Statistics dopers - a football pickems question.

I run a football pickems contest for the members of a football booster club (Go Ravens!). It is a pick 5 and works as follows:

Each player picks the winner from 5 NFL games in the upcoming week. The picks are ‘straight up’ not ‘against the spread’.

The player weights each pick from 1 to 5. For example if the player is most confident that the Steelers will win (I know…that’s a stretch) he/she applies a weight of 5 to that game. Similarly, he/she applies 4,3 2 and 1 to the other games. The best possible score therefore is 15.

Now, the question. There are 3 games tomorrow (Thanksgiving day). Am I correct in requiring all the entries to be in by noon tomorrow? Is there an advantage to people who wait to see the results of the Thursday games and then make their picks from the remaining 13 match-ups? Obviously they can’t wait for the three games tomorrow and then choose winners from those games - that’s a given.

I have told players that I need the picks whether they pick teams that play tomorrow or not. Am I justified in doing this or does it make no difference statistically.

Edit: Each player picks the winner from 5 of the 16 NFL games in the upcoming week. The picks are ‘straight up’ not ‘against the spread’.

I would think that players who wait until after Thursday are at a disadvantage, since their pool of games is from 13 instead of 16. I don’t see any sneaky advantage they get either.

Hrm. Unless there is injury news. Should Petyon Manning have a brain aneurysm tomorrow afternoon, that would be a real advantage to have waited.

I think what you should say is that it’s fine to have your picks in by some later date/time. But then every player has the right to re-submit their picks up until that date/time, just in case some event occurs that changes the odds. As long as everyone has until the same deadline, it doesn’t matter so much what that deadline is.

I don’t believe that would work. If I picked (in error) the losers of tomorrow’s games I could then go back and re-pick three more games happening on Sunday.

That ain’t fair.

It seems to me you have three choices:

  1. Require all picks by Wednesday.
  2. Require all picks by Saturday, and forbid people from picking the Thursday games.
  3. Give the players a choice, in which they submit picks by Wednesday only if they’re picking a game on Thursday.

Any choice is equally “fair”, since it treats each player the same way. Your question seems to be, if you go with option (3), is there an inherent advantage to being a Wednesday picker or a Saturday picker?

Let us disregard certain rare cases, such as late injury news suggested by muttrox or a team clinching a playoff spot by virtue of a Thursday result. To see how rare these occurrences are, ask yourself how often the point spread for a Sunday game changes materially between Wednesday and Saturday. My understanding is that this is very rare.

Let us further suppose an “efficient market” in football betting, in which all public information is reflected in the point spread. The point spread translates to a probability of victory–a 10-point favorite is perhaps 60% likely to win, and so forth. (I don’t know the percentages, but they can be calculated.)

Under such a hypothesis, the rational way to bet is to assign 5 points to the favorite with the biggest point spread, 4 to the next biggest, and so on. As a Saturday picker, you have 13 games to choose from, versus 16 on Wednesday, so you should submit on Wednesday if heavy favorites are playing in any of the Thursday games.

Of course, in the real world, your players don’t bet in such a mechanical fashion. If they did, the game wouldn’t be much fun. Rather, they try to “beat the market” by applying their knowledge of the game, or even their hunches.

Under such a scenario, there are advantages to waiting until Saturday. You have more time to read about the games, listen to the TV commentators, absorb more information, and even allow your hunches to develop.

So the players will face a trade-off: more time to think about their picks, but fewer games to pick from. In the cold light of rationality, the second consideration is more important, and Wednesday pickers will have an advantage. But in the real world, many bettors will view the first as more important, and will submt on Saturday.

But again, it’s fair, because everybody will have the same options.

I agree that any option is equally fair, if it is applied to all contestants. So if I were you, I would go with whichever rule made the commissioner’s (i.e. your) life easier.

I think a fourth choice is also feasible and fair

  1. Pick any Thrusday games you wish by Wednesday. If you pick n Thursday games, you then have until Saturday to pick the remaining 5-n games.

That could possibly introduce an advantage to picking the Thursday games, because it would allow one to adjust one’s strategy based on the success of those picks. For example, “I lost my Thursday pick; therefore, if I’m going to have a chance for the week, I need to roll the dice on a couple of underdogs that no one else will pick.” The all-Sunday player can’t make that adjustment.


I find the concept of weighing your picks highly intriguing. I think that that would add a unique twist to the betting.

I find the idea of betting “straight up” kind of silly, as anybody can pick the favorite.

Combined, I think that this makes for a terrible system. As Freddy alluded to, anyone with a lick of sense would just look at the spreads, pick the biggest favorite as their #5 weight, the next as their #4 weight, and so forth. At the end of the season, (unless some football fanatic edges out a win by picking the underdog in close games) you’re going to end up with half of the participants tying for first place.

I recommend that you change the system to using the spread. Then you’ve got a hell of an interesting system.


It might be helpful if the posters here would do as the OP asked, and provide a rigorous STATISTICAL answer to the question, rather than educated guesses… :dubious:

Then provide your own answer rather than finding fault with ours.

I am not a statistician, so I can’t. But I will find fault where fault lies. If you don’t like it, don’t post answers that are nothing more than opinion and guesswork to a question asking for fact. :wink:

I’ve participated in similar pools to this one for a while and that doesn’t work, well at least almost never. As long as the pool of participants is large enough there will be statistical outliers from the norm. Whether through luck or skill, one or more pickers will almost always outperform that strategy. It’s the same reason you can’t pick all higher seeds in the NCAA tourney. Even in the best year for such a strategy (I think one year two #1s and two #2s went to the FF) you would have likely lost to someone who got lucky.

My response.


OP here. It is not a bad system at all. Though it is ‘straight up’ this season has seen it’s share of favorites losing especially those with large points spreads that one would logically apply the weighting of 5. Bears against Dolphins being one example.

Keep in mind you are betting on a total of 85 games through the 17 weeks in which you could amass a total of 255 points.

Right now, the leader after 11 weeks has 135 from a total possible of 165. About 82%.