How the NFL draft works, and why April grades are useless

The grades arrive before the players do. Within a day of the seventh round finishing, all thirty-two classes have been given a letter by somebody working from the same college tape the clubs worked from, minus the medical files, minus the interview notes, minus the board a scouting department spent nine months arguing over. Some of those grades will turn out to be right. None of them can be known to be right for about four years.

Which is a waste, because the question worth answering is not who won the weekend. It is how does the NFL draft work as a machine: who is allowed into it, how the order gets built, where the extra picks come from, what it costs to move up, and what a club is actually buying when it spends a selection.

Nearly all of that is written down. The eligibility rule, the selection procedure, the wage scale and the option year sit in the collective bargaining agreement and the league's own constitution and bylaws. The compensatory formula does not, which is a story in itself. Everything else is checkable, and once you can see the machine the annual noise around it becomes very easy to tune out.

How does the NFL draft work? Seven rounds, one board, and a card

The draft is formally a selection meeting. Thirty-two clubs take turns choosing from a pool of eligible players across seven rounds, spread over three days: one round on the opening night, two the following evening, the last four in a single long session on the third day.

Thirty-two clubs times seven rounds gives 224 selections before anything is added. Compensatory picks are then bolted on, which pushes the total into the mid-250s in a normal year.

• 7Rounds
• 224Selections before compensatory picks
• 32Compensatory picks awarded league-wide
• 4Maximum compensatory picks for one club

Fixed by the bylaws and the collective bargaining agreement rather than set each year. Compensatory picks awarded for the loss of minority coaches and executives sit outside the 32.

What a club wins with a selection is narrower than most people assume. It does not acquire a player. It acquires the exclusive right to sign one, and nobody else may negotiate with him. Those rights are not permanent: a drafted player who never signs eventually returns to the following year's pool, though under the current wage scale this almost never happens, because there is so little left to argue about.

The pick becomes official when a card is handed to the league's representative at the club's table. Name, position, college, club. Not when the commissioner reads it out, not when the club posts it, not when the player's phone rings. The card is the instrument, and the gap between the card going in and the announcement is where every argument about leaked picks in the last decade has lived.

The three-years-removed rule, and who it is really for

Eligibility has one substantive test. A player must be three full seasons removed from the graduation of his high school class.

That is the whole rule. Not three years of college football, not a degree, not a minimum age in years and months. A player who never attends college at all becomes eligible on the same timetable as one who starts thirty games in a major conference. Players who have exhausted their college eligibility enter the pool automatically. Anyone with eligibility remaining has to declare himself in writing by a deadline in January, and the college side then runs its own withdrawal window with its own conditions about agents and representation, which is where players who change their minds get caught.

The stated justification is development and safety: three years of college strength and conditioning between a teenager and grown professional linemen. There is something in that, and the physical argument is stronger in football than in any other sport with a draft. Compare the route into professional ice hockey, where clubs select eighteen-year-olds and then park them in junior or college leagues for years. Football does not run a farm system, so it outsourced one, and the age rule is what keeps the outsourcing tidy.

The other half of the explanation is less flattering and just as real. The rule restrains supply, and it was agreed between the league and a union whose members are all, by definition, already in the league. Nobody at the bargaining table represents the twenty-year-old it excludes. He is not a member yet, and if he arrives two years later he will be competing with the people who agreed to keep him out. A rule with no constituency to defend against it is an easy rule to keep.

It has been tested. An antitrust challenge in the early 2000s, brought by a player who had been out of high school for two years, succeeded at first instance and was reversed on appeal, on the ground that a rule which is the genuine product of collective bargaining is shielded from antitrust attack by the labour exemption. The practical effect of that ruling is that the eligibility rule cannot be broken by a court. It can only be changed by the two sides who wrote it, and only one of them has any reason to.

There is also a supplemental draft, held after the main one for players whose circumstances changed after the declaration deadline. Clubs bid by naming a round, and the winner forfeits its pick in that round of the following year's draft. It is a real mechanism, it costs a real asset, and it has gone unused for long stretches, which tells you how rarely the case for it arises.

The order is built backwards, and a soft schedule pushes you up it

There is no lottery. The order is a direct function of the previous season, and it runs in two blocks.

