WWTBAM Bored

According to Nate Silver, the U.S. has only about a 1 in 3 chance of advancing to the second round of the World Cup. They ran simulations of each scheduled match 10,000 times to calculate the odds of each of the 32 teams (which are divided into eight groups and play the other teams in their group on a round robin basis) finishing first or second in the group (the top two teams in each group advance). Odds in the USA's group

Germany 88.9%
Portugal 43/7%
USA 34.5%
Ghana 32.9%

The good news is that Portugal doesn't have that much better of a chance than the USA of advancing. The odds of the US winning or tying each of their three qualifying matches:

Ghana 36%W, 35%L, 29%T.
Portugal: 32%W, 39%L, 29%T.
Germany: 13%W, 67%L, 20%T.

Best odds of advancing, not surprisingly is host Brazil, with a 99.4% chance of advancing. The US has a "friendly" (exhibition) match with Belgium on June 12, but it is supposedly a closed doors match. The first official match against Ghana is June 16.

http://fivethirtyeight.com/interactives/world-cup/

Nate Silver also predicts that there will be no ties during the group stage (so his image of infallibility will be gone by the end of Italy-England, most likely, on Saturday).

If US starts with a win, then I like their odds to advance, if Germany beats Portugal.

Appa23 wrote:Nate Silver also predicts that there will be no ties during the group stage (so his image of infallibility will be gone by the end of Italy-England, most likely, on Saturday).

That's not remotely close to a fair summary of Silver's analysis. It's true that in each individual game, he believes that a tie is less likely than a victory by one team or another. But for most games he rates the chance of a tie in the 20-30% range.

I think Silver (and the rest of the world) would be astonished if the group stage concluded with no ties. If the chance of a tie were 20% in each of the 48 group stage games, the chances of no ties at all would be 0.00223%. --Bob

Bob78164 wrote: I think Silver (and the rest of the world) would be astonished if the group stage concluded with no ties. If the chance of a tie were 20% in each of the 48 group stage games, the chances of no ties at all would be 0.00223%. --Bob

In the 2010 World Cup, which had the same number of teams and groups, there were 14 ties in 48 matches, or about 29%. Group E was the only one of the eight groups that had no ties in the six group matches. The U.S. and England had two ties in their three total matches in Group C. In Group F, four of the six matches resulted in ties, including all three of New Zealand's matches.

Bob78164 wrote:

Appa23 wrote:Nate Silver also predicts that there will be no ties during the group stage (so his image of infallibility will be gone by the end of Italy-England, most likely, on Saturday).

That's not remotely close to a fair summary of Silver's analysis. It's true that in each individual game, he believes that a tie is less likely than a victory by one team or another. But for most games he rates the chance of a tie in the 20-30% range.

I think Silver (and the rest of the world) would be astonished if the group stage concluded with no ties. If the chance of a tie were 20% in each of the 48 group stage games, the chances of no ties at all would be 0.00223%. --Bob

No, it absolutely is a fair synopsis of his prediction for the group stage -- in every group, he predicts that one team will go 3-0, one team will go 2-1, one team will go 1-2, and one team will be 0-3.

I do not know if the simulations of the games were done in isolation (just doing a thousand simulations of Portugal-US playing) without any consideration of the outcome of the previous game(s) (Portugal is tight after losing to Germany, US playing loose with 3 points in its pocket, US being desperate after a tie or loss to Ghana, etc.)

It is apparent to me that the simulations did not consider the real world implications of a team having won its first two games, and likely have clenched its berth in the knockout round, so it plays several bench players -- let alone a team that might be fine with finishing 2nd rather than first, to get a better draw.

Appa23 wrote:If US starts with a win, then I like their odds to advance, ...

That game's going to make or break the group for the US. A win there, and they can get by with two ties against the better teams. A loss make it unlikely, because they'd probably have to win both games against the Europeans. A tie is hardly better; they'd have to win one and tie the other to pull it off.

