When breaking bad news to someone, it's often a good idea to lead with a joke to lighten the mood. For instance, did you know that if you visit Ludwig van Beethoven's grave site, you can hear all of his symphonies playing in reverse? The great musician, it seems, is decomposing. (In case it's not obvious, that was the joke.)
Speaking of decomposing, (note: appreciate the smooth transition), if my leagues are at all indicative of the dynasty landscape as a whole, I'd say there's about a 33% chance you have noticeable problems with your roster's composition, leaving you with a team that is flabby and inefficient. (That was the bad news.)
But worry not, because roster composition problems are one of the simplest and quickest problems to address. In fact, I consider this one of the lowest-hanging fruits in dynasty. Within a few weeks, you should be able to take your bloated, lumbering roster, trim away the extra fat, and leave behind a sleek and nimble beauty. All for little-to-no cost.
What Do I Mean By Composition
All dynasty leagues allocate a certain number of roster spots to each team. For the most part, you are free to fill those spots however you choose. You could concievably devote your bench entirely to extra quarterbacks, for instance. (Please don't devote your bench entirely to extra quarterbacks.)
The breakdown of players by position is your team's composition. For instance, in one of my leagues with 24 roster spots, my team has 3 quarterbacks, 7 running backs, 8 wide receivers, 5 tight ends, and 1 defense. (Kickers? Pfaw. Dynasty leagues are better without them.)
It's easy to think that you should just roster the best players available, but that's really not the case. The quality of players available on waivers tends to be pretty close, so you should roster the best players available that will leave you with a reasonable overall composition.
The reason for this is two-fold. First, if all of the best available players are quarterbacks, (in your opinion), then that means the league doesn't value quarterbacks very highly; you can leave most of those passers on the street and grab them at your leisure in the future, effectively using the waiver wire as an extension of your roster.
Second, your bench spots are essentially developmental spots. They're the minor leagues, they're your farm system. The point of your bench is not to roster the players who will score the most points. It's to roster the players who have the best odds of one day cracking your starting lineup. (Secondary consideration can be given to rostering players based on current or anticipated trade value.)
After all, how much value do you get out of a player you never start? How much does that player contribute to you winning games? None and nothing.
What Are the Implications?
Let's look at byes, because they're the easiest way for backups to see your starting lineup. Over the past three years, the NFL has scheduled byes on seven, eight, and nine different weeks; for convenience sake, let's assume that there are eight bye weeks in a given year. (We'll also assume that byes are evenly distributed, with four teams on bye each week. This isn't the case, but it greatly simplifies the math.)
Obviously you need at least one backup at each position to cover for byes. But it's also possible through luck of the draw for two players to be on bye on the same week. Naively, the odds of your top two quarterbacks winding up with the same bye are 1 in 8, or 12.5%. (Actually, since you're not rostering two quarterbacks from the same team, they're really 3 in 31, or ~9.6%.)
As a result, there's a roughly one-in-ten chance you'll need a third quarterback next year just to keep the lights on all season.
Now, let's look at wide receiver. Assuming you start three receivers, you'll need at least four players to cover byes. But the odds of two of your top four receivers sharing a bye are actually 55%. (I'll spare you the math.) Which means there's a better-than-even chance you'll need a fifth receiver.
But things actually get messier from there, because the odds that three of your top five receivers share the same bye are actually 14.6%. Looking strictly at byes, you're more likely to need a sixth receiver than you are to need a third quarterback, and this is without even considering any flex positions.
If your goal is to reduce your odds of being unable to start anyone because of byes below 10%, here are the number of players you'd need to roster based on the number of players who are started. Because you'll never start two quarterbacks, kickers, or defenses from the same team, the numbers for those positions are different than they are at receiver, running back, or tight end (where teams can and frequently do start two players from the same team).
As a result, I've listed two results where they differ; QKD is the quarterback, kicker, and defense value, while RWT is the running back, wide receiver, and tight end value.
1 QKD Starter: 2 roster spots
1 RWT Starter: 3 roster spots
1.5 Starters: 3 roster spots
2 Starters: 4 roster spots
2.5 starters: 5 roster spots
3 starters: 6 roster spots
3.5 starters: 6 roster spots
4 starters: 7 roster spots
(Yes, this largely just simplifies down to “roster double the amount of players you start each week”, but isn't it nice to know that there's actual math behind it?)