The eighteen clubs that missed the playoffs take picks one through eighteen in reverse order of record. Worst record picks first. That is the entire rule for the front half, and it is why the argument about tanking is sharper in football than in basketball, where the lottery deliberately breaks the link between finishing last and picking first. In the NFL the link is not broken at all. Finish worst, pick first, every time.

The back half is where it gets interesting, because the fourteen playoff clubs are not sorted by record or by seed. They are sorted by how far they got. Every club eliminated in the wild card round picks ahead of every club that reached the divisional round, regardless of what either did in the regular season. A thirteen-win club that loses its opening playoff game picks before a nine-win club that wins one and loses the next. Then come the two conference championship losers, then the Super Bowl loser at 31, then the winner at 32.

So playoff results genuinely reorder the tail of the first round, and a club can improve its record, win a home playoff game, and drop four slots for its trouble. If the seeding side of that is unfamiliar, the route from a regular season record to a playoff bracket is its own piece of arithmetic.

Ties are settled by strength of schedule, and the direction of the tiebreaker is the part people get backwards. The club whose opponents won fewer games picks earlier. The logic is that two clubs on the same record are not equally good if one of them got there against a much softer set of opponents, so the soft schedule is treated as evidence that you were worse than your record made you look. It is one of the few places in American sport where a governing body openly says that a win-loss record is not a sufficient measure of a team.

If strength of schedule is also identical, the divisional and conference tiebreakers are applied in reverse, and if a tie survives even that the league draws lots at the scouting combine. Clubs left tied after the draw do not simply hand one of them the earlier slot in all seven rounds. They alternate, so the advantage is split rather than compounded.

1. 1The regular season endsClubs are sorted by winning percentage. No lottery, no ballot, no committee.
2. 2The eighteen non-playoff clubs go firstThey take picks one to eighteen in reverse order of record, worst record picking first.
3. 3Ties go to the weaker scheduleThe club whose opponents won fewer games picks earlier, on the reasoning that the same record against softer opposition is a worse record.
4. 4Playoff clubs are ranked by how far they gotWild card losers fill the next block, then divisional losers, then the two conference championship losers, then the Super Bowl loser at 31 and the winner at 32.
5. 5Record breaks ties inside each playoff blockWithin a block the worse regular season record goes first, so advancing further always costs you slots.
6. 6Rounds two to seven copy the orderThe order is set once and reused, which is why a club's slot in round five mirrors its slot in round one.
7. 7Compensatory picks are addedThey attach to the ends of rounds three through seven, ordered by the value of the free agents that earned them.
8. 8Trades are applied lastEvery pick that has changed hands moves to its new owner, which is why the published order looks nothing like the standings.

The same order is reused in rounds two to seven, before compensatory picks and trades are applied.

Compensatory picks are a rebate for losing players you could not keep

A club that loses good free agents and does not replace them from outside gets extra picks. That sentence is the whole idea, and almost every complication in the system comes from defining "loses", "good" and "from outside".

Only certain departures count. The player has to be an unrestricted free agent whose contract expired and who signed elsewhere inside a defined window. A player the club released does not count, because the club chose to let him go. A player signed after the window closes does not count either, in or out. That last detail drives real behaviour: clubs protecting an award will simply wait, let the market's first week pass without them, and add veterans later at a price that no longer costs them a pick.

The awards are then netted. Qualifying losses are ranked against qualifying signings, matched off against each other, and the unmatched losses generate picks. Lose four and sign one, and three losses remain. The round each pick lands in depends on how valuable the departed player was.

Valuable how? This is the part the league does not publish. The formula is run by the Management Council and is built mainly on the size of the contract the player signed with his new club, measured against league-wide salary levels rather than in raw dollars, with adjustments for how much he actually played and whether he was recognised with postseason honours. Salary does most of the work. The reason is not laziness. It is that the open market's valuation of a player is the least game-able input available, because a club cannot inflate what somebody else chose to pay.

The consequences are worth stating plainly. Since contract value dominates, a club is rewarded for developing players other people want to pay, not for developing players it can keep cheaply. And because the award is netted, a club that spends aggressively in free agency will receive nothing at all, no matter how many of its own players walk out of the door.