We'll see if ESPN's ratings go up or down based upon the results of that first game. Of course, they have the benefit of being in the same hemisphere for a change, so there aren't any 4am games to schedule life around!

Appa23 wrote:

Bob78164 wrote:

Appa23 wrote:Nate Silver also predicts that there will be no ties during the group stage (so his image of infallibility will be gone by the end of Italy-England, most likely, on Saturday).

That's not remotely close to a fair summary of Silver's analysis. It's true that in each individual game, he believes that a tie is less likely than a victory by one team or another. But for most games he rates the chance of a tie in the 20-30% range.

I think Silver (and the rest of the world) would be astonished if the group stage concluded with no ties. If the chance of a tie were 20% in each of the 48 group stage games, the chances of no ties at all would be 0.00223%. --Bob

No, it absolutely is a fair synopsis of his prediction for the group stage -- in every group, he predicts that one team will go 3-0, one team will go 2-1, one team will go 1-2, and one team will be 0-3.

He makes no such prediction. As I said, for each individual game, a victory by the favored team is (under his analysis) more likely than a tie. It follows that aggregating the single most likely result of each game will result in no ties (because a tie is not the single most likely result in any individual game).

But nowhere does Silver present that as a prediction. He is neither stupid nor innumerate. And it would require one of those two qualities to believe, or predict, that the single most likely result will occur on 48 separate occasions. --Bob

Appa23 wrote:
No, it absolutely is a fair synopsis of his prediction for the group stage -- in every group, he predicts that one team will go 3-0, one team will go 2-1, one team will go 1-2, and one team will be 0-3.

No, if you look at the link I posted, it contains an interactive graphic that allows you to click on each of the 32 teams and see what their odds of winning each of their three matches are. When you link to the site, Brazil is highlighted, and it shows their changes are (W-L-D) against Mexico (87-3-10), Croatia (88-3-10) and Cameroon (91-2-8). Those don't always add to 100 because of rounding. So, even though Brazil is a heavy favorite, they have roughly a 10-12% chance of losing or tying each of their games, and the chance of at least one loss or tie are about 29%.

Simulations like Silver's and those that predict the Super Bowl or NCAA March Madness results are based on large numbers of trials using the relative strengths and weaknesses of each team. They tend to be fairly accurate based on the law of large numbers which is the same principle that is used to determine actuarial rates. When the actuarial tables say that there is a 1.5% chance of a man of a particular age dying in the next year, obviously those chances are higher for people who are sick or are in more dangerous occupations than for people who are healthy. However, those tables are based on years of results involving thousands of men of that age, sick and healthy, and, on average, about 1.5% of men of that age die in the next 12 months.

The law of large numbers says that as the number of observed results increases, the odds increase that those results will be close to the true probability. Another example is tossing a coin. Toss it ten times and you could easily get seven or eight heads. Toss it 10,000 times and the odds are the percentages of heads will be very close to 50%.

Things that might hurt a particular team's chances of winning a particular match (injuries, red cards, internal dissension, being "tight") in the long run balance out the chances of the same type of thing happening to the other team. Brazil's chance of winning any particular match in the real world will go up or down based on those factors, but in the long run, their chances of winning each match are close to 90%. However, 90% isn't 100%. Just as if you pick a number from one to ten, it's unlikely you will draw a one, it's not statistically shocking as if you bought the winning Powerball ticket. Similarly, based on what we know today, it's not shocking if Brazil doesn't win a particular match; it would be shocking if they failed to qualify.

All I know is that a lot of work here will be done by a group of people in the Cafeteria on their lap tops. Management here is mostly British and the TVs around the building get switched from CNN to the World Cup as soon as it starts.

Silver makes television appearances touting his predictions for a variety of events (elections, Super Bowls, . . .). He actually titled this effort "World Cup Predictions". Yes, he is using percentages, and he reflects that there is a (lesser) chance that other results might occur, but he is all about touting how he can accurately pick the outcomes of events via his simulation models.

(BTW, he actually does pick each game by bolding the most likely result.)

The U.S. national team coach, Jurgen Klinsmann, won't be placing any bets for his team:

“We cannot win this World Cup, because we are not at that level yet,” Klinsmann told me over lunch in December. “For us, we have to play the game of our lives seven times to win the tournament.”

He leaned back in his chair on the terrace at a Newport Beach restaurant, not far from where he lives in Southern California. Then he shrugged and said, “Realistically, it is not possible.”

More here:
http://soccer.si.com/soccer/planet-futb ... ?eref=sihp

Appa23 wrote:Silver makes television appearances touting his predictions for a variety of events (elections, Super Bowls, . . .). He actually titled this effort "World Cup Predictions". Yes, he is using percentages, and he reflects that there is a (lesser) chance that other results might occur, but he is all about touting how he can accurately pick the outcomes of events via his simulation models.

(BTW, he actually does pick each game by bolding the most likely result.)

There is a difference between making a prediction and the degree of confidence. If I pick a ball at random from a container with 51 white balls and 49 black balls, my prediction is that I will draw a white ball, but it's virtually a 50/50 proposition, so there's not that much of a degree of confidence. On the other hand, if there are 99 white balls and one black ball, my degree of confidence is much higher. When Silver examines a national election or a major election in which there have been a substantial number of other polls, he may examine twenty or thirty different polls. If all of them say the same thing, then there is virtually no chance the other result may occur, and if they all say the same thing, then there's a high degree of confidence they will be correct (if they've polled correctly).

On the other hand, Silver himself gives Brazil a 45% chance of winning the World Cup. That is far and away better than the chances of any single one of the other 31 teams (Argentina is second with 13%), but, on the other hand, is still less than a 50% proposition. Play the World Cup hundreds of times and Brazil will win far more often than any other team. However, the World Cup is only played once... upsets do occur (especially considering that matches after the first round could be decided on penalty kicks), and the other 31 teams combined stand a better chance of winning than Brazil does. Still, Brazil's chances are better than the chances of the favorite in the NCAA basketball tourney once it gets to the round of 32, which may have a 10-20% chance of winning.

The point isn't that upsets can occur: it's that Silver is saying as much if you understand his article. He predicts Brazil will win because it's the most likely scenario, but not that it's overwhelmingly likely. On the other hand, he often picks elections with a much greater degree of certainty.

Appa23 wrote:Silver makes television appearances touting his predictions for a variety of events (elections, Super Bowls, . . .). He actually titled this effort "World Cup Predictions". Yes, he is using percentages, and he reflects that there is a (lesser) chance that other results might occur, but he is all about touting how he can accurately pick the outcomes of events via his simulation models.

(BTW, he actually does pick each game by bolding the most likely result.)

I've seen his television appearances and I've been a reader of his since his Baseball Prospectus days. He is quite good at articulating the difference between a probabilistic model and a prediction. FiveThirtyEight does the former, and that's what he's done with the World Cup. --Bob

Bob78164 wrote: I've seen his television appearances and I've been a reader of his since his Baseball Prospectus days. He is quite good at articulating the difference between a probabilistic model and a prediction. FiveThirtyEight does the former, and that's what he's done with the World Cup. --Bob

It's still a prediction. The difference is the degree of confidence, which is much lower in most sport situations. It's the difference between picking the winner of a very evenly matched football game and picking the winner of Alabama vs. Georgia State in football. In both cases someone picks the result that he believes is more likely to occur; the only difference is the degree of confidence.

Of course, sometimes things like this occur:

Super Bowl 42 Simulations Are In: New England Wins

All the services that ran simulations had the Patriots winning 75-90% of the time. One of them was interesting:

As reported in USAToday, the SportSims [a service that had New England winning 79% of the time] analysis concludes the Giants' keys to victories are to sack Tom Brady three times, contain Wes Welker while not turning the ball over themselves. If they do that, SportsSims suggests, their likelihood of winning goes up to 53%.

Actual statistics:

Brady sacked five times.
Giants turn the ball over one time.
Wes Welker 11 catches for 103 yards and no touchdowns.

And one more statistic:

Giants 17, Patriots 14

silverscreenselect wrote:Of course, sometimes things like this occur:

Super Bowl 42 Simulations Are In: New England Wins

All the services that ran simulations had the Patriots winning 75-90% of the time. One of them was interesting:

As reported in USAToday, the SportSims [a service that had New England winning 79% of the time] analysis concludes the Giants' keys to victories are to sack Tom Brady three times, contain Wes Welker while not turning the ball over themselves. If they do that, SportsSims suggests, their likelihood of winning goes up to 53%.

Actual statistics:

Brady sacked five times.
Giants turn the ball over one time.
Wes Welker 11 catches for 103 yards and no touchdowns.

And one more statistic:

Giants 17, Patriots 14

You are an evil, evil man.

I predict that there will be no ties during the remainder of the 2014 Major League Baseball season. Thank goodness for sports that don't believe in kissing their sister (or brother, as the case may be)....



lb13

I predict that it will not snow in Omaha on January 1, 2015.
I predict that it will not snow in Omaha on January 2, 2015.
I predict that it will not snow in Omaha on January 3, 2015.
I predict that it will not snow in Omaha on January 4, 2015.
I predict that it will not snow in Omaha on January 5, 2015.
I predict that it will not snow in Omaha on January 6, 2015.
I predict that it will not snow in Omaha on January 7, 2015.
I predict that it will not snow in Omaha on January 8, 2015.
I predict that it will not snow in Omaha on January 9, 2015.
I predict that it will not snow in Omaha on January 11, 2015.
I predict that it will not snow in Omaha on January 12, 2015.
I predict that it will not snow in Omaha on January 13, 2015.
I predict that it will not snow in Omaha on January 14, 2015.
I predict that it will not snow in Omaha on January 15, 2015.
I predict that it will not snow in Omaha on January 16, 2015.
I predict that it will not snow in Omaha on January 17, 2015.
I predict that it will not snow in Omaha on January 18, 2015.
I predict that it will not snow in Omaha on January 19, 2015.
I predict that it will not snow in Omaha on January 20, 2015.
I predict that it will not snow in Omaha on January 21, 2015.
I predict that it will not snow in Omaha on January 22, 2015.
I predict that it will not snow in Omaha on January 23, 2015.
I predict that it will not snow in Omaha on January 24, 2015.
I predict that it will not snow in Omaha on January 25, 2015.
I predict that it will not snow in Omaha on January 26, 2015.
I predict that it will not snow in Omaha on January 27, 2015.
I predict that it will not snow in Omaha on January 28, 2015.
I predict that it will not snow in Omaha on January 29, 2015.
I predict that it will not snow in Omaha on January 30, 2015.
I predict that it will not snow in Omaha on January 31, 2015.

I DO NOT predict that it will not snow in Omaha in January, 2015

littlebeast13 wrote:I predict that there will be no ties during the remainder of the 2014 Major League Baseball season. lb13

Strangely, there is a possibility of ties in MLB (discounting the infamous All-Star Game of a few years ago); or at least there used to be. They may have changed it over the years.

If a game had gone the required number of innings to be an "official" game, but was tied at the time of a rain suspension, and the game was never resumed (usually because it didn't matter for the teams' postseason rankings, or for practical reasons), the game was technically a tie for the record books, and the batting/pitching stats counted. It generally wasn't recorded as such in the agate of the papers because that determination usually was at the end of a season (plus, there wasn't a spot for ties).

SpacemanSpiff wrote:

littlebeast13 wrote:I predict that there will be no ties during the remainder of the 2014 Major League Baseball season. lb13

Strangely, there is a possibility of ties in MLB (discounting the infamous All-Star Game of a few years ago); or at least there used to be. They may have changed it over the years.

