I think it’s difficult to give a one-size-fits all generic answer to this question. Deciding when to file a patent application is always going to involve a balancing of cost, risk and potential reward.
My impression is that having a patent application pending is unlikely to make or break an angel investment round. While it may be more helpful in later rounds, I think it’s unwise to file a patent application strictly for the purpose of attracting investors. Most won’t be impressed by a “patent pending” label, and at least some of those who are will be smart enough to see through a weakly crafted patent application. Do it well or don’t do it all.
Doing it well requires filing a patent application after it’s too early, but before it’s too late. I’ve seen many patent applications that were filed “too early” as well as many that were filed “too late.”
The ones that were filed “too early” generally suffered either from a too-cursory description and drawings. This is usually a result of efforts to save costs. An insufficient description typically winds up causing problems when it becomes necessary to narrow claims during prosecution (which happens in almost all cases). If there isn’t enough detail, drawings and written description support to make valuable narrowing amendments, then a patent application can wind up stuck between having disclosed the basic gist of the invention but not having enough detail to get the claims allowed. By the time this is discovered, it’s usually too late to do much about it.
In some cases, patent applications get filed “too early” in the sense that they are filed before the inventor really understands what features of the invention are really valuable. Since invention and product development are inherently iterative processes, it can be easy to overlook the truly important elements of an invention when it hasn’t been tested very much. This is the main reason that I advocate liberal use of provisional patent applications. A year is usually enough time to get through at least a one or two prototype iterations, typically enough that at least the outlines of the important elements become apparent by the time the provisional anniversary arrives. But to avoid the problems outlined in the paragraph above, the provisional also needs to be as thorough as possible at the time that it is filed.
Patent applications that are filed “too late” are generally the ones that get filed after the inventor has created prior art against himself, or the ones where someone else creates prior art or files their own patent application for the same or similar invention. While the former is certainly within the control of the well-informed inventor, the latter is entirely unpredictable. It’s the sort of thing that is only completely apparent in hindsight. Unless you know a lot of details about someone else working on the same problem, it’s very difficult to anticipate what prior art someone else might create or when (or if) they might file their own patent application. The risk of this is actually pretty low in most cases, but because it’s impossible to quantify, it appears to be a greater risk.
So, my general suggestion is the same as always: file a provisional patent application when you are able to adequately describe the invention and how it will work, and then spend the next year testing the assumptions built into your provisional application. Whether or not you need angel funding to get that done is something only you can answer.
Based on my answer on Quora to the question: