It's 30 minutes before kickoff on Sunday. You have one flex spot and two receivers who look exactly alike on paper. One put up 90 yards last week. The other went a round earlier in your draft. How are you going to choose who to start?
There is a forecast for that exact decision sitting in plain sight, and many fantasy managers never look at it: the player prop lines that sportsbooks post for nearly every starter, every week. How many yards, how many catches, how likely to score. Money rides on every one of those numbers, and they move when the news does: an injury report, a role change, a weather forecast, a bad matchup.
So I tested them. Three NFL seasons, 720 regular-season games, every line as it stood an hour before kickoff, turned into fantasy points and graded against what actually happened.
The test
Take two players at the same position who both played that week, or any running back, receiver or tight end up against another for a flex spot. Which one should you have started? Then count how often each approach picked the player who actually scored more.
I put the betting lines up against the three habits most managers actually use, and against a smarter mix of two of them:
- Last week: start whoever scored more last week.
- Last three weeks: start whoever has scored more over his team's last three games.
- Draft position: start whoever went earlier in drafts (average draft position from Fantasy Football Calculator).
- Draft position + last three weeks: the combination of the two, a weighted average of what a player's draft slot usually scores and his last-three-weeks average. Draft position carries most of the weight early in the season, and recent form's share grows as the weeks go by.
The full rules for each approach are in the methodology.
The scoreboard
How often each approach picked the higher scorer, weeks 2 to 17 of the 2023, 2024 and 2025 seasons:
| Approach | Standard | Half-PPR | PPR |
|---|---|---|---|
| Last week's score | 58.6% | 59.1% | 59.4% |
| Last three weeks | 61.3% | 61.5% | 62.0% |
| Draft position | 61.1% | 61.9% | 62.2% |
| Draft position + last three weeks | 63.4% | 63.8% | 64.2% |
| Vegas player props | 66.6% | 66.8% | 67.2% |
Out of every 100 start/sit calls, the lines get about eight more right than going by last week, about five more than draft position or the last three weeks alone, and about three more than the combination. If you make five real start/sit decisions a week, that is two to six more right calls a season, depending on what you would have gone by instead.
Two calls in three is not a guarantee. A third of the time, the other guy scores more anyway. That is football.
Close calls
The hardest decisions are the ones where the two players look alike on paper, like the two receivers at the top. So I pulled out the close calls: players drafted within one round of each other whose averages over the last three weeks were within 3 fantasy points. That is about one call in every 20 to 25.
| Close calls only | Standard | Half-PPR | PPR |
|---|---|---|---|
| Last week's score | 50.6% | 50.8% | 50.6% |
| Last three weeks | 52.1% | 51.7% | 52.2% |
| Draft position | 50.3% | 50.5% | 50.9% |
| Draft position + last three weeks | 52.1% | 51.8% | 51.9% |
| Vegas player props | 60.2% | 59.6% | 59.9% |
On these calls every habit is a coin flip, and combining them does not help: none of them gets above 52.2%. The lines still pick the right player about 60% of the time. That is an edge of about eight calls in 100 over the best of the rest, more than double their edge across all calls, and they were ahead of each of the other approaches in at least 43 of the 48 weeks. Tighter or looser versions of "close" told the same story. When a call feels like a coin flip, that is when the lines matter most.
What about the matchup? I also tried adjusting the combination for the defense each player faced, using the points that defense had allowed: fantasy points to his position and, separately, points on the scoreboard, from last season and the weeks so far. The effect was near zero, less than a tenth of a call per 100 either way. When the adjustment changed a pick, it was right about half the time.
The House (almost) always wins
The lines out-picked last week's score, the last three weeks and draft position in all 48 weeks I tested, in both standard and PPR scoring. In half-PPR they missed once: one week in 48 when draft position did better. Against the combination, they won 46 or 47 of the 48 weeks, depending on the scoring.
They led in every season, 2023, 2024 and 2025, and at every stage of the year. Early on, draft position is the better of the two single habits; by November, recent form has caught up with it, and the combination beats both at every stage. The lines beat all of them the whole way through, including weeks 15 to 17, the fantasy playoffs, where they picked right 67.6% of the time in PPR against 63.7% for the best of the rest.
Expert projections did not catch them either
Habits are one thing; experts are a tougher test. So I also ran the test against a popular expert source: one site's weekly projections. They are good. They beat the last-three-weeks habit by four to five calls in every 100.
The lines still beat them, by about one more right call in every 100:
| Head to head | Standard | Half-PPR | PPR |
|---|---|---|---|
| Expert projections | 65.7% | 65.8% | 66.0% |
| Vegas, same calls | 66.4% | 66.6% | 67.0% |
I scored the site's projected stats with the same scoring rules as the lines. These are the scoreboard's calls, minus the 2% where the site had not projected both players, which is why the Vegas numbers differ slightly from the ones above. An edge of one call in 100 is thin, and it did not hold every week: the lines led the projections in 36 to 39 of the 48 weeks. But they finished ahead in every season and every scoring format.
Where the lines help most
Same test, split by the kind of decision, PPR scoring. "Flex" is a running back, receiver or tight end up against one of the others, and "Both" is the combination of draft position and the last three weeks:
| Call | Last 3 weeks | Draft spot | Both | Vegas |
|---|---|---|---|---|
| QB | 56.8% | 56.8% | 58.7% | 60.9% |
| RB | 64.0% | 62.6% | 65.9% | 69.4% |
| WR | 61.8% | 63.6% | 65.1% | 67.1% |
| TE | 57.2% | 58.3% | 59.1% | 61.9% |
| Flex | 62.1% | 62.0% | 63.8% | 67.3% |
The lines win every row. The biggest edges are at running back and in flex calls, probably because that is where workloads swing most from week to week, and the books price the swings. Quarterback and tight end calls are hard for everyone, and there the lines' lead was least steady: they beat the combination in 29 of the 48 weeks at each position.
Why the House is so hard to beat
A prop line is not an opinion. It is a price, and the sportsbook, whether a casino or a betting app, takes real bets at it. If a book sets a receiver's yardage line too low, the people who know better bet the over, and the book pays for the mistake. So books have every reason to get the number right, and to move it as soon as the betting says it is wrong.
That pressure comes from both directions. Books pay oddsmakers and build pricing models to set the first number. Then bettors test it, including professionals who make a living finding lines that are off. Their bets push the line toward what the best-informed people believe, and the more money behind a view, the more it moves the number. The line you see an hour before kickoff is what survives all of that.
Experts work under different incentives. An analyst who ranks a receiver too high loses a little credibility, not money. Consensus rankings pool many analysts, which helps, but not enough to beat the market. The lines in this test were a consensus too: the middle of up to ten sportsbooks, each with its own money at stake.
That does not make the lines perfect. Books usually take smaller bets on player props than on point spreads, so props feel less of that pressure than the main game lines do, and the lines still miss a third of the calls. But over three seasons, no habit, no combination of habits and no expert projection I tested beat the lines in a single season. When two of your players look alike, the number with money behind it is the one to trust.
To put the lines to work, the methodology page shows how to turn a prop line into a projection, what the lines say about kickers, and what to do about it.
Player lines: The Odds API. Draft positions: Fantasy Football Calculator ADP (fantasyfootballcalculator.com). Box scores and schedules: nflverse (CC BY 4.0). Nothing here is betting advice.