If a game had gone the required number of innings to be an "official" game, but was tied at the time of a rain suspension, and the game was never resumed (usually because it didn't matter for the teams' postseason rankings, or for practical reasons), the game was technically a tie for the record books, and the batting/pitching stats counted. It generally wasn't recorded as such in the agate of the papers because that determination usually was at the end of a season (plus, there wasn't a spot for ties).

From Wikipedia, I'm too lazy to verify this:

In North American Major League Baseball, a game may end in a tie only due to weather or, historically, darkness (a called game due to darkness is unlikely to happen now that all Major League parks have floodlights; darkness also means reaching the curfew prohibiting innings from starting after 1 AM local time). Before 2007, Tie games ended by weather were replayed from the start, but since 2007, the games are continued from where they left off. A tie game may also be declared if a game is tied, the two teams are not scheduled to play again for the remainder of the year, and the game does not affect playoff implications.

TheConfessor wrote:I predict that it will not snow in Omaha on January 1, 2015.
I predict that it will not snow in Omaha on January 2, 2015.
I predict that it will not snow in Omaha on January 3, 2015.
I predict that it will not snow in Omaha on January 4, 2015.
I predict that it will not snow in Omaha on January 5, 2015.
I predict that it will not snow in Omaha on January 6, 2015.
I predict that it will not snow in Omaha on January 7, 2015.
I predict that it will not snow in Omaha on January 8, 2015.
I predict that it will not snow in Omaha on January 9, 2015.
I predict that it will not snow in Omaha on January 11, 2015.
I predict that it will not snow in Omaha on January 12, 2015.
I predict that it will not snow in Omaha on January 13, 2015.
I predict that it will not snow in Omaha on January 14, 2015.
I predict that it will not snow in Omaha on January 15, 2015.
I predict that it will not snow in Omaha on January 16, 2015.
I predict that it will not snow in Omaha on January 17, 2015.
I predict that it will not snow in Omaha on January 18, 2015.
I predict that it will not snow in Omaha on January 19, 2015.
I predict that it will not snow in Omaha on January 20, 2015.
I predict that it will not snow in Omaha on January 21, 2015.
I predict that it will not snow in Omaha on January 22, 2015.
I predict that it will not snow in Omaha on January 23, 2015.
I predict that it will not snow in Omaha on January 24, 2015.
I predict that it will not snow in Omaha on January 25, 2015.
I predict that it will not snow in Omaha on January 26, 2015.
I predict that it will not snow in Omaha on January 27, 2015.
I predict that it will not snow in Omaha on January 28, 2015.
I predict that it will not snow in Omaha on January 29, 2015.
I predict that it will not snow in Omaha on January 30, 2015.
I predict that it will not snow in Omaha on January 31, 2015.

I DO NOT predict that it will not snow in Omaha in January, 2015

Will it snow in Omaha on January 10, 2015? --Bob

Bob78164 wrote:

TheConfessor wrote:I predict that it will not snow in Omaha on January 1, 2015.
I predict that it will not snow in Omaha on January 2, 2015.
I predict that it will not snow in Omaha on January 3, 2015.
I predict that it will not snow in Omaha on January 4, 2015.
I predict that it will not snow in Omaha on January 5, 2015.
I predict that it will not snow in Omaha on January 6, 2015.
I predict that it will not snow in Omaha on January 7, 2015.
I predict that it will not snow in Omaha on January 8, 2015.
I predict that it will not snow in Omaha on January 9, 2015.
I predict that it will not snow in Omaha on January 11, 2015.
I predict that it will not snow in Omaha on January 12, 2015.
I predict that it will not snow in Omaha on January 13, 2015.
I predict that it will not snow in Omaha on January 14, 2015.
I predict that it will not snow in Omaha on January 15, 2015.
I predict that it will not snow in Omaha on January 16, 2015.
I predict that it will not snow in Omaha on January 17, 2015.
I predict that it will not snow in Omaha on January 18, 2015.
I predict that it will not snow in Omaha on January 19, 2015.
I predict that it will not snow in Omaha on January 20, 2015.
I predict that it will not snow in Omaha on January 21, 2015.
I predict that it will not snow in Omaha on January 22, 2015.
I predict that it will not snow in Omaha on January 23, 2015.
I predict that it will not snow in Omaha on January 24, 2015.
I predict that it will not snow in Omaha on January 25, 2015.
I predict that it will not snow in Omaha on January 26, 2015.
I predict that it will not snow in Omaha on January 27, 2015.
I predict that it will not snow in Omaha on January 28, 2015.
I predict that it will not snow in Omaha on January 29, 2015.
I predict that it will not snow in Omaha on January 30, 2015.
I predict that it will not snow in Omaha on January 31, 2015.

