Fantasy, in Theory: Tiny Edges

A few easy ways to chase extra fractions of a win throughout the season

The majority of writing on the subject of fantasy football is devoted to the very hard things that will give you a huge advantage in your leagues. The two primary reasons for this is that they’re very hard, and that they will give you a huge advantage in your leagues.

Examples in this area include “who should I draft?”, “who should I start?”, and “who should I trade for?”

Today, I wanted to pay a little bit of attention to the very easy things that will give you a minuscule advantage over your league. These things might only result in one extra win every three or four years… but in a game where margins can often by razor-thin, one extra win can cascade to one extra championship.

Or it could be the difference between a 3-win season and a 4-win season. Like I said, the odds are against any of these factors ever mattering. But since they’re so easy, there’s no reason not to do them anyway. They’re essentially free points.

What the Flex

Let’s say you’re in a league that starts 2 running backs, 3 wide receivers, 1 tight end, and a flex. The way we tend to conceptualize it is that our top two backs, our top three receivers, and our top tight end are all “starters”, and our top remaining players are our “flexes”.

From a roster-construction standpoint, this is exactly the right frame of mind. But if your league management software requires you to declare a specific starter as your flex, this mindset is the wrong one for declaring starters.

Instead, when declaring our weekly starters, we should first decide which position we would like to flex. Then, in our lineup, we should use that flex starter spot for the player at that position with the latest start time.

As the name indicates, the flex position is all about maximizing flexibility, and the earlier that starter locks, the less flexibility you have. Why should we voluntarily sacrifice flexibility if we don’t have to?

Let’s say that your league requires two starting receivers and a flex position. Your top three receivers are A.J. Green, Emmanuel Sanders, and Tavon Austin. When you’re working on trades and waiver additions, you consider Green and Sanders your starters and Austin your flex.

But let’s say that Denver and Los Angeles are playing on Thursday this week. In this case, you should slot Sanders and Austin as your starting receivers and put Green in the flex. This way, if A.J. Green were to, say, injure his hamstring in practice on Friday, you would have the option to replace him with any other receiver, running back, or tight end on your roster.

If you’d put Austin in the flex and he’d locked at kickoff, leaving Green as one of your starting wide receivers, then the amount of options available for dealing with his unexpected absence would be greatly reduced.

Again, it’s very rare that this would actually make a difference in the outcome of a game, but not unheard of. Typically we see a handful of players a year who are surprise inactives in late games.

On the off chance that one of them will be yours, you should try to preserve your flexibility for as long as possible.

Variance for Fun and Profit

Questions about how to use variance are as old as fantasy football itself. Should you start the defense facing your opponent’s starting quarterback? Should you start a quarterback / wide receiver pair?

The primary rule is, on a weekly basis, you start the players who you expect to score the most points. This is important, so I will repeat it: you start the players who will score the most points.

If it’s Thursday morning and you’re deciding between two defenses, one of which is facing your opponent’s top quarterback, you start the defense who you think will score the most points. If it’s Saturday and you’re debating between a wide receiver who plays with your quarterback and one who doesn’t, you start the wide receiver who you think will score the most points.

The reason is that variance is inherent in the fabric of fantasy football. Copious, heaping, tilting, crushing amounts of variance. Some guys are going to blow up for insane days. Some guys are going to put up complete duds.

You probably would have loved your chances against a team with Kaepernick, DuJuan Harris, Kapri Bibbs, Robert Woods, Rishard Matthews, and Lance Kendricks last week. And then Kaepernick put up nearly 400 yards, Harris and Bibbs both took receptions 50 yards to the house, Woods had 160 receiving yards against one of the best defenses in the league, Matthews scored a pair of touchdowns, Kendricks chipped in seven catches for 90 yards, and your team got boat-raced.

We think we know how players are going to perform, and over a long enough timeline fantasy projections have a track record of success, but the error bars around each individual projection are potentially huge. And when a lot of players are left to play, all of those error bars add together to produce a truly staggering range of possible outcomes. That squad I listed above could have easily turned in the worst weekly performance in your league’s history, too.

So early in the week, when a lot of players haven’t played yet, it’s a fools errand trying to guess whether a little bit more variance on the margins will matter or not. You start the players who you think will score the most poinst.

But… late in the week, the calculation changes. After the Sunday afternoon games are done, in most contests there will only be one or two players left to play. Suddenly, variance starts to make a proportionately larger difference.

Should you start the defense facing your opponent’s quarterback? Early in the week, the answer is “if you expect it to score the most points, yes.”

If you’re up by 30 points after the Sunday afternoon games, and the only players left are your opponent’s quarterback and your defense, and your league scoring system is heavily performance-based so that defenses can, (and often do), go negative… the answer is “not if you can help it”.

In this situation, there’s far less uncertainty. You know that in order to win, your opponent needs both a great day from his quarterback and a bad day from your defense. If you’re starting the defense facing his quarterback, though, a great day from his quarterback is pretty much by definition a bad day from your defense.

In other words, instead of needing two things to go his way, your opponent only needs one. The story of his comeback is much easier to tell.

At this point of the week, there are three other defenses that have yet to play. If one of them is on your roster or available on waivers, and it’s not a completely terrible option, you should give some strong consideration to playing it to protect your lead.

Again, to be perfectly clear: If today is Wednesday, you start the players you think will score the most points. If it’s Thursday, you start the players you think will score the most points.

If it’s Friday or Saturday, you start the players you think will score the most points. If it’s Sunday morning or early Sunday afternoon, you start the players you think will score the most points. If it’s Sunday evening and your matchup is close, you start the players you think will score the most points.