Compensatory picks sit at the ends of rounds three through seven, after the last regular selection in each round. Nothing is inserted into the middle of a round, so no club's own picks move because another club received an award. A club may collect a maximum of four in a year, and the league-wide total is held at thirty-two. If the formula produces fewer than that, the remainder are handed out in draft order, continuing past the end of the seventh round, so the total always lands on the same number. A separate award, sitting outside those thirty-two, goes to clubs that develop minority coaches and executives who are then hired as head coaches or general managers elsewhere, paid as third-round picks in each of the two following drafts.

For most of the mechanism's life these picks could not be traded. They were awarded to a club and had to be used by that club, which made them oddly inert: real assets that could not be turned into anything else. That changed in 2017, and the change matters more than it sounds.

Once compensatory picks became tradable they stopped being a consolation and became currency. A club can now run a deliberate strategy of developing players, declining to overpay to keep them, collecting the picks that generates, and then spending those picks as trade capital rather than as selections. That is a coherent way to build a roster, and it is only available because the picks can move. It also quietly rewards the same discipline the cap rewards, which is why clubs good at one tend to be good at the other.

The clock does not take the pick away, and that is exactly the point

Each selection is timed. As the rules currently stand the first round allows ten minutes, the second seven, rounds three through six five, and the seventh four. Those allowances have been shortened more than once and the league publishes the current table with its draft procedures, so check it there rather than trusting any article, this one included.

Here is what almost nobody knows about the clock: running it out does not forfeit the pick.

The selection stays open. What happens instead is that the clubs behind you become free to hand in their own cards and move ahead of you, and once they do, your pick has effectively slid down the board. You still get a player. You just get him a slot or two later than you were entitled to, having watched a name you wanted disappear in the meantime. It has happened in the first round, in front of everybody, which is a far worse punishment than a forfeiture would be.

Understanding that changes what the clock is for. It is not a decision timer. By the time a club is on the clock, the decision has largely been made, because the board was set weeks earlier in a series of meetings whose entire purpose was to rank players by grade and stop anybody from arguing about need in the room. Scouts stack their positions, area scouts defend their reports, the general manager and the head coach fight it out, and the result is a horizontal board where every player sits in a value tier rather than on a wish list.

So when the clock starts, the war room is usually quieter than the television edit suggests. The remaining work is narrow: two or three names left in the top tier, a decision about which one, and a live question about whether somebody will pay enough to make moving back the better answer. The phones are for that second question. A trade has to be reported to the league by both clubs, and the league confirms the terms with both before anything is exchanged, which is the moment that catches out anyone who thought a handshake was enough.

The clock does not reset when a pick changes hands, either. Whoever acquires it inherits whatever time is left rather than starting a fresh allowance, which is why trades on the clock happen at speed and why their terms are almost always simple. Nobody negotiates a complicated deal with ninety seconds showing.

1. 1The board thins to one nameA player the club graded in a higher tier is still available, and the two clubs picking in front of it both need him.
2. 2The call goes outThe club holding the pick is contacted directly. There is no auction house and no central exchange, just a phone line.
3. 3A price is agreed in picksThe currency is current and future selections, occasionally with a player attached. Both sides are arguing about slots, not about points on a chart.
4. 4Both clubs report the trade to the leagueNeither club can act on a handshake. The league takes the terms from both sides and confirms that they match.
5. 5The clock keeps runningThe acquiring club inherits whatever time was left on the original clock rather than starting a fresh allowance.
6. 6The card goes inName, position, college and club, handed to the league representative at the table.
7. 7The pick is official at the tableNot when it is announced and not when it is reported, which is why the news so often breaks before the commissioner speaks.

The sequence is the same whether the move is three slots or twenty. Only the price changes.

Why the old trade value charts were wrong about the top of the first round

For a long time, clubs priced pick trades using a points chart. One number per slot, decaying steeply from the top, and a trade counted as fair when the points roughly balanced. The best known version came out of a front office in the early 1990s and spread across the league because it worked, in a specific and limited sense: it described the prices at which trades were actually clearing.

That is the flaw, and it is a subtle one. The chart was built by observing completed trades. It encoded the market's revealed prices, not the picks' underlying value. If the whole league was overpaying to move up, the chart faithfully recorded the overpayment and then handed it back to everybody as a standard. A tool built from consensus cannot detect that the consensus is wrong.

And it was wrong in a particular direction. The decay from the first pick downward was far steeper than the actual difference in what those picks produced. Moving up ten slots at the top of round one cost a fortune in chart points, on the assumption that the players available in those slots were dramatically better. The evidence never supported a gap that large. Scouting is not precise enough to reliably separate the third best player in a class from the eleventh, and the further down the board you go the flatter the true differences become, right at the point where the chart says they should be collapsing fastest.