I DO NOT predict that it will not snow in Omaha in January, 2015

Will it snow in Omaha on January 10, 2015? --Bob

That one was too close to call. Could go either way.

Appa23 wrote:Silver makes television appearances touting his predictions for a variety of events (elections, Super Bowls, . . .). He actually titled this effort "World Cup Predictions". Yes, he is using percentages, and he reflects that there is a (lesser) chance that other results might occur, but he is all about touting how he can accurately pick the outcomes of events via his simulation models.

(BTW, he actually does pick each game by bolding the most likely result.)

By the same logic, he's "predicting" that only one team will advance from Group A. Brazil is a virtual lock to advance, but each of the other three teams has less than a 50% chance to advance. Do you really think Silver expects Group A to advance only one team? --Bob

silverscreenselect wrote:

SpacemanSpiff wrote:

littlebeast13 wrote:I predict that there will be no ties during the remainder of the 2014 Major League Baseball season. lb13

Strangely, there is a possibility of ties in MLB (discounting the infamous All-Star Game of a few years ago); or at least there used to be. They may have changed it over the years.

If a game had gone the required number of innings to be an "official" game, but was tied at the time of a rain suspension, and the game was never resumed (usually because it didn't matter for the teams' postseason rankings, or for practical reasons), the game was technically a tie for the record books, and the batting/pitching stats counted. It generally wasn't recorded as such in the agate of the papers because that determination usually was at the end of a season (plus, there wasn't a spot for ties).

From Wikipedia, I'm too lazy to verify this:

In North American Major League Baseball, a game may end in a tie only due to weather or, historically, darkness (a called game due to darkness is unlikely to happen now that all Major League parks have floodlights; darkness also means reaching the curfew prohibiting innings from starting after 1 AM local time). Before 2007, Tie games ended by weather were replayed from the start, but since 2007, the games are continued from where they left off. A tie game may also be declared if a game is tied, the two teams are not scheduled to play again for the remainder of the year, and the game does not affect playoff implications.

I made my statement based on this new rule that was put into place prior to the 2007 season, but was unaware that there was still a caveat to the rule that could result in a game being declared a tie (Though tie games in baseball never did show up in the standings... ties only existed as a way to make the stats from that game count). '07 was the last year I ran my baseball trivia game on the Bored, and one of the questions I asked was which two teams played the final tie game in MLB history.... it was the Astros and Reds in this game on 6/30/05...

http://www.retrosheet.org/boxesetc/2005 ... IN2005.htm

lb13

littlebeast13 wrote: I made my statement based on this new rule that was put into place prior to the 2007 season, but was unaware that there was still a caveat to the rule that could result in a game being declared a tie (Though tie games in baseball never did show up in the standings... ties only existed as a way to make the stats from that game count). '07 was the last year I ran my baseball trivia game on the Bored, and one of the questions I asked was which two teams played the final tie game in MLB history.... it was the Astros and Reds in this game on 6/30/05...

http://www.retrosheet.org/boxesetc/2005 ... IN2005.htm

lb13

What I don't understand is why so many people acted as though Bud Selig was worse than Hitler for letting the 2002 All Star Game end in a tie. It is just an exhibition, and it doesn't matter who wins (but now it does thanks to that horrible rule that the ASG determines home field in the World Series). The alternative was to let non-pitchers pitch in the ASG, and the same people who bitched about the tie would have bitched about that as well.