Sportsbooks

Find positive EV opportunities on FanDuel

FanDuel posts one of the deepest prop menus in the market. Depth cuts both ways: more prices to check, and more places where a price is stale, correlated, or built from a ladder that does not hold together.

Updated October 21, 2026 · 9 min read

quant+ is not affiliated with, endorsed by, or sponsored by FanDuel.

  • Line and alternates. The main line plus every alternate rung the book posts for that market. Each rung is a separate price and needs its own fair value.
  • Best price across books. The highest number available anywhere for the side you want, which is not the same as the number at the book ranking first overall.
  • Consensus line. The line where the market broadly settles. A projection far from consensus is either an edge or an error, and the burden of proof is on you.
  • Fair price. A probability derived from the market with the margin removed, expressed as odds. This is the comparison that decides everything downstream.
  • Model projection. An independent estimate of the player's distribution over the market, produced from the inputs below rather than from the price itself.
  • Book count. How many venues post that exact market. Two books quoting a prop is a different information environment from twenty.

Field names only. No projection, price or probability is shown because none can be verified here.

Multiplying legs is not how parlays are priced

A same-game parlay is the most common place a bettor convinces themselves they have found a large edge. Take two legs priced near even money, multiply the decimal prices together, and the parlay appears to pay roughly three to one. Compare that against multiplying your own fair probabilities, and the gap can look enormous.

The gap is usually not an edge. It is the correlation between the legs. A quarterback throwing for more yards and his receiver posting more receiving yards are the same event seen twice. Multiplying their prices treats them as independent, which they are demonstrably not, and a book that prices same-game parlays correctly charges you for that dependence.

The honest version of this calculation replaces the product with a joint distribution. If the two legs are positively correlated, the true joint probability is above the product, which means the parlay is not as bad as naive multiplication suggests. If they are negatively correlated, the opposite. Either way, the naive product is the wrong number, and a scanner that reports edges on it is reporting an artefact.

Same-game parlay, stated precisely

  • leg one, fair probability p1
  • leg two, fair probability p2
  • naive product: p1 times p2
  • correct joint: p1 times p2, plus a correlation term that is usually positive for same-team, same-direction legs

The correlation term is not a small correction. For same-team, same-direction legs it is routinely large enough to change the sign of the apparent edge.

Hit rate over the last ten games is mostly noise

Almost every prop tool on the market leads with a rolling hit rate. It is the easiest number to compute and one of the least informative, for reasons that are structural rather than a matter of degree.

Start with the sample. Ten games is ten observations, and the quantity being measured is a count that swings naturally from game to game. If a player averages a value close to the line, a ten-game window will frequently show six or seven overs without anything having changed about the player.

Then consider what the window is actually measuring. A player's opportunity count is set by the game script: how many plays their team ran, whether the game stayed close, whether they were rested in the fourth quarter. Those conditions are not stable across games, so a hit rate blends the player's ability with the situations they happened to be in.

Finally, the line moves. A hit rate computed against a fixed number is measuring the number, not the player. If the market has already adjusted the line upward after a hot run, the historical hit rate against the old line is describing a bet that is no longer available.

Counts like receptions, strikeouts and shots are non-negative and discrete, and their variance usually exceeds their mean. A Poisson model assumes variance equals the mean, which understates the spread, so intervals and tail probabilities built on it are too narrow. A negative binomial, which adds a dispersion parameter and lets variance exceed the mean, is the more honest default for these counts.

  • Why the distribution matters. The probability that a count exceeds a line depends on the whole distribution, not just its average. Two players can share the same projected mean and have materially different probabilities of clearing a specific number, because one has a tighter spread of outcomes than the other.
  • What a usable projection looks like. A distribution over the outcome, conditioned on expected opportunity, expressed as a probability of clearing each rung of the ladder. The probability for the rung you are looking at is then read directly instead of being guessed from a count of recent results.

Four things that move a projection, in order of weight

If a prop model does not expose these, it is not really a model. It is a screen with a sort arrow.

Role and usage

Snap share, route participation, target share, carry share. Usage is what converts team-level expectation into an individual opportunity count. A player's average is mostly a record of the role they had, and roles change faster than season averages adjust.

Opponent tendency

Opponent defence by position, by coverage shell, by down and distance, and by pace of play. A defence that concedes volume to one position is not the same as a defence that is bad overall.

Game environment

Implied team total, spread, and expected number of plays. A receiver on a team expected to score fourteen points has a different opportunity distribution from one on a team expected to score thirty, even at the same target share.

Availability

Injury designation, practice participation, and whether the player is genuinely expected to be active. Participation reports are noisy on purpose and should move a projection by less than a naive reading suggests.

