I've mentioned before that the reason I am an all-grain brewer is because the home cook in me wanted to take the leap from brewing with kits to writing my own recipes. This is the approach I take in recipe formulation. I should also say, I'm focusing this post on writing a recipe to match the BJCP style guidelines. Not everyone brews to style and that's fine. It is, however, important to have a base from which to work. Like the best Jazz musicians, it's important to learn the rules before you go about breaking them.
When I approach the formulation of a new recipe, I begin by evaluating the style I want to explore, usually with the BJCP Style Guidelines app on my phone or tablet. Once I've understood the parameters of the style, I then seek out commercial examples of the style using the list at the bottom of each style profile. I evaluate them against the characteristics outlined and make notes. When writing your own recipes; the only constraint is your own personal palate. Being a great homebrewer requires you to train your palate and objectively evaluate commercial and homebrewed examples in order to apply that knowledge into making your own. In other words, tasting great beer is essential to making great beer. Find the best resource for commercial examples near you and taste as many examples as possible.
Next, I will look for "clone" homebrew recipes of the commercial beers that I most want to emulate. I usually consult the books Clone Brews, Brewing Classic Styles, as well as many of the online forums out there such as Beer Advocate and Home Brew Talk. Sometimes you can even go directly to the source and have your questions answered from the brewer who made the commercial example. The brewing community can be remarkably open and welcoming in that sense. It can't hurt to ask, the worst that could happen is that the brewer declines to share his or her recipe.
Once I've compiled enough data, I compare and contrast ingredients both common and differing. Identifying common ingredients can open up your creativity. Invention and creativity are essential to making great beer personal. Ironically, I use a spreadsheet for this, noting when clone recipes have the same ingredients in common. While I'm spreadsheeting, I am still tasting those commercial examples so I can determine what I like about the beer and identify what flavors I want to ultimately have in my own beer. I ask myself If I can taste that caramel flavor in the caramel 40? Would I prefer more? Less? Is it piney from the Simcoe or is there perhaps another hop in there that's been misidentified? What flavor comes through in the yeast? Is it neutral or does it impart its own unique characteristic and is that the best yeast for this particular style? How clear is this beer? Which head retention best serves to highlight the flavor of the beer? And which do I prefer when objectively evaluating each beer?
Now that I have a base to work from comes the fun part; writing the recipe. I identify how I want this beer to look, taste, smell, and feel. And it's with that in mind that I set about to tailor the beer to my interpretation of the given style. I could bump up the amount of a specialty malt, change a hop or hop schedule, dry it out with dextrose, the sky is the limit. Part 2 of Recipe Formulation will be about finding that balance in your creativity to produce a beer that meets your own expectations. For example, I once made a beer with a Stone malt profile, a Bell's hop profile, which I then fermented with Conan yeast harvested from a can of Heady Topper and propagated on my stir plate to the appropriate pitching rate. Remembering that beer now, I feel a strong desire to make it again.
Do your research and brew aided by collected knowledge. Have a viewpoint and use that sensory research when you write the recipe. Your beer will be better as a result, and isn't better beer the whole point?