1
$\begingroup$

I generally buy and sell weekly as stock prices fluctuate with volatility. I am exploring the idea of using a machine learning algorithm to consider various economic conditions (inputs) and provide insight (output) into whether any given day is a safe day to buy, given market conditions, for a particular stock. Thus I'd like to determine the risk related to investing at any given time. I'm only trying create a model of current conditions; I'm not seeking to predict prices.

For example, it seems more likely that a market correction may occur (and it is bad to buy) when the S&P 500 is pretty high, VIX is low, and the dollar is high.

Similarly, it seems less likely a market correction may occur (and it is good to buy) when indexes are not extremely high, the VIX is relatively higher, and the dollar is relatively lower.

Here is a compiled list of possible factors an algorithm could use as input:

  • Fear/volatility index (VIX)
  • GDP or GDP growth
  • Current interest rate
  • Current unemployment rate
  • Current Federal funds interest rate
  • Employee pay
  • Housing market data
  • Gold prices
  • Dollar value
  • Time of year (e.g. September is historically a bad month to invest)
  • Oil price
  • Bond prices
  • Index growth
  • Time until next Fed meeting
  • Month price change %
  • Day price change %

...and other factors (suggestions welcome).

Any thoughts?

$\endgroup$
  • $\begingroup$ Considering that you're "not seeking to predict prices", it sure sounds like you are at least trying to predict the direction of the overall stock market, i.e., predicting prices. It's easy to make money in the stock market: buy low, sell high (if you short, then you sell high first, then buy low). The trick is determining when prices are low and high, which sounds like what you are trying to do, under the guise of "model of current conditions". $\endgroup$ – Mark L. Stone Sep 21 '16 at 1:39
  • $\begingroup$ Fair enough. The goal is to predict potential price movement based on overall conditions and any inputs available. A gradient scale from 0 to 1 (bad to good) is what I'm after. $\endgroup$ – Chad Johnson Sep 21 '16 at 1:42

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.