How consensus ADP is built and why 8,007 drafts beat one expert

What the number on the board actually is

Every pick number on the Parlae board is an average of where real people took that player in a draft this week. Not a projection. Not a ranking someone typed. On the evening of September 3 the PPR board was built from 7,848 drafts completed between August 27 and September 3. The day before, the same feed reported 8,007. The count moves because the sample is a rolling window: drafts older than about a week fall out, new ones come in, and the average is recomputed daily.

So when the board says Jahmyr Gibbs goes at 1.02, that is shorthand for "across 1,975 drafts in the window, his average pick was 1.5 overall." When it says Josh Allen goes at 3.10, it means his average was pick 33.5. The round and pick form assumes a 12 team league because that is the shape of the drafts in the sample.

Where the sample comes from

The drafts are run on Fantasy Football Calculator, which publishes its data publicly. Its methodology line is short: the numbers come from fantasy football mock drafts in the stated window, and computer selections are filtered out so that only human picks count. The drafts are 12 team, 15 round. Each scoring format is its own pool. On September 3 the PPR pool held 7,848 drafts, the half PPR pool 3,140, the standard pool 1,722, and the 2QB pool 7,045 over a longer window that opened August 4.

We do not blend those pools. If you switch the board from PPR to standard, every price is recomputed from the standard drafts alone. That is why the same player can carry a different pick number on each view, and why a player with a thin standard sample can look jumpy there and steady in PPR.

We also refuse to print a price nobody paid. A player needs at least 15 drafts in a format before the board shows an ADP for him there. Below that, the row says there is no draft data. Filling that gap with a projection would turn a measurement into an opinion, and the whole value of the board is that it is not one.

Source: Parlae board, Sep 3 2026, reading the feed's own sample counts.

What travels with the average

An average alone hides the shape of the crowd, so the feed carries three more numbers for each player: his earliest pick, his latest pick, and the standard deviation of where he went. Read together they tell you how much the room agrees.

Gibbs: average 1.5, earliest 1, latest 4, standard deviation 0.7. Everyone agrees. Ja'Marr Chase: average 3.9, earliest 1, latest 9, deviation 1.0. Broad agreement. James Cook: average 11.5, earliest 4, latest 20, deviation 2.7. That is a player some rooms treat as a top five pick and others let slide into round two.

Then there is Josh Jacobs. Average 55.8, earliest pick 1, latest pick 119, standard deviation 29.4 across 434 drafts. That spread is not indecision. The Packers placed him on the commissioner exempt list on August 30, and the window still contains drafts from before and after that news. The average is telling you two markets are mixed together, and the deviation is how you know.

Why the crowd beats one list

An expert's rankings are one person's ordering of who is better. ADP is the record of what thousands of people did when it was their turn and it cost them something to be wrong. Those are different questions and only one of them is answered by watching the room.

Think about what you actually need on the clock. You do not need to know whether Puka Nacua is better than Chase. You need to know whether Chase will still be there in eleven picks. An expert cannot tell you that, however good the expert is. The crowd can, because the crowd is the thing that will take him.

There is a fair objection here. FantasyPros publishes an expert consensus, and its support pages explain that it is built from rank points across a pool of analysts rather than a simple average, weighted toward the analysts with the best accuracy record. That is a real and useful product. It is the crowd version of "who is better." It is still not "when will he go," which is the question a draft asks you fifteen times.

The other advantage of a price is that it can only be wrong in one direction at a time. If the room is too high on a player, that shows up as an ADP you can decline to pay. If the room is too low, that shows up as a player who keeps falling past his price. Either way you get a number to act on. A ranking that disagrees with the room gives you a debate.

What ADP cannot tell you

It is not a projection. A pick number says where people draft a player, not what he will score, and the two drift apart every year. It is format specific, so a PPR number tells you nothing about a standard league. And it lags the news, because a seven day window keeps yesterday's drafts in the average for a week. The Jacobs spread above is the clearest example we have this season: the board was still averaging in drafts from before the exempt list decision, which is why the same player showed a 23.8 pick slide between the two board days on file.

The right way to use it is as the price of admission. Look at where the room takes a player, decide whether you would pay that, and if you would not, find the next player at that position whose price you would pay.

What to do with this

1. Open the board on your league's scoring format before you look at a single name, because the PPR, half PPR and standard prices are separate samples and a player can move a full round between them.
2. For any player you are targeting, read the earliest, latest and deviation figures on his player page. A tight spread means you must pay the price. A wide spread means a patient drafter can get him a round later.
3. When a player's spread looks like Jacobs, find out what happened before you trust the average. A wide deviation in a short window almost always means news landed inside the window.