Home » predictions
Category Archives: predictions
There is a substantial gap between what data websites and betting websites believe. Well, at least they have the order right!!
Here is what the betting websites believe:
|Country||Probability of winning|
Here is what the data websites believe:
Update1: Thanks to suggestions from ‘faustus’, Table 2 and Table 3 are now updated to include numbers of wins / draws, and some grammatical errors in previous versions are corrected.
The Group Stage of 2014 FIFA world cup is now complete, and in fact the round of 16 has already started. Congratulations to the 16 teams that have advanced to the round of 16 after 48 very exciting games. In all 136 goals were scored, averaging 2.83 per game. This is far greater (~35%) than the 2010 FIFA world cup in South Africa, when 101 goals were scored after the group stage.
How did my predictions fare so far? I received an email from ESPNFC Bracket Predictor that I am better than (or equal to) 51.26% or all brackets, which basically means I am barely above average. Not bad for a person who watches football once every 4 years, and for whom the first thing that comes to mind when someone says football is an oval ball usually thrown and caught with the hands. However, when you think of the fact that majority who entered in that bracket competition is likely Americans, it may be sort of an underachievement.
Let’s see which federations were represented in this World Cup by how many teams:
Table 1. Participants of the 2014 FIFA world cup, broken down into continents.
|AFC (4)||Australia, Asia||Australia, Iran, Japan, S. Korea|
|CAF (5)||Africa||Algeria, Cameroon, Ghana, Ivory Coast, Nigeria|
|CONCACAF(4)||N. America||Costa Rica, Honduras, Mexico, US|
|CONMEBOL(6)||S. America||Argentina, Brazil, Chile, Colombia, Ecuador, Uruguay|
|UEFA (13)||Europe||Belgium, Bosnia and Herzegovina, Croatia, England, France, Germany, Greece, Italy, Netherlands, Portugal, Russia, Spain, Switzerland|
A small cartoon to enjoy below, acknowledging the introduction of spray foams in this world cup, before we delve into which continent is the best?
Which continent has played the best football in 2014?
To be fair, we must first admit that the climate is tropical – hot and humid, and many footballers from Europe, N. America, Asia, Australia, Africa may not feel at home. Of course this analysis does not consider or adjust for this, nor is this a statistical inference. It is pure descriptive statistics. Below is the breakdown of all intercontinental games. This table includes the 5 games that was played between two European teams. I have added the last column which represents how many games were won by the first continent (W1), how many games were drawn (D) and how many games were won by the continent that is listed second(W2).
Table 2. Summary of intercontinental games. (Number of games = 48)
|Continent-1||Continent-2||No. games||Goals-1||Goals-2||Total Goals||W1-D-W2|
|N. America||S. America||3||4||3||7||1-1-1|
Excluding the 5 games played between two European teams, we have 43 games and 115 goals. Here is how each continent performed. As expected, the top two continents were South America and Europe (Table 3), although North America was not too far behind and 3 of the 4 teams in fact advanced to the round of 16.
Table 3. Performance of each continent against all continents combined. (Excludes 5 intra-European games)
|Continent||Games||Goals For||Goals Against||Goal Diff||Goal Ratio||% teams advanced||W-D-L|
For those interested, I report in detail the 9 games played between Europe and S. America. As shown in Table 2, South America beat Europe 14-7. It could be debated, however, that Europe may not be represented versus South America by their best teams.
Table 4. S. America v. Europe games (N=9)
|S.America team||Goals||Goals||Europe team|
|Argentina||2||1||Bosnia and Herzegovina|
Hence, my very preliminary analysis based on only 43 games, and including only 9 games between Europe and S. America, shows that S. America is playing the best football in 2014 world cup.
Anand vs. Carlsen world championship 2013 will start in little less than 6 hours. It has been described as a battle between experience and youth, and arguments have been made for both sides. Carlsen is arguably the best player the world has ever seen, and has an Elo rating of 2870, that no other player has ever seen on earth. He is not known to have a very deep opening knowledge, but has superior midgame and endgame knowledge and stamina and tends to grind his opponents to gain wins from equal positions after opening. He is known for precise plays, and has a good intuitive feel for the position. On the other hand, Anand is a 5 time world champion, has won world championship in 3-different world championship, and has been playing top level chess for more years than Carlsen’s age. He is known to have good opening preparations, but recently has been seen unmotivated or disinterested and tends to make mistakes at all three phases of the game. The bookies think Carlsen has 2:1 odds, or about 67% chance to win the match. The official website is chennai2013.fide.com
What does the data say?
Given an Elo difference between the 2 players of 95 points, which roughly translates to 0.6 points for Carlsen in a game, and 0.4 points for Anand in a game, we can predict what will happen during the match. Here, we need to assume a probability of a draw. I did 30,000 simulations, 10,000 each by assuming 40% draw rate, 50% draw rate, and 60% draw rate. The results don’t differ a lot.
