How market prices become fantasy projections
The short version: de-vig the props, integrate the ladders, convert to your scoring, rank by value over replacement (for the full-season draft board and again for each week's start/sit), then let an accuracy-weighted crowd argue with the result.
Why market prices?
Betting markets are an independent signal. People stake money on these numbers, and when a season rushing-yards line drifts from reality, someone profits by pushing it back. That does not make the market smarter than analysts or ADP, and we don't claim it is. It makes the market a second opinion worth checking: where our board and the draft room disagree, there is a specific, verifiable price behind the disagreement, and you can go look at it.
The honest caveats, because they're real: books shade lines to balance action, and season-long props are thinner markets than game lines. That's why no single book's number is read raw here. Every line is de-vigged, multiple books are pooled with prediction markets, and threshold ladders are integrated into a full range of outcomes rather than quoted as one median. Where no market prices a player at all, the board says so instead of pretending.
De-vigging player props
A sportsbook's over/under isn't a fair probability: the book builds in a margin (the vig). We strip that margin so both sides of a line sum to 100%, which turns a posted line into an honest market expectation for the stat. Before the season, that prices season-long totals (passing, rushing, and receiving yards and touchdowns, receptions, interceptions); once games kick off, the identical de-vig runs on each week's game props, so the weekly board is priced exactly the way the draft board is. A one-sided quote (a touchdown price with no “won't score” side to balance it) is corrected with a standard margin assumption instead of a paired one.
Touchdown markets usually price only whether a player scores at all, not how. That single number is split into rushing and receiving touchdowns using the player's own projected mix, so the split cells stay tagged as market-derived rather than passed off as a model guess.
Prediction-market ladders
Prediction markets carry threshold ladders: separate yes/no contracts for a player clearing 900, 1,100, or 1,300 rushing yards, each with its own live probability. A ladder is a survival curve, a full boom-to-bust range of outcomes rather than a single median, and we integrate it to get the expected value rather than averaging rungs. Thresholds from different venues describing the same stat are pooled into one denser curve before integrating, and a lone threshold is fit to a distribution instead of being read as a coin flip.
In-season the same integration runs on per-game player ladders (touchdowns, passing, rushing and receiving yards, receptions) for that week's matchup, so weekly projections carry the same range of outcomes the season board does.
From stats to points to value
Expected stats become fantasy points under your scoring (PPR, half-PPR, or standard). Points alone don't rank players: a QB's 20 points and a TE's 12 mean different things. So the board ranks by VORP: points above replacement at each position, computed for your exact league settings. Replacement is the last startable player at most positions, and the weekly streaming pool for kickers and defenses, because churning the waiver wire there yields near-top production, so only elite K/DST clear the bar. Change teams, roster slots, or scoring and every value on the board recomputes. Alongside the projection, each player carries a floor-to-ceiling range so you can see how wide the outcomes are, not just the midpoint. Tiers break where consecutive players are separated by a per-game points cliff, capped in size so a shallow gradient still tiers, and each band header shows the points you give up by waiting out the tier. The volatility meter between floor and ceiling rates the relative width of that p10-p90 band (spread divided by projection), ranked within the position, deliberately not the absolute spread, which would just flag the biggest projections as the riskiest.
Where no market prices a stat, the number falls back to a consensus of external projection sources (a trimmed mean, corrected for the optimism bias built into expert projections), and only where even those are silent does our model (team totals × projected usage) fill in. The board tags each: a market number, a consensus number, and a model number are never passed off as one another, and it shows what share of each player's projection is actually market-priced.
ADP and Edge
ADP is a live aggregate of real drafts across the major platforms, in the scoring you have selected, shown as-is with no fallback: a player nobody is drafting has no ADP, and the board says so. Edge is the gap between where the market values a player (his value rank, or his Fair $ in auction) and what the draft room pays for him (his ADP, or the Avg $ the room bids). It is a disagreement flag, not a verdict. A green edge means the room lets him go later or cheaper than the market prices him; a red edge means the room takes him earlier or bids him past his price. Which side is right is the question the rest of this page is about.
What this can and can't tell you
The standard objections to pricing fantasy off betting markets are mostly right, so here is exactly how each one is handled, and the two things this board doesn't do yet.
Prop lines are medians. Fantasy scoring is a mean.
A 1,000-yard season line means the market thinks over and under are equally likely. But your fantasy score sums averages, and for skewed stats the average sits above the middle: touchdown distributions have a long right tail, so a player whose median is 8 TDs might carry an expected value closer to 8.5. Reading lines raw would systematically misprice every skewed stat. We don't read them raw: each stat gets a distribution fit and a per-stat skew correction before it becomes points.
Thin markets get filtered, not trusted.
Season-long markets can be illiquid, and an illiquid price is a soft price. Prediction-market rungs with wide bid-ask spreads are dropped before pricing, and a lone rung far out in the tail is never allowed to price a stat by itself. If the only quote on a stat is a stale 90%-yes contract, we'd rather fall back to a fill than pretend that's a market opinion.
When there's no market, we say so.
Plenty of stats have no usable price. Those fall back to a blend of outside projection sources, and every such number is labeled as a fill on the player page. A fill is never dressed up as market data.
Players who can't play carry no value.
A player who's suspended or ruled out is dropped to zero draft value and takes up no replacement slot. A market can keep re-quoting a frozen line long after the player is off the field, and that stale price is not allowed to prop up a rank. Players expected back from injury keep their value; the games-played caveat below is how we handle their risk.
What we don't do yet.
Two real gaps. First, no per-player games-played model: everyone is projected over the same expected-games number. Season medians already embed some missed-game risk, so players with injury history can look cheap on our board, so discount their edge accordingly. Second, no multi-year backtest: the site launched in August 2026 and prior-season market captures don't exist, so we can't honestly show you historical accuracy yet. Instead we archive every board and grade it against real results, in the open, starting with Week 1 of the 2026 season, so the track record builds where you can see it.
The crowd is the second opinion
Alongside the market, thousands of head-to-head votes ("who do you start?" in season, "who do you draft first?" before it) build an independent crowd ranking. The mechanism is plain Elo: every vote is a match between two players, ratings are seeded from consensus ADP, and a player's rating is shrunk toward that seed until he has votes of his own. The ± marker beside a name appears only once a player has enough votes for the crowd's read to mean something. Votes aren't equal: each voter's start/sit calls are graded against the market's own probabilities once games play out, and proven voters carry up to 3× weight. Calls decided by under 3 points are not graded; that close is a coin flip, not a read. Where the crowd and the market disagree, you're looking at a genuine difference of opinion between two independent signals, and that disagreement is the most interesting thing on the site.
What this is not
Projections, not picks. Nothing here is betting advice; the site takes no positions and books no bets. Market prices are used as information about player expectations, the same way you'd read a weather forecast.