Why are my evaluation stats giving weird results when training a neural network?

I am training an RNN which uses LSTM layers on stock data. Upon training and getting the evaluation stats, I get the same stats almost every time, which are very bad. Here is what my stats look like:

   Column    MSE            MAE            RMSE           RSE            PC             R^2            
col_0     2.14646e+04    9.68327e+01    1.46508e+02    1.77569e+00    NaN            -7.75692e-01 

First of all, is there any reason why the Pearson Correlation is Not a Number? Occasionally, I get a real number for that, but it is within .02 to .06. Secondly, is my data being messed up the reason that none of the other stats will change? I twist and tweak the hyperparameters of my model, but everything besides the PC stat stays EXACTLY the same. What can I do to fix this? Thanks for the help in advance.