The best known academic treatment of this, by Cade Massey and Richard Thaler, made the argument in the form that stuck. Clubs are systematically too confident in their ability to rank players, that overconfidence is priced into the trade market, and once you subtract what a pick costs to sign, the most efficient selections in the draft are not at the very top of the first round.

That last step is the one that reframes everything. A pick's worth to a club is not the player's value. It is the player's value minus what he has to be paid. Surplus value, in other words, and the surplus curve is much flatter than the price curve, because the very top picks come attached to the largest rookie contracts.

• Old chart price
• Surplus after contract cost

Pick 110040

Pick 86044

Pick 203542

Pick 401834

Pick 70922

Both series are invented, indexed so the largest value is 100, and chosen to show the shape of the argument rather than the price of any real pick. No trade or contract is being described.

Show the numbers

Worked example: what a steep chart charges against what a pick returns

Item	Old chart price	Surplus after contract cost
Pick 1	100	40
Pick 8	60	44
Pick 20	35	42
Pick 40	18	34
Pick 70	9	22

Read those two series against each other and the whole modern argument for trading down is visible in one picture. The price series falls off a cliff. The surplus series barely leans. Anywhere the gap between the two is widening, a club moving up is paying for a difference that does not exist.

What replaced the old charts was not a single better chart but a method. Model the expected production of a pick from historical outcomes at that slot. Price the rookie contract that comes with it, which the wage scale makes knowable in advance to the dollar. Subtract one from the other to get surplus. Discount future picks for time and for the uncertainty about where they will land, which is where the market's old rule of thumb comes from, that a pick a year away trades at roughly what a pick a round later would fetch today. Public versions of this exist, several of them built from what drafted players go on to earn on their second contracts, which is a neat trick: it uses the open market's later verdict on the player as the measure of what the pick produced.

None of this makes trading up irrational. It makes it expensive, and it shifts the burden of proof onto the club that moves. If your evaluation of a specific player sits genuinely far above the consensus, paying above the model is correct, because the model is built on average outcomes and you are claiming this one is not average. The clubs that get into trouble are the ones that move up for a player they have graded exactly where everybody else has, and pay a premium for nothing but the certainty of getting him.

The rookie wage scale is why early picks became assets rather than liabilities

Before 2011, drafted players negotiated their own contracts, and the results at the top of the first round were absurd. A player who had never taken a professional snap could be among the highest paid in the league at his position, and a club that picked first in a weak year was handed a liability dressed up as a prize. Holdouts ran into training camp. The money that went to unproven rookies came out of the same pool as the money that went to proven veterans, which is why the players' side eventually traded it away.

The current system replaces negotiation with a scale. Every slot carries a substantially predetermined value, every drafted player signs for four years, and the structure is constrained so that clubs cannot smuggle the old deals back in through escalators and creative guarantees. There is very little to argue about, which is why rookie contracts now get signed in batches with no drama at all.

The effect on how clubs think is enormous, and it runs straight into the accounting rules that govern what a club can spend. A drafted player is cheap, and he is cheap for four years, at a price fixed before anybody knows whether he is any good. That is the definition of an asset. It cuts the other way too: because the cost is fixed and low, a pick that misses does very little damage to the cap, which is precisely why the surplus curve is flatter than the old charts implied.

It also partly rehabilitated the top of the first round. The research that made the case against overpaying for early picks was done under the old system, where those picks came with genuinely punishing contracts. Cut the cost of a top pick sharply, as the wage scale did, and the surplus available in those slots rises. The critique of the old charts did not stop being true, but the specific claim that the first overall pick was a poisoned chalice stopped being true along with it. A cheap quarterback on a fixed four-year deal is the single most valuable structural position in the sport, and everything about how aggressive front offices behave during that window follows from it.

The fifth-year option, and a decision made a year too early

First-round picks come with one extra feature: a club option for a fifth season, which the club must decide on after the player's third year.

Notice the timing. The club is deciding whether to commit to a player's fifth season having seen three of them, and it makes that call in the spring, before the fourth has been played. It is a forecast, not a verdict.