What about not-byes?
There are other ways for players on your bench to see your starting lineup. For instance, there could be an injury to one of your starters. But assuming injury rates are relatively equal across positions, (they aren't, but again... simplifying assumptions), the ratios you need to guard against injury should be the same as the ratios you need to cover for byes. After all, the more players you start at a position, the more likely one of them gets hurt.
Importantly, the injuries concern does not apply to fantasy defenses, which never miss a game. Which is just one more reason not to burn many roster spots on defenses unless you simply have a ton of roster spots to burn and there are rarely more than 6 or so options available on waivers.
What about a player exceeding expectations and outperforming one of your presumed starters? Again, the more players you start, the more likely this becomes. In a league that starts 36 receivers a week, a receiver only needs to be top-36 to be worth starting. In a league that starts 12 quarterbacks, a quarterback needs to be top 12. It's easier to be one of the 36 best receivers than one of the 12 best quarterbacks.
(This math can change based on what the rest of your team looks like. If your top three receivers are Antonio Brown, Julio Jones, and DeAndre Hopkins, the odds of an end-of-the-bench guy earning starts based on performance decline precipitously.)
So What Should Your Roster Look Like?
In a perfect world, if all else was equal, your roster composition would mirror the ratios dictated above. All else is never perfectly equal, and sometimes talent or injury pushes you away from those ideal ratios. If you have Aaron Rodgers and Andrew Luck, you're probably going to want to roster more than 2 (or even 3) quarterbacks.
But those ratios provide a good target to hit, and if you're deviating from them you should at least be able to justify why.
Enough theory, though, let's give some examples. Imagine a league with 20 roster spots that starts 1 quarterback, 2 running backs, 3 wide receivers, 1 tight end, and 1 defense with no flex. To get your odds of severe bye conflicts below 10%, you'd need to roster 2 quarterbacks, 4 running backs, 6 receivers, 3 tight ends, and 2 defenses. (If there are a decent number of options on waivers, by all means feel free to run with just one defense most of the time.)
This fills 17 out of 20 spots, leaving 3 to play with, 4 if you forewent the second defense. You could easily fill those final three spots with an extra running back, wide receiver, and tight end, or you could skip the tight end to add another wide receiver.
Your final composition, then, would be 2 quarterbacks, 5-6 running backs, 7-8 wide receivers, 3 tight ends, and 1-2 defenses.
Imagine a second league with 24 roster spots that starts 1 quarterback, 2 running backs, 3 wide receivers, 1 tight end, and two flexes, this time with no kickers or defenses. Imagine also that the league gives tight ends 1.5 points per reception, so you're reasonably likely to use one in your flex.
Your presumed starters might be 1 quarterback, 2.67 running backs, 3.67 wide receivers, and 1.67 tight ends. According to the chart above, your minimum targets should therefore be 2 quarterbacks, 5-6 running backs, 6-7 wide receivers, and 3-4 tight ends.
Filling out the rest of your spots, maybe you end up with a target of 3 quarterbacks, 7 running backs, 9 receivers, and 5 tight ends. (You might notice how close this is to the actual roster composition of my team that I mentioned above. This is not a coincidence.)
Again, there's a little bit of wiggle room in that. If there's another quarterback you love and you want to carry an extra, that's totally fine. But if you're deviating from that prescribed ratio by more than a player or two at a position, you should have a phenomenal reason. The further you are from the ideal, the better your reason should be.
“I've been hit really hard by injuries” is one such phenomenal reason. “My top players at this position are all superstars so I don't need as much depth” is another. Again, I'm not saying never deviate, I'm saying deviate with a degree of intentionality.
Yes, sometimes this means passing on a "superior" player at the "wrong" position, but bear in mind that pretty much every player available on waivers is a longshot, so you're not meaningfully impacting your odds of getting a hit on talent alone. But you're substantially improving your odds of getting a hit based on team need.
As long as you keep your target ratios in mind, your team will be optimally suited to handle whatever ravages and curves the season might throw your way. But rosters that are not intentionally composed have a nasty habit of unintentionally decomposing.