But if it’s Sunday evening, and most of your match has already been settled, and you’re either up big or down big, then it doesn’t hurt to start considering variance for your last player or two of the week.

I typically ask myself “what needs to happen for a big comeback, here?” If I woke up on Tuesday and found that the trailing team had come back to win, what’s the story of that comeback?

Then, if I find myself with a huge lead, I try to make that story more difficult to tell. If I find myself in the trailing position, I try to make it easier.

Taking a Zero To Seal a Win

Speaking of variance, you know what’s the ultimate low-variance play? Starting a player on bye, (or, in leagues that allow it, leaving a starting spot unfilled). That’s a move guaranteed to get you the exact same score every time.

Of course, the score that it gets you is a zero, but this is the best way to illustrate the difference between maximizing your points scored and maximizing your odds of winning.

On Thursday, assuming you have players on your roster who are expected to score points, you’re never going to take a zero. Because early in the week, points are more important than variance.

But if you have a 3 point lead on Sunday night with no one but your defense yet to play, and defenses are capable of putting up negative scores, sacrificing points to minimize variance increases your chances of winning.

The biggest risk to taking a zero in a close game is that a mid-week score correction will put you from slightly ahead to slightly behind. When gauging this risk, it’s best to ask yourself how frequently players put up negative scores.

For defenses in high-performance scoring systems, negative scores are going to be relatively common. For a wide receiver or running back, they’re technically possible but extremely rare— a player would essentially need to lose a fumble on his first touch, get injured, and not return.

The second-biggest risk is that playoff seeding will come down to tiebreakers and those extra points at the end of the year will prove useful. There’s a similar effect at play here— if it’s week 3, there’s a lot of variance left during the course of the season, so this is a non-factor.

If it’s week 12 and you know you’re tied with another owner who has a very similar point total, then sure, pay attention to how the tiebreakers are looking before you take an intentional zero.

The last, (and most important), consideration is whether your league even allows taking a zero. In some leagues, leaving a starting spot unfilled will cause your entire team to score zero points, which is sub-optimal.

In others, starting a player on a bye is considered “bad form” and highly discouraged. The advantage to be gained is minimal enough that it’s probably not worth upsetting people over, but that’s for each individual owner to decide.

Taking a Zero To Avoid a Loss

There’s one last reason an owner might want to take a zero that has nothing to do with minimizing variance. In redraft leagues with shallow benches, sometimes bye weeks leave an owner facing a conundrum between dropping a quality player for a short-term fill-in or taking a zero at a position.

This problem also faces dynasty teams who are loaded up with quality prospects. Is it worthwhile to drop a 22-year-old receiver who could be a long-term fixture just to grab a journeyman who can get you six points for a week?

Most of the time, I would say the answer is no. The truth is that the majority of fantasy football games are not particularly close.

In a standard scoring league I play that starts 1 quarterback, 2 running backs, 2 wide receivers, 1 tight end, and 1 flex, for instance, only 24% of games have been within 10 points, and only 7% have been within 5 points.

Let’s say that Cole Beasley is available on waivers. Beasley is the 23rd-ranked wide receiver in that format, has topped five points every time, and has topped 10 points 37.5% of the time— he’s a very solid short-term fill-in, probably better than many owners might have available.

But in order for that fill-in to matter, my game would have to be within 10 points, (24%), I’d have to be on the losing side of the ledger, (50%), and Beasley would have to score more than 10 points, (37.5%). Multiply all those percentages out and there’s about a 1-in-20 chance that taking a zero instead of picking up Cole Beasley costs me a win.

This math is greatly simplified to illustrate the point— the actual odds are higher than that, and will partly depend on how good the rest of my team is and how good my opponent is.

In fact, using the same league as an example, (because it’s my only ESPN league and therefore the only league where owners designated players as weekly flex starters), I can check how often a team’s flex starter made the difference between a win and a loss. (This operates under the theory that most teams use their worst starter in the flex slot, which as part one indicates, shouldn’t be the case if everyone is managing their team optimally.)

That league has featured 90 team-games so far. Of those 90 games, six would have gone from a win to a loss had the team taken a zero at the flex position, and one more would have become a tie. That means taking a zero at the flex position would have impacted just 7.8% of games if we count the tie as a full loss, or 7.2% if we count the tie as half a loss.

And again, the caliber of average flex player should be a bit higher than whatever options are available on the street. In practice, I’d estimate that for most owners, there’s about a 5-8% chance that taking a zero at a lower-scoring position winds up costing a win. Which, depending on what you’d have to drop, could easily be worth the risk.

Framing Expectations

None of these tactics are going to make much of a difference in your final won/loss record. I’ve been playing fantasy football for around 15 years and I’ve only been in a handful of games where a late swap to minimize variance was even on the table.

In fact, I wrote about two of those situations in recent years, here and here. In the former, my opponent’s decision not to minimize variance was the wrong one and cost him a victory. In the latter, my decision to maximize variance was the right one… but it didn’t even matter to the final outcome.

Which is actually a good illustration. These situations are rare, and often even when they come up they don’t wind up mattering. A big underdog who correctly opts to maximize variance is still probably going to wind up losing. That’s why they’re a big underdog.

But at the same time, given how easy they are, there’s no reason not to chase those tiny edges where you can find them. At the end of the day, the big, hard parts of fantasy football, (such as valuing players), are going to make the biggest difference in your final record. But that doesn't mean it hurts to be on the lookout to steal an extra win on the margins from time to time.

Follow @AdamHarstad

More articles from Adam Harstad

See all

More articles on: Strategy

See all

More articles on: Timeless

See all