Finding a Function for Data from Curve Fitting with a Chi^2 Boundary

I have MATLAB's Curve Fitting Toolbox. I've used it to find coefficients for a function given data. I've used it to find fits for polynomials. I would now like to know if there is a way for MATLAB to find a function given data. I have some data that cannot be easily represented with a basic polynomial or other such function. I've read through several previous questions, and they make the extremely valid point that there is an infinite amount of functions that would fit some data. So, I'm wondering if there is a way for MATLAB to give me the functions that have the best chi^2 goodness of fit with the data (as in the line from the curve fit fits the data very well). Perhaps the best 3 or 5 or 10 functions; the number does not matter. Would the chi^2 be a sufficient boundary? If so, how would I accomplish such a feat?

Also, if this is easier to do using a different language or program (such as Python or Wolfram Alpha), I would appreciate it if you could direct me to where I could learn more.