Two things changed the option in the most recent agreement, and both changed behaviour. First, once exercised it is fully guaranteed. It used to be a much softer commitment, which meant clubs picked it up almost reflexively and worried about it later. Now it is real money, guaranteed at the moment of the decision, and clubs sometimes decline it on players they fully intend to keep, simply because they do not want to hand over the guarantee a year early.

Second, the price is keyed to what the player has done rather than to where he was drafted. The old system priced the option by draft slot, which produced the ridiculous result that two players with identical careers owed different amounts because one went eighth and the other went twenty-eighth. The current version sorts players into tiers using playing time thresholds and postseason honours, with the top tier priced at the same figure a club would owe to tag him and the lower tiers built from averages across a wider band of salaries at his position.

That single change made the option a much sharper instrument. A first-round pick who plays constantly and earns recognition prices himself into a genuinely expensive fifth year, which forces his club to decide early whether it wants him long term. A first-round pick who has been merely adequate stays cheap enough to keep without much thought. The system now separates the two cases automatically, which is what a good rule does.

How does the NFL draft work for the players nobody picks?

The seventh round ends and the pool does not empty. Every eligible player who was not selected becomes a free agent immediately, able to sign with any club, and the phones start ringing before the final round is over.

This is the part of the weekend that gets a fraction of the coverage and produces a surprising share of the football. Rosters run to ninety players through the offseason before cutting to fifty-three, plus a practice squad, and no club fills ninety spots with seven or eight draft picks. Every one of them signs a large group of undrafted rookies within hours, most on three-year deals rather than the four a drafted player gets.

The honest comparison is not between drafted and undrafted. It is between the very late rounds and undrafted, and there the difference in expected outcome is thin. Scouting cannot reliably distinguish the 240th best player available from the 260th. What separates the two groups is not quality. It is process. A player taken in the seventh round goes where he is sent. An undrafted player chooses.

That choice is worth something real. An undrafted free agent with two offers can look at both depth charts, count the players in front of him, ask which club runs a scheme that suits what he does, and pick the roster where the path is clearest. Agents work this hard, and it is the one moment in a young player's career where he has any control over his destination. Clubs know it, which is why the recruiting is aggressive and why a club with a thin position group can beat a better club for the same player without spending more money.

The system has a second door as well. Practice squads exist so that clubs can develop players who are not ready, and a player who spends a year on one is being paid to learn a professional offence while his club keeps a claim on him. A meaningful number of the league's starters came through that door rather than through a draft slot, which is the strongest available argument that the seven rounds are not a ranking of talent so much as a ranking of certainty.

What the combine, the pro day and the medicals are each good for

The scouting calendar produces four kinds of information, and confusing them is the most common error in draft coverage.

Tape is the primary input, and nothing else is close. Everything that follows is a check on the tape, a tiebreaker between similar players, or a search for reasons to remove someone from consideration. A club that changes its mind about a player on the strength of a workout has usually not changed its mind about the football. It has found an explanation for something it already suspected.

The combine is mostly a medical event. This is the single most misreported thing about it. The drills are televised and the value is elsewhere: it is the only occasion on which every club's doctors examine every significant prospect under identical conditions, with the same imaging and the same records in front of them. A medical grade can move a player further than any forty-yard dash, and it moves him quietly, because nobody publishes it. Interviews matter for the same reason. Time with a prospect in a room is scarce, and the combine hands it to all thirty-two clubs at once.

Measurements matter too, in a dull and useful way. College programmes list players at flattering heights and weights. The combine establishes what a player actually is, down to arm length and hand size, and at some positions those figures are genuine thresholds rather than trivia. The athletic testing is best used as a filter for outliers at both ends rather than as a ranking tool. Knowing that a player is exceptionally fast tells you something. Knowing that he is a tenth of a second faster than another player tells you close to nothing.

Pro days are the player's home fixture and should be read that way. They happen on campus, on the player's own surface, with his own throwers, run and timed by people who want him to do well. Hand timing is generous. Drills are designed around what he is good at. None of that makes a pro day useless: it is often the only chance to see a player who was not invited to the combine, a second look at someone recovering from injury, and a chance to run position-specific work that the combine's standard script does not cover. It just means the numbers arrive with a thumb on the scale, and every club knows it.

