Has anyone experience with technical chart analyses? The only method I found useful so far is the moving average.

I'm trying to create a trading bot and he is already fetching data each minute. Now (after a few days) the time has come to create some kind of algorithm to decide when to buy/sell.

  • 4
    That's a broad topic. When I can tell you is, do not let the bot execute buying decisions for you that you didn't program specifically. Have it put the information in front of you and aggressively alert you. You also want to keep running aggregates of trend data that you base your algos on. Everything folds back into itself periodically.

    In case you're not familiar, this is heading in the direction of high frequency trading (HFT), which is an excellent way to go broke if you're not involving a human. The big players have servers located in the same DC as the exchange servers with paid priority access to trade data. They base trades on real time transactions and in many cases make millisecond trade decisions. They also burned half a trillion dollars in 2008.

  • 2
    The problem with trading bots is that Blackrock already exists and their bots get updates way more often and faster than yours.

    Also, chart analysis is bullshit. Analyze news and reports instead.
  • 0
    @SortOfTested thank you for your reply.

    The bot won't be a black box. Each decision should be reasonable for 100% so no ml or ai will be involved. Further, I will run the bot in "demo mode" with virtual money at first to check if it works like it was planned.

    Also its more a fun project so even if it works good enough the bot will not get more than 50 or 100€.

    I know about the points you mentioned and I know that stock exchange is something you cannot really calculate since humans are involved and thay do not do rational decisions sometimes. Nevertheless I think the idea to have an automated trading bot is really interesting.
  • 0
    @Oktokolo you are absolutely right. Since my goal is not to make billions of dollars but maybe profit at all that's OK for me. It's not about beating one of the largest companies in the world but about leaning to make decisions on rapidly changing data.

    Do you have a good idea how news can be analyses? I'm thinking about rss feeds but there is the problem that text that is written for humans needs to be interpreted.
  • 0
    Well, human language processing is a huge (and mostly unsolved) research field.
    But you could start with parsing the financial statements of the corporations.
  • 1
    Moving average is a basic one, as you already mentioned. I usually also use the stochastic oscillator, rsi, and bollinger bands.
Add Comment