Methodology

Team ratings

Every team has an Elo rating: a single number for how strong it is, built from every NFL game result since 2010. Beat a good team and your rating rises a lot; beat a bad team and it barely moves. Bigger wins move it a little more, with a correction so blowouts by heavy favourites don't inflate ratings.

Each offseason, ratings are pulled 40% of the way back toward average, because rosters change. Home teams get a small boost (worth about 1.5 points on the spread). We chose these settings by testing which ones best predicted the results of the 2015–2025 seasons.

Simulating the season

After each game day we play out the rest of the 2026 season 10,000 times. In each simulated season, every remaining game gets a winner and a final score based on the two teams' ratings. Ratings keep updating within each simulated season, so a team on a hot streak in one simulation stays hot. That keeps the range of outcomes realistic.

Each simulated season then gets standings, playoff seeds using the full official NFL tiebreakers, and a simulated postseason through the Super Bowl.

Building the draft order

Every simulated season produces a full 2027 first round using the NFL's rules:

• Picks 1–18: non-playoff teams, worst record first.
• Picks 19–24: Wild Card round losers; 25–28: Divisional round losers; 29–30: Conference Championship losers.
• Pick 31: Super Bowl loser. Pick 32: Super Bowl winner.

Within each group, ties go to the team with the easier strength of schedule (the combined record of its opponents). If that's level too, the NFL's division or conference tiebreakers decide (the team that loses the tiebreaker picks first), and a coin flip is the last resort.

We check this logic by replaying real seasons: it reproduces the actual first-round order of the 2024, 2025 and 2026 drafts exactly.

The rooting guide

The rooting guide uses the same 10,000 simulated seasons. For each game this week, we split the simulations into the ones where the home team won and the ones where the away team won, and compare your pick's top-5 odds and average position in each group. The bigger the gap, the more that result matters to you.

“Root for the Texans over the Titans: +22.6% top-5 odds” means your pick finished in the top five 22.6 percentage points more often in the simulations where the Texans won than in the ones where they lost. Games with almost no effect are tucked away at the bottom.

For a pick you own through a trade, the guide follows that pick, not the team you got it from.

The prospect board

The board ranks the top draft-eligible college players: about 110 in all, ten at each of eleven positions, in one overall order. It's built and updated by hand each week using film, production and the wider scouting conversation. It isn't scraped or copied from anyone else's board. Position rankings are simply the overall order filtered to a position, so the two always agree.

The board lists players who are eligible for the 2027 draft. We don't predict who will declare, and every projection assumes the player enters the draft. Arrows show how many spots a player moved since the previous week's board.

Stats and percentiles

Stats come from CollegeFootballData.com and update weekly. Each position gets its own stat set: passing efficiency and yards per attempt for quarterbacks; catches, yards and yards per catch for receivers and tight ends; sacks, tackles for loss and pressures for pass rushers; and so on. Box scores don't capture offensive line play, so linemen have no stat percentiles.

A percentile compares a player with every qualifying FBS player at his position this season, not just the players on our board. “90th percentile” means he's ahead of 90% of them. To qualify, a player needs a minimum amount of playing time for his team's games so far (for example, 14 pass attempts per game for quarterbacks). The exact rule appears under each player's stats.

Draft range, landing spots and fit

We simulate the first round 10,000 times, once for each simulated NFL season, using that season's draft order and who owns each pick. In each mock draft:

• Every prospect gets a draft value from his board rank, adjusted for position. Quarterbacks, edge rushers and tackles tend to go earlier than their rank suggests; running backs, linebackers and safeties later.
• We add uncertainty, so a player can rise or fall from his rank. It's widest in the preseason and narrows week by week as the season gives us more information.
• Teams pick in order, taking the best available player with a nudge toward positions they need.

Top-5, top-15 and round-1 odds are the share of mock drafts in which that happened. The projected range covers the middle 80% of outcomes. Landing spots are the share of mock drafts in which each team took him.

Fit (0–100) combines three things: the team's need at his position (40%), how well his projected range lines up with the team's likely pick (30%), and the chance he's still on the board when the team picks (30%). Every number behind the score is shown in plain language next to it.

These are deliberately simple and adjustable. The positional adjustments, uncertainty and weights are all settings, so the method can be tuned as the season teaches us more.

Reading the odds

• No. 1 / Top 3 / Top 5 / Top 10: the share of the 10,000 simulations in which the pick landed in that range.
• Projected pick: the median pick. Half the simulations land at this pick or earlier.
• Weekly change: compared with the odds after the previous week's games.
• Traded picks: odds belong to the pick, so they're shown under whichever team owns it.

With 10,000 simulations, small moves (under about half a percentage point) can be random noise rather than real change.