A projection is an average, not a promise
Most fantasy football platforms and commentators publish a projection for each player: a single number such as 14 points. That figure is best read as the middle of a range of plausible outcomes. A player projected for 14 points will rarely score exactly 14. Some weeks he will score far fewer, some weeks far more, and the projection is the expected value across all of those possible weeks.
Thinking in ranges instead of single numbers is the biggest habit change that probability brings to fantasy football. Two players with the same projection can be very different to own, depending on how spread out their outcomes are.
Floor, ceiling and variance
A player’s floor is a low-end outcome, his ceiling a high-end one, and variance describes how widely results are scattered between them. A standard deviation is the usual way to put a number on it, though in practice most people judge it by looking at a player’s range of weekly scores.
Here is a deliberately simple illustration. Imagine two players, each with a projection of 12 points.
- Player A scores either 4 or 20 points, each equally likely, giving an average of 12.
- Player B scores either 11 or 13 points, each equally likely, also giving an average of 12.
Now consider what you need.
- If you need at least 15 points from that slot to win, Player A gives you a 50% chance and Player B gives you 0%.
- If you need at least 10 points, Player A gives you 50% and Player B gives you 100%.
Real scoring is not two-point distributions, but the lesson carries over. If you are a clear favourite in a head-to-head week, steady players protect your lead. If you are a clear underdog, a high-variance player gives you a better chance of an upset. Your decision depends on more than the average, and it is also why a player with a slightly lower projection can be the right choice.
Matchups and conditional thinking
Matchups mean the opponent a player faces, the game’s likely style, and conditions such as weather. A strong opposing defence may lower a projection and a weak one may raise it. This is conditional probability in action: the chance of a good game given who he is playing differs from the chance in general.
Be careful with how much weight matchups receive.
- Small samples. Early in a season, a defence’s record rests on only a few games. A sample size that small can make an average defence look elite or poor.
- Changing circumstances. Injuries, coaching decisions and changes to the offence can alter how a team plays, which means last month’s numbers may no longer apply.
- Interactions. If a team’s starting quarterback is ruled out, the outlook for its receivers changes. Updating projections as news arrives is a practical form of thinking about conditional probability.
Luck, streaks and regression
Fantasy outcomes include a great deal of chance, because one long play or one touchdown can swing a week’s score. Players who score a lot of touchdowns in one stretch often score fewer in the next, a pattern consistent with regression to mean. Chasing last week’s big score is a common error; underlying involvement in the offence, such as how often a player is targeted or handles the ball, tends to be a steadier guide than a single scoring spike, though it is still imperfect.
Another trap is sunk cost. A player you selected early in a draft is not owed a place in your lineup. The decision each week is about who is likely to help now, not about justifying a past choice.
Over a season versus a single week
A single week is heavily influenced by luck. Across a full season the better manager tends to do better, because small edges in decisions accumulate, but even then a good decision process does not guarantee a good result. Weekly team totals are often roughly described by a normal distribution, meaning most weeks land near the middle and unusually high or low weeks are less common. The law of large numbers says that averages settle down over many weeks, but it makes no promise about any particular week, or even about a single season.
Contests that involve money
Free leagues with friends are games. Paid contests are different. Entry fees are typically pooled and some portion is retained by the operator, which means the average entrant gets back less than they pay in. That is another case of negative expectation. Skill can matter in these contests, but so does luck, and results over a small number of entries can mislead.
If you play for money, decide in advance how much you are comfortable spending and keep to those limits. Do not chase losses with extra entries, and do not borrow to play. Rules about paid fantasy contests vary by location. If playing is causing you stress or financial harm, a local support service can help.