Training for finish time and training to win are two different things; that is why Olympic chamions are rarely world record holders. Most of what you can find about training comes from coaches who think of state championships for high schoolers, national championships for college runners and the Olympics for elite runners - these schedules are all designed to peak for one specific race, which is not the way record setters train. The way to train for a personal best finish time is to get in the best possible condition and then do a bunch of races, waiting for the conditions to be right for a great race - this is very different from what most people end up doing. For example, I have marathoners choose the race they hope do do well in, plus another 4 weeks before it and a third four weeks after it; the first race is planned as a "dry run," which, if the conditions are right, can be turned into an all-out race [coincidentally, Lydiard also suggests this]. If neither the first race nor the second works out, there's still a last-chance third race. Most people find the cost of three marathons (two of which will be sacrificed) to be prohibitive, but elite runners simply make some calls to see who will give them free entry.
Planning one's race schedule involves a lot of guessing and I'm going to give a few guidelines that might help you sort through possibilities.
Guideline 1 ("Jack Foster's Rule"): For every mile you race, you need a minimum of one training day before your next race (and preferably two, long-term). For example, you can race the week after a 10K, but racing a 10K every week can only be done for a while before you need a break. You need four weeks after a marathon before you can race again, but you can race a marathon every 8 weeks for a surprisingly long time.
100 mile racers can only expect to race twice a year by this guideline and that makes an important point. This guideline implies all-out racing, going after every single second. One can enter races far more often if one uses them as just fast training runs and not all-out efforts.
Guideline 2: You can race 8-10% of your mileage short-term, 4-5% long-term. This guideline is the same as the first one, if you're running 70 miles per week. The more miles you run in training, the more racing you can do; this is the advantage of high mileage. If you're doing much less than 70 miles per week, this guideline means you'll be racing less frequently than the first guideline suggests.
Guideline 3: Though there is a wide variability in the mileage runners can tolerate, there seems to be a "sweet spot" for those running once a day, every day. Looking at what runners did before coaching advice was widely available, I found a remarkable consistency in the amount of time run each day, regardless of pace. For the most common distances, here's what I recommend:
1 Mile - 40 minutes/day
5K - 50 minutes
10K - 60 minutes
1/2 marathon - 65-70 minutes
marathon - 75 minutes
100 mile (trail) - 105 minutes
For racers who are moving from the early years of running as they feel and racing whenever they want to a more formal structure, this will probably suggest a huge increase in mileage, an increase that may take a long time to realize. There are runners who do well with much higher mileage than this suggests; they should be running more than seven times per week (if they use just the longest run done each day, the guideline still works fairly well).
How to use the guidelines
Look at your training in the past couple of months and find the average pace you've run or the pace you most commonly run. Using this pace, guideline #3 will tell you how many miles per week you should run. Then, knowing your mileage, you use guideline #2 to find out how many miles you can race between now and your goal race. Guideline #1 then tells you how to space those races out through the weeks.
I'm doing a trail 50K at the end of this April and doing "marathon" training. Given the climate here in Minnesota in the winter, I'm averaging 9 minute miles in good weather and hoping to run 8's in the spring. This puts me at 55-60 miles per week at the moment, aiming for 65. I have 14 weeks (I think) until the race, which is about 14x60=840 miles of training. 5% of 840 is 42 miles of races. Ideally, my schedule would then look something like: 5K on Feb.2, 10K on March 2, 15K on March 16, 1/2 marathon on March 30, 8K on April 13. Looking at what races are actually available (and getting sticker shock at some entry fees), I'll probably turn many of these into time-trials, i.e. simulated races on my own courses.
Four weeks before my race, there isn't a local marathon for me to use as a dry run; the best I can do is the Get Lucky 1/2 Marathon ($62?!) Four weeks after the Chippewa Moraine 50K is the Med City Marathon, which I'll consider as a back-up - though I'll undoubtedly miss the cut-off for sending in an entry.
1 day ago