I'm a Tensorflow Newbie. I am trying to find weight of holdings given value ratios for each holding. Ex) Input has 1000 stocks and each has three values : A, B, C. Output is weight for each stock, adding up to 1 There are 10 such datasets, each for different date.
How do I set up appropriate shape and placeholders and prediction? Thinking to set up loss function and use optimizer to train but confusing. Any insights/references will be helpful to structure this. Thanks.
