales of the adviser to traders (beginners),
such a solution can be user-friendly)) even for those who work with binaries.
Again, I'm talking about the stage of monetization of the National Assembly through the sale to traders.