Math/statistical challange: stock price prediction algorithm

  • Status: Closed
  • Prize: $300
  • Entries Received: 4
  • Winner: kannasanthosh

Contest Brief

NOTE: this math/statistical challange requires programming or advanced Excel skills
_______

INTRODUCTION:

As most people know, when a bad news hits a company the stock price of that company falls and the inverse happens if it's a good news.

In the attached two images I show you three recent real-world examples of what happened to the stock price and volumes of NVidia and Amazon when bad news hit them.

As you can see there is a sudden increase of traded volume (bars on the bottom) and the price falls because people sell in panic, both things happening for some tens of minutes, allowing an investor able to act immediately (let's say after just 1-2 minutes of seeing the sudden change in volume and price) to earn money for that fall predicting it, with a financial operation called "short selling" (earn while the price falls), closing the position when the fall is over.
Similar thing can be done in the opposite direction when there's a good news: the investor can immediately buy the stock, selling when the rise finishes, earning money for the rise...
_____

REQUEST:

What I ask you is to find the best algorithm for earning the most money by automatically detecting the best moment to start a position ("short sell" or buy) and the best moment to close it, applied to the attached 12MB big file that contains the real price and volumes of the Amazon stock with 1 minute resolution of last 10 years, for your performance tests.

The algorithm that, applied to this data set, virtually earns the most money, wins my award (and you can use it like me for earning money on the stock market of course;)

A very simple example algorithm could be this:
- start a position immediately after there's a sudden 4x increase of volume together with 0.50% change in price ("short sell" if -0.50% or buy if +0.50%)
- close the position 1 hour later or when the price reverts back of half of the total reached change, the sooner

RULES:
- to calculate the earnings of your algorithm, you must start the calculation with a virtual investment of 10.000$ and subtract a virtual (like real world) commission equal to 0.06% price + 10$ fee every time you virtually start and every time you virtually close a position
- the algorithm can use only the data provided by the 12MB file, that are prices and volumes. Its actions must be only based upon prices and volumes (and time).
- you cannot cheat by overfitting code to specific single patterns in the provided data set. The algorithm must make sense as general rule to predict the future cases, though it can be complex

NOTE:
the headers of the 12MB txt file are these:
date time open(first price) highest lowest close(last price) volume

Recommended Skills

Winner
Featured Contest
Post a Contest Like This

Top entries from this contest

View More Entries

Public Clarification Board

  • visweswaran1998
    visweswaran1998
    • 6 years ago

    Could you explain briefly how the Machine Learning Model failed please, It has predicted the future values of your data set

    • 6 years ago
    1. interservices
      Contest Holder
      • 6 years ago

      If I remember well, you took multiple columns as volumes, while there is only only with volume as described in this clarification board.

      • 6 years ago
    2. interservices
      Contest Holder
      • 6 years ago

      If I remember well, you took multiple columns as volumes, while there is only one column (the last) with volume, as described in this clarification board.
      Furthermore, please clarity your return over 10.000$ initial investment, considering the specified costs on every buy and sell

      • 6 years ago
  • kannasanthosh
    kannasanthosh
    • 6 years ago

    Hi mate.. Please make this a sealed contest. I am ready with my entry. Do let me know if I should submit as it is.

    • 6 years ago
  • visweswaran1998
    visweswaran1998
    • 6 years ago

    See Entry 2 which has 100% of prediction

    • 6 years ago
    1. interservices
      Contest Holder
      • 6 years ago

      Hi Visweswaran!
      as wrote yesterday in this public clarification board, the headers of the 12MB file are:
      date time open highest lowest close volume

      where...
      open is the price of the first trade of the minute (at the beginning)
      highest is the highest price traded during the minute
      lowest is the lowest price traded during the minute
      close is the price of the last trade of the minute

      • 6 years ago
  • lourdesreynaldo
    lourdesreynaldo
    • 6 years ago

    Good evening, could you detail what's in each column? The first is a date, the second is a time, but I do not know if the next four are prices, and the last volume? Thank you.

    • 6 years ago
    View 1 more message Processing...
    1. interservices
      Contest Holder
      • 6 years ago

      open is the price of the first trade of the minute (at the beginning)
      highest is the highest price traded during the minute
      lowest is the lowest price traded during the minute
      close is the price of the last trade of the minute

      • 6 years ago
    2. lourdesreynaldo
      lourdesreynaldo
      • 6 years ago

      Thank you.! I start working on the project.

      • 6 years ago

Show more comments

How to get started with contests

  • Post your contest

    Post Your Contest Quick and easy

  • Get tons of entries

    Get Tons of Entries From around the world

  • Award the best entry

    Award the best entry Download the files - Easy!

Post a Contest Now or Join us Today!