The Same Fifteen Picks. Fifty Points Apart.

Does it matter which of your picks you put the big numbers on? In a confidence pool you rank your games. Your best one carries 15 points, your worst carries 1. Everybody has a theory about the order.

Here is the answer, and it comes in two halves that sound like they cannot both be right. The order is worth nothing to your score. Over the long run it adds exactly zero. And the order nearly doubles your chance of winning — 22 in 100 instead of 15 in 100.

Most people hear "it does not raise your score" and stop. That is the mistake. Scoring more and winning are not the same goal, and once you see the difference you will start noticing it everywhere.

First, why the order is worth nothing

Say you get 8 of your 15 right. Put those 8 on your top numbers and you score 92. Put them on your bottom numbers and you score 36. That is a 56-point swing; a whole week in our family pool usually scores between 53 and 86. The order can move you further than the entire normal range of a week.

So it matters enormously. But here is the catch: you do not know which 8 are the right ones. If you did, you would not need a ranking. Every game is close to a coin flip, so your big numbers are as likely to sit on a loser as a winner. Average that over a season and the order gives back exactly what it costs; the total is zero.

That is a real result and it is worth sitting with. Ranking your picks does not make you better at picking. It only decides how much your luck is worth.

Now, why it doubles your chance of winning

You do not win a pool by scoring well. You win by scoring more than nine other people. That is a different question, and it has a different answer.

We ran 120,000 weeks. In every one, all fifteen games are coin flips and you disagree with the crowd on eight of them. The only thing that changes is where your big numbers go. Watch what happens to the gap between your score and the best score against you.

Look at where the two curves sit. They are centered in the same place — both average −19.7, matching to two decimals. That is the first half of the answer, drawn: the order changed nothing about how you score.

Now look at their width. Putting your big numbers on the games where you differ from the crowd makes the curve wider. Wider means more of it lands right of zero; right of zero is where you win. Same middle, more tail, 22% instead of 15%.

The part that surprised us

We expected this to matter most when you disagree with everybody. It is the reverse. The fewer games you differ on, the more the order matters.

With only four disagreements you more than double your chances — and if you bury those four at the bottom of your sheet, you have thrown away the only thing making you different from everyone else. Notice what that means. If you mostly agree with the crowd, where you put your points is the only decision you have left.

What to take home

Ask what a strategy is actually for. "Raises your score" and "wins you the thing" are different promises, and a lot of advice quietly swaps one for the other.

You will see it once you look. A fund manager who trails the market copies the market, because being averagely wrong keeps his job while being differently wrong does not. A team up three scores runs the ball into a wall; it lowers their expected points and raises their chance of winning. Neither is playing badly. They are playing a different game to the one on the scoreboard.

And the reverse holds, which is the harder half: when you are behind, safe is the risky choice. Playing it close to the crowd guarantees you finish near the crowd, and near the crowd is a loss. We checked this on our own pool — a leader holding a 12-point edge with ten weeks left who sides with the crowd on every game drops from a 28% chance of finishing first to under 4%. That means siding with the crowd, not duplicating one rival's whole sheet; copy one person exactly and you hold your lead over that person for ever, while the other eight run past you.

What the simulation assumes: fifteen independent coin flips, nine rivals who take the crowd side about three times in four and rank their confidence at random, and ties scored as a loss for you.

And what happened when somebody rebuilt it. A second version, written separately in Stan with its own assumptions, put the three strategies at 16.3%, 13.7% and 10.5% where ours put them at 22.4%, 19.3% and 14.5%. Six points apart on the level — and the ratio between best and worst came out 1.55 against our 1.54. Trust the ratio, not the percentage. How much better one ordering is than another survived a rebuild; how often you actually win did not, because that depends on what your particular pool is like.

Our own rankings for this week are in the pool guide, sealed before kickoff so we cannot improve them afterwards. The rest of the paper is at thesportspage.net.

What this does not say: that any of it makes you better at picking games. It does not. Every number here assumes you are guessing, because that is what our own testing says we are doing. This is how to lose slowly and win occasionally — not how to be right.