The draft grader explained: what a grade means and what it cannot see

A grade is a comparison, not a prediction

The draft grader answers one question: did you pay more or less for each player than the room pays for him? It compares the overall pick you spent on a player against the average overall pick the market spends on him, using the same consensus ADP that runs the rankings page. Nothing else goes in. Not projections, not our opinion of the player, not what happens in Week 1.

That makes it honest about what it is. A grade of A does not mean your team will win. It means you bought your players at prices the market would call cheap. A grade of D means you paid up, repeatedly. Whether paying up was right is a question the season answers, not the grader.

Pick numbers, never rounds

The first rule is that every comparison is made in overall pick numbers. Round against round is not good enough. In a 12 team league, pick 13 and pick 24 are both "round two," but they are eleven picks apart, and eleven picks is nearly a full round of value. So the grader converts your round and your draft slot into an overall pick number in snake order, and compares that to the player's ADP, which is already an overall number.

Take a player at 2.07 from slot 6 and the grader knows you spent pick 19. If the market takes him at pick 37, you took him 18 picks early. If the market takes him at pick 8, you got him 11 picks late. The rank column is never used; only the price.

The opportunity cost rule

The second rule is the one most graders skip, and it is the reason a first round pick is almost never called a reach.

A pick can only be a reach by as many picks as you could actually have waited. From slot 1 in a 12 team league, your next turn is pick 24, twenty three picks later. If you take a player with an ADP of 20 at pick 1, a naive grader calls that a 19 pick reach. But he would have been gone long before pick 24. You had no way to get him cheaper. The real cost of that pick is zero, and the grader scores it as market price.

The rule in full: if the market takes a player before your next turn, taking him now costs you nothing. If the market takes him after your next turn, the reach is measured from your next turn, not from his ADP. Value in the other direction is always given in full: if you took him later than the market does, you get every pick of that.

The scale, published

Each graded pick produces an opportunity adjusted number of picks, positive for value and negative for a reach. Those are averaged across your roster and converted to rounds at your league size. Then:

A roster dead even with the market scores 74. Every full round of average value per pick adds 9 points; every full round of average reach subtracts 9. The score is clipped to 0 through 100 and mapped to a letter.

Source: the grader's published cutoffs, which are the same in the code and on the page.

A single pick is labelled a steal or a reach only when its adjusted difference is more than half a round. Inside half a round, it is labelled market price, because a six pick difference in a 12 team league is draft room noise and calling it a verdict would be pretending to more precision than the data has.

A worked example

The grader page shows an example: a 12 team league, slot 6, PPR, six graded picks, grade B plus, score 80. The six picks averaged 8 picks later than the market takes those players, counting only the picks the drafter could have waited.

Check the arithmetic. Eight picks in a 12 team league is 0.67 of a round. Multiply by 9 points per round and you get 6. Add 6 to the baseline of 74 and you get 80, which is the floor for a B plus. The example roster included a receiver taken at 2.07 whose market price was 4.01, a steal of 18 picks, and a quarterback taken at 4.07 whose market price was 5.10, a reach of 11. You can redo every line of it by hand, which is the point.

What the grade cannot see

Now the list that matters more than the letter.

It cannot see the future. A team that paid market price for twelve players who all get hurt grades B and finishes last. A team that reached for the right rookie grades C and wins the league. The grade is about the prices you paid, and prices are set before games are played.

It cannot see roster construction. Six steals at wide receiver and no running back is six steals, and the grader will say so. It does not know you need two backs on Sunday. The live draft assistant handles that; the grader does not pretend to.

It cannot see bye weeks, stacking, your league's exact scoring beyond PPR, half PPR and standard, or a keeper you kept. It does not know your league mates.

It cannot see players the market has not priced. If you drafted someone with fewer than 15 drafts in the sample, the grader lists him as uncovered and excludes him. It never guesses a price for him, because a guessed price would let us hand out a grade the data does not support.

And it lags the news exactly as much as the board does. The market price for Josh Jacobs on September 3 was an average of drafts before and after the Packers placed him on the exempt list on August 30. A grade that used that price is a grade against a mixed market, and the deviation on his player page tells you so.

What to do with this

1. Grade the draft, then ignore the letter and read the picks. The two numbers worth keeping are your biggest steal and your biggest reach, because those tell you which prices you read right and which you misjudged.
2. If your worst reach was a player the market takes before your next turn, the grader already scored it as market price; the mistake was not the pick, it was not knowing he was safe to wait on. Check the deviation next year.
3. Put the grade beside your roster shape. A high grade with one running back is not a good draft; it is a good shopping trip, and the waiver bids guide is where you fix it.