The most likely outcome is that Carlsen wins with 6.5-3.5 or 6.5-4.5 points. After 12 games, there is about 80% chance that Carlsen wins, 10% chance the match goes into tie-breaks and 10% chance that Anand wins. We will come back and check after approximately 3 weeks, how good these predictions are 🙂
Please leave a comment for your own predictions. If you wish to just vote in a poll it is here
Political pundits should pay some attention to science and numbers, not to feelings.
More on election predictions: As far as states are concerned: Nate Silver was right on the money – he predicted all the 50 states accurately. There were others (Drew Linzer, Josh Putnam, Markos Moulitsas and many others) who did that too. Even I predicted 48 states correctly.
The pundits on conservative blogs and media (Jonah Goldberg, Karl Rove, Fox News commentators, etc.) need special mention here. Not because their absence from the list above is conspicuous – but because (1) they attacked anyone who used statistical models for their predictions and (2) they were so wrong, it was not even close!
Tarun Wadhwa sums it up here!
2016 elections will happen in 4 years. Everyone who is excited about the elections, will again depend on the ideas of political analysts. We will again have a hard time figuring out what is the truth and what is “a spin”. A lot of predictions have been made for the 2012 elections and it is essential that we evaluate today’s analysts; who did the best this time? whose ideas were completely crazy? who were feeding us the wrong information because of their ideological biases or political leanings? who stuck to the math, irrespective of where they stand on the political spectrum? After tallying the predictions with what actually happens, it would not be easy to answer all the above questions objectively, but we can at least get a feel for it.
Political pundits depend on the same polls that we see; they may be privy to a little more information than us, but they are the pundits because they have a better insight than most of us. However, they are not immune to confirmation bias. Most would embrace polls readily if the findings are similar to their expectations, but on the other hand some may (subconsciously) dismiss polls and try to find methodological flaws with reports that are opposite to their expectations. Some analysts would even knowingly feed the wrong information to manipulate the outcome of the polls or to make political gains. It is for us to find out who is credible.
There is a consensus among most analysts about which states are blue and which are red and which states belong to the “swing state” category. There are 11 swing states with hectic campaigning activity pre-elections: CO, FL, IA, MI, NE, NH, NC, OH, PA, VA, WI. Yesterday, I predicted that Obama gets 290 electoral college votes and Romney 248 – that’s because I think that 3 states: FL, NC and VA will go republican.
Brad Plummer has a blog in the Washington Post that aggregates the positions of many well known political analysts. It is good to have this in one place, just in case the original predictions get taken down after the elections. That is a good list to start and I will try to update it here if I can. Here is the current list – and for some of the “sane” predictions, I have tried to add the list the “swing states” that were chosen to go republican. For some, I have just added “too many” states – which I can revisit if Romney does win the elections to find out who did best.
Real clear politics has Obama up 303 – 235 and have FL and NC as the republican swing states.
Nate Silver, FiveThirtyEight: Obama 303, Romney 235. Republican swing states: FL, NC. I think this was updated a few days ago because this morning I see that FL has swung blue. I give Nate Silver a lot of credit for taking a clear, unambiguous stance, over and over again, in spite of the risks involved.
Sam Wang, Princeton Election Consortium: Obama 303, Romney 235. Republican swing states: FL, NC
Drew Linzer, Emory University: Obama 332, Romney 206. Republican swing states: NC
Michael Barone, The Examiner: Obama 223, Romney 315. Republican swing states: NC, FL, OH, IA, CO, PA, VA, NH, WI
Ezra Klein, The Washington Post: Obama 290, Romney 248. Republican swing states: FL, NC, VA
Larry Sabato, UVA Center for Politics: Obama 290, Romney 248. Republican swing states: FL, NC, VA
Josh Putnam, Davidson College: Obama 332, Romney 206. Republican swing states: NC
Jay Cost, Weekly Standard: Romney victory.
Philip Klein, The Examiner: Obama 277, Romney 261. Republican swing states: FL, NC, VA, CO, NH
Ross Douthat, New York Times: Obama 271, Romney 267. Republican swing states: FL, NC, VA, CO, NH, IA
Jamelle Bouie, The American Prospect: Obama 303, Romney 235. Republican swing states: FL, NC
George Will, The Washington Post: Obama 217, Romney 321. Republican swing states: NC, NV, FL, OH, IA, CO, PA, VA, NH, WI.
Ben Domenech, The Transom: Obama 260, Romney 278. Republican swing states: FL, NC, VA, CO, NH, IA, OH (not sure)
Markos Moulitsas: Obama 332, Romney 206. Republican swing states: NC
Karl Rove: Romney 285, Obama 253.
James Pethokoukis: Romney 301, Obama 227. Too many.
Dick Morris, FoxNews: Romney 325, Obama 213. Too many.
Jim Cramer, CNBC: Obama 440, Romney 98. This guy is just insane. Who is going to believe his stock picks after such an outlandish electoral prediction?
Dean Chambers, UnskewedPolls.com: Romney 311, Obama 227. Too many.