The visits are the quiet part. Each club may bring a limited number of draft-eligible players to its own facility before the draft, a cap that exists to stop the wealthiest clubs simply outspending everybody for access. Those visits get used for medical rechecks, for long interviews, and for the sort of background work nobody discusses publicly. When a club spends one of its limited visits on a player, that is information. When somebody reports the visit and calls it a smokescreen, that usually is not.

The one category shrinking is generalised cognitive testing. The old general knowledge test that produced a decade of leaked scores and bad jokes has been retired in favour of assessments aimed at football-specific processing, which is at least trying to measure the right thing. Whether it measures it well is a separate question, and the clubs are not publishing their answer.

Two structural problems sit underneath all of it. Prospects now prepare for the combine with specialist coaching, so the tests measure preparation as much as ability. And the college calendar has changed around the draft: with a longer postseason on the college side and players increasingly choosing not to play in games that no longer serve them, the tape a club is grading and the body it is examining can be the better part of a year apart.

Why a draft grade issued in April tells you nothing

Consider what a grader actually knows on the Monday after the draft.

He knows which players each club selected and roughly where the consensus had them ranked. He does not have the medical files, which are the largest single source of disagreement between clubs and the public. He does not have the interview notes or the background reports. He does not have the club's board, so he cannot tell whether a selection was a reach or the last name in a tier. He does not know which trades were offered and refused, which is often the most revealing thing a front office did all weekend. And he has no idea how any of these players will respond to the professional game, because nobody does.

What is left to grade is agreement with consensus. A club that takes the consensus fifteenth ranked player at pick fifteen gets a good grade for having added precisely nothing over the market. A club that takes a player twenty spots earlier than the public board gets marked down for the one thing that could plausibly generate an edge, which is disagreeing with everybody else and being right.

Then there is the counterfactual problem, which does not go away with time. A draft class is not good or bad in isolation. It is good or bad relative to the players who were available at those specific slots, and relative to what those picks could have been traded for instead. Judging a class without both comparisons is like judging a poker hand by whether it won.

A fair evaluation window looks roughly like this. The first season is mostly noise, because rookie playing time is driven by roster context: a good player behind an established starter plays less than a mediocre one on a thin depth chart. The second season is where signal starts to appear. The third gives a real read on which players are going to be contributors. The honest verdict lands somewhere in years four and five, when the class reaches its option decisions and its second contracts, because that is when money gets committed by people with something to lose.

Even then the class is not finished. A player can look like a hit for three years and be gone by his second contract. Another can spend two seasons on a practice squad and then start for a decade.

How to judge a draft class honestly

If you want to assess a class rather than react to one, five questions do most of the work, and not one of them can be answered in April.

How many of them got paid again, by anybody? The second contract is the market's verdict, and it beats games started or snaps taken by a distance. A player who earns a real second deal, from his own club or another one, was a successful pick. A player who starts thirty games on his rookie contract and then attracts no bidders was not. Count a class that way and most of them look very different from their April grade.

Did the club beat its slot, or merely fill it? A first-round pick who becomes a solid starter is not obviously a success, because a solid starter is roughly what a first-round pick is supposed to produce. The interesting question is whether the class beat the cost of the slots it used. Late picks who become starters are where drafts are actually won, and they are also the picks nobody grades.

What did the class cost in picks, including the ones given away? A club that moved up twice and surrendered next year's second and third rounder has not had a good draft merely because the players it took are good. The picks it spent were assets and they belong in the calculation. Equally, a club that traded back three times and came away with four extra selections should be judged on all of them, including the ones it used on players nobody has heard of.

Did the process make sense independently of the result? This is the uncomfortable one, because outcomes are noisy enough that a bad process produces a good class often enough to be dangerous. Did the club move back when the board was flat and move up only where it disagreed sharply with the consensus? Did it pay a premium for certainty it did not need? A front office with a sound process will beat one that gets lucky, over enough drafts. Drafts come once a year, so "enough" is a long time, which is exactly why patience is the rarest quality in the business.

Was the class built for the club's actual window? A team a year from contending and a team four years away should draft differently, and the same class can be right for one and wrong for the other. This is where the draft joins up with everything else about how a roster gets paid for, and why it pays to read the two together rather than treating the way American football is put together as a set of separate puzzles.

Ask those five and you will be roughly four years behind the grades, which is about the right distance. The people who made the picks will still be arguing about them then. That is the tell. Nobody inside the building thinks they know in April either.