Depth is the defining feature, and it has consequences

Published reviews converge on the same description. Each of the points below is attributed, and each carries a research consequence rather than a verdict.

Alternate line breadth

A published comparison counted sixteen alternate total lines at FanDuel against nine at its main competitor on the markets examined. More rungs means more surfaces to price, and a ladder that long will occasionally contain a rung that has not kept pace with the main line. It also means a naive scan generates far more candidates, most of which are inside the margin.

Source: Sportsbook Review comparison, accessed 2026-10-06

Live betting depth

Reviews consistently describe extensive in-play offerings, including player props during the game. In-play prices refresh faster than pre-match ones and are the least stable surface on any book. Any scan of in-play markets needs a latency budget attached to it, because an edge that decays in seconds is an edge only if you can act in seconds.

Source: BettingUSA review, accessed 2026-10-06

Corporate context

FanDuel Group is controlled by Flutter Entertainment. The controlling interest was acquired by Paddy Power Betfair in July 2018, and the New Jersey operation launched that same year. Public coverage has reported a planned prediction-market product aimed at states where conventional sports wagering is not available. That is a different product line with different pricing mechanics, not a feature of the sportsbook.

Source: Wikipedia corporate record, accessed 2026-10-06

Prop research in five moves

Note that three of the five moves happen before any price is compared.

Establish expected opportunity

Project usage first: snaps, routes, targets, minutes. The opportunity count is the dominant term in any prop projection, and it is available before the price is.

Model the distribution

Convert expected opportunity into a full distribution over the outcome, using a count model whose variance is allowed to exceed its mean.

Read every rung

Compute the probability of clearing each alternate line, not only the main one. The best price is frequently on a rung nobody is looking at.

Handle dependence explicitly

For anything combined into a same-game parlay, replace the product of leg probabilities with a joint estimate. If you cannot estimate the dependence, do not price the combination.

Check the price is live

Confirm the quote age and re-check before acting. Prop markets on a deep menu are exactly where a stale number is most likely to survive unnoticed.

Grade on the close

Record the closing line for the same market. Prop closes are where a projection edge becomes measurable, and where a well-calibrated model separates from a lucky one.

What this page is not claiming

Nothing here asserts that FanDuel prices props badly. Deep menus are expensive to maintain and the markets in them are competitive. The claim is narrower: a deep menu produces more candidate prices, and a candidate is not an edge until it has been compared against a fair value derived independently of the book you are shopping.

Nor does quant+ place bets. The prop research surface produces probabilities, fair prices, best available prices and book counts. The wager is placed by you, in the FanDuel app, at a price you have verified. quant+ has no account access, no affiliation and no sponsorship relationship with FanDuel.

For the underlying theory, start with the expected value guide, and for cross-book prop comparison open the prop research tool.

DraftKings and FanDuel pricing side by side · Expected value calculator

Price the prop, not the streak

Build a distribution for the market, read every rung of the ladder, and compare each price against fair value instead of against a highlight reel.

Scan the FanDuel board · Open prop research

Questions people ask

How do you find positive EV prop bets on FanDuel?

Project the player's expected opportunity first, turn that into a distribution over the outcome, and read the probability of clearing each alternate line off that distribution. Then compare that probability with the implied probability of the price FanDuel is offering across every rung of the ladder. The comparison has to be against a value you produced, not against another book's price for a different line.

Why is a same-game parlay hard to price?

Because the legs are not independent. Two legs describing the same team, the same drive or the same player move together, so the probability of both landing is not the product of their individual probabilities. Books price same-game parlays with a correlation adjustment. Multiplying the leg prices yourself will produce a number that is wrong in a direction you have not identified, and a scanner reporting that number is reporting an artefact.

Is a hit rate over the last ten games a good prop signal?

Rarely on its own. Ten games is a small sample of a count variable whose game-to-game variation is driven as much by opportunity and game script as by ability, and a hit rate computed against a past line describes a price that may no longer exist. It is a useful sanity check on a projection and a poor substitute for one.

What does quant+ show for a FanDuel prop?

The line and its alternate rungs, the best available price across books, the market consensus line, a fair price derived with the margin removed, an independent model projection, and the number of books posting that market. The pattern of those fields is usually more informative than any single one of them.

Sources

  1. FanDuel vs DraftKings: Which Sportsbook Is Better? | Sportsbook Review, sportsbookreview.com
  2. FanDuel Sportsbook Review | BettingUSA, bettingusa.com
  3. DraftKings vs FanDuel | Action Network, actionnetwork.com
  4. FanDuel | Wikipedia, en.wikipedia.org