I recently read "The Myth of the Rational Market" by Justin Fox. This book is largely a history of the development of academic models of the stock market. I found it interesting although perhaps a bit long. I was already familiar with these models in general terms which probably made the book easier to follow.
As the book notes models are simplifications of reality. As such they are never completely correct but may nevertheless be useful. However you must be aware of their limitations. Unfortunately financial modelers often have excessive faith in their models and this can lead to problems. One common problem is failure to properly take tail risk into account. Tail risk is the risk from rare events. Because the events are rare they are easy to ignore but this can be dangerous as one bad day can wipe out several profitable years. This is especially a problem with new financial products that don't have a big experience base to reveal problems.
Another problem is a preference for mathematically elegant models even when they don't match reality. For example modeling stock prices is easier if you assume they move continuously but real stock prices sometimes jump discontinuously. A continuous model can still be useful but you shouldn't forget that stock prices don't always move continuously. Unfortunately academic economists have a tendency to fall in love with their models and come up with elaborate explanations for why apparent discrepancies with the real world are illusory. As with strained claims that real world bubbles don't exist because their models don't predict them.
The recent financial crisis has left financial modelers somewhat discredited. Although as the book observes the problems were primarily outside the stock market which has been the primary focus of academic work. I don't think the answer is to throw out models entirely but to be more cautious in relying on them. Particularly when they are being used to sell new financial products.