Home Community Crypto-ML Roadmap, Feature, and Suggestions Update: Models are training on the latest data to help avoid these recent drops

Viewing 9 reply threads
  • Author
    Posts
    • #8061
      admin
      Keymaster

      As an update, we have seen repeated behavior in this 6000-7000 channel whereby it appears a breakout may occur, but we instead see a sharp drop. The ML is currently learning from this and updates are expected to be released later today.

    • #8149
      matthew55
      Participant

      Hello,
      I’ve noticed we now have a “Signal Management” table within the Member Area.

      https://p187.p2.n0.cdn.getcloudapp.com/items/xQug2mmy/Screenshot%202020-04-01%20at%2008.23.25.jpg?v=241f7cd049b29e04e8e63ee0e22f050e

      Can you please explain what “Precision” means, and how it’s calculated?

      I guess the closer it gets to 1.00 the better, but would like to know what we can expect from it every, let’s say, 0.05 increases

      Thank you

      PS: Nothing to do with this specific question, but would be nice if we could format our posts with BBCodes or HTML via a specific interface, instead of having to manually write our own codes 🙂

    • #8153
      martyf
      Participant

      c’mon machine, when you gonna pull outta this long….. that 7300 was nice…. …. hrmmmmm

    • #8154
      kw
      Participant

      Would anyone else be interested in a Telegram group or similar? I’d be interested in a more real-time discussion. We could always start a group, but Justin may prefer to own it.

    • #8156
      Justin
      Moderator

      Hi everyone,


      @matthew55
      –yes, I think we can enable BBCodes or something similar. Will look into it. As for the “precision,” that is described here: https://crypto-ml.com/blog/machine-learning-upgrade-to-5-0-deep-neural-networks/#Statistical_Measures

      And generally yes, the closer to 1 the better.

      @kw –agreed on a Telegram group. I will get that going.


      @martyf
      –yes, $7,300 was extremely nice and I was hoping for the same. The predictions seemed to generally stall out (go flat or slightly dip) but didn’t flip hard negative enough to cause a sell. They’re sitll calling for a little upward movement. We’re monitoring the behavior very closely as I am expecting better closures.

    • #8158
      Justin
      Moderator

      Alright, Crypto-ML Telegram channel is in place.

      Despite being in the Crypto space, I’m not much of a Telegram (or any social media) user. So if anyone has recommendations, let us know.

      https://t.me/cryptoml

      I think an ideal use for this is to discuss manual closes (even partial) on trades that have been great but may be now exhausted. We of course want the machine to do this for us, but potentially with our collective brain power, we can identify areas of opportunity.

    • #8162
      LeoTheLion
      Participant

      Regarding the updates/learning, would it be possible for the precision of the models to be updated daily? I imagine that this is a computation that is being constantly updated. This would help users make more educated trades.

    • #8170
      kw
      Participant

      Thanks Justin for creating the telegram channel! It appears as if it’s read-only however, I do not seem to have the ability to write a message even though I joined the group.

    • #8173
      denome
      Participant

      Oh yeah it’s read-only for sure. I joined it and I was just expecting some announcement.

    • #8276
      Justin
      Moderator

      Dang, sorry about that…my freshness to Telegram coming through.

      Discussions need to happen in a Group rather than a Channel.

      So here’s the link to the Group: https://t.me/crypto_ml_chat

      Thanks for your patience!

Viewing 9 reply threads
  • You must be logged in to reply to this topic.