RGB Sensor’s New Job: Cryptocurrency Trade Advisor

[XenonJohn] dabbles in cryptocurrency trading, and when he saw an opportunity to buy an RGB color sensor, his immediate thought — which he admitted to us would probably not be the immediate thought of most normal people — was that he could point it to his laptop screen and have it analyze the ratio of green (buy) orders to red (sell) orders being made for crypto trading. In theory, if at a given moment there are more people looking to buy than there are people looking to sell, the value of a commodity could be expected to go up slightly in the short-term. The reverse is true if a lot of sell orders coming in relative to buy orders. Having this information and possibly acting on it could be useful, but then again it might not. Either way, as far as out-of-left-field project ideas go, promoting an RGB color sensor to Cryptocurrency Trading Advisor is a pretty good one.

Since the RGB sensor only sees what is directly in front of it, [XenonJohn] assembled a sort of simple light guide. By enclosing the area of the screen that contains orders in foil-lined cardboard, the sensor can get a general approximation of the amount of red (sell orders) versus green (buy orders). The data gets read by an Arduino which does a simple analysis and sends alerts when a threshold is crossed. He dubbed it the Crypto-Eye, and a video demo is embedded below.

Could this have been done purely in software? Certainly, but there’s a certain charm to the Crypto-Eye being a standalone device that uses a simple visual input to make buy and sell predictions like a Speak & Spell.

Inventive crypto trading is just a side project for [XenonJohn], he’s better known around these parts for his outstanding contributions to one-wheeled electric vehicles, like this 3000W Electric Unicycle, which also happens to feature an Arduino with 80’s-style voice feedback, just like the Crypto-Eye.

15 thoughts on “RGB Sensor’s New Job: Cryptocurrency Trade Advisor

    1. Without patching into the data stream?
      As in, can it watch/read the graphic display by colour, using the existing program’s display, instead of having to hack into the data stream or subscribe to a data stream?

      1. They provide an api for Pele who develop software tools. Trading bots are the most common example of this. Reading into the buy and sell order history does not give the clearest representation of which direction the trading is going though. The point is to create a proactive trading strategy, not a reactive. This is a cool tool but is basically about as accurate as asking a magic 8-ball what to do and averaging the response till it makes sense.

        1. By ‘they’ do you mean that particular data/application source, or that others provide an API?
          I notice that the screen says “login or register to continue”, so this may be a way to have something watch a screen/monitor for you.
          API is definitely the way to go for anything serious (I stream and generate Level 1 for 24 securities, and trade four with 2ms to 6ms algo turnaround, with extensive multi-threading with Java on a hex core), but I’m curious if ‘watching’ a monitor (graphic vs. a data stream) can be done without a camera.

          (And then there’s the issue of is the graphic source live or 20 minute old data, then tick or aggregate or other representative data.
          And response time: by the time it’s said its result, a person hears it, then decides, then responds, they’re so far behind the current price…
          We’ll ignore the wisdom of his algo for now – lol – but he may be using that as entry confirmation or timing.
          Wonder what he’ll do when he notices crossing indicators…)

      2. Somewhat unrelated, but I was looking into something that would watch my screen and make decisions based on on-screen displays, perhaps with ocr, but I couldn’t find any good framework for this. All the OpenCV, etc, stuff seems to demand a camera source, not working with simple screen capturing. Any advice, y’all?

Leave a Reply to ChrisCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.