Picking Breeding Stock

How does one increase the odds of getting a nice puppy from a breeding? Is going to a famous or top-name kennel sufficient? What about the small breeder who only breeds a litter every year or two? Would they be a better bet?
When two dogs are bred, the laws of genetics take over in all cases and will determine the quality of dogs produced. Two dogs of so-so genetics will produce only so-so dogs ON AVERAGE at best. Every once in awhile, even so-so dogs can produce a beautiful, show-winning specimen. But it will be rare, if ever. And it will likely not breed true.
When choosing a breeder one has to look beyond size and even reputation. Look at the dogs themselves. A big breeder whose emphasis is on breeding top show dogs may produce many puppies by a chosen sire using several bitches. If they take the top dog or dogs and show them, their reputaion as a "top breeder" based upon show results will likely remain intact or even be enhanced. But these litters are still governed by the statistics of genetic law. There will be a standard distribution (bell curve) about some mean or average quality level. That level is set primarily by the parents' levels of quality. There will be a few dogs of better quality, many about the same as their parents and a few that are worse in some respects. The advantage a big breeder has is that they get to see more of the distribution faster than a small breeder simply because the big breeder whelps more puppies every year. But the fact remains that the distribution of quality in the puppies from either breeder will have some standard deviation about the mean quality, and the tightness of that deviation (or how similar ALL the puppies are) depends on what the parents are like. Two very different dogs (or dogs with vastly different positive and negative attributes) will have wide deviation and a wide range of quality among the puppies. That's why some litters have puppy siblings that are so different. Take two dogs that have many of the same strengths, and maybe one has a weakness that is not in the other, and you will get litters that are very strikingly similar among the siblings.
So why doesn't every breeder do this? Because sometimes they need to go fix a fault that is hard to correct or they see a dog that is popular and has some features they want. If they don't look AT THE WHOLE DOG when selecting a sire or dam to bring this feature into their line, they may end up with more disimilarities in the puppies than they planned upon. A big breeder maybe can "experiment" a little to gain a certain feature or look. They rigorously select from all the pups and keep the best of the bell curve distribution. But the rest of the puppies may be very different from that one special puppy the breeder keeps, especially if they are used as breeding stock themselves.
Rather look at the parents. And their pedigrees if you know the dogs in them. Then look at the puppies. Is the litter tight within itself on most features and faults? Or is there a wide disparity between puppies in the litter. Better a litter that is consistent than one with a single super-star puppy and the rest average. Especially if you plan on breeding one of the pups yourself.