CONCACAF Qualifying For World Cup 2018 Prediction Contest: Round 2 [R]

https://docs.google.com/spreadsheets/d/1RGsJdCklTrh12WWERZ3YOUMkOyfqJfU4NgJ0LsMs3x0/edit#gid=0 has people's predictions. Round 2 starts in about 9 hours with Nicaragua-Suriname. Predictions for both legs must be made before the first leg starts. I don't know if I'm going to post prediction results during the round or wait until after all 20 games.

 

Here are bartecky's predictions (he started a Conversation with me saying he was unable to post):

Saint Vincent 1-0 Guyana
Guyana 2-0 Saint Vincent

Antigua 4-1 St. Lucia
St. Lucia 0-2 Antigua

Puerto Rico 1-2 Grenada
Grenada 2-1 Puerto Rico

Dominica 1-4 Canada
Canada 4-0 Dominica

Dominican Republic 0-1 Belize
Belize 2-0 Dominican Republic

Guatemala 5-0 Bermuda
Bermuda 1-2 Guatemala

Aruba 0-2 Barbados
Barbados 2-0 Aruba

St. Kitts 1-3 El Salvador
El Salvador 4-0 St. Kitt

Curacao 1-0 Cuba
Cuba 1-1 Curacao

Nicaragua 2-1 Suriname
Suriname 0-1 Nicaragua

 

@barnabyjonez

You posted predictions for games that already happened. Since your predictions weren't the actual scores, I don't think you cheated, but I still made a rule that predictions for both legs must be made before the first leg starts. Therefore the only Round 2 predictions I will score for you are:

Puerto Rico-Grenada
Dominica-Canada
Dominican Republic-Belize
Guatemala-Bermuda
St. Kitts and Nevis-El Salvador

You missed Round 1, which was 4.2 percent of the competition, and half of Round 2. Round 2 is 8 percent of the competition. For Round 1 and Round 2 combined you missed 8.2 percent. You can still do well if you predict the rest of the competition.

 

https://docs.google.com/spreadsheets/d/1RGsJdCklTrh12WWERZ3YOUMkOyfqJfU4NgJ0LsMs3x0/edit#gid=0 has the Round 2 first leg prediction results. I'm also going to type some of it here. Here are everybody's points (out of 40), correct results, and exact scores from the 10 first leg games:

1. dinamo_zagreb: 19, 6, 1
2. ChaChaFut: 15, 5, 1
3. ioppolo: 14, 4, 2
4. jagum: 13, 4, 2
Tied 5. SetPeace: 12, 4, 0
Tied 5. bartecky: 12, 4, 1
Tied 7. Sebsasour: 10, 3, 1
Tied 7. italiancbr: 10, 4, 0
Tied 9. IASocFan: 9, 3, 1
Tied 9. drt2k3: 9, 3, 0
Tied 9. Outburstz: 9, 4, 0
Tied 12. PaulieJay: 7, 2, 1
Tied 12. chec: 7, 2, 1
Tied 14. EvanJ: 6, 2, 0
Tied 14. blacksun: 6, 2, 0
Tied 14. zenryan: 6, 2, 1
17. Dr. Gamera: 5, 2, 0
18. barnabyjonez: 4, 1, 1 (only predicted in time for five of the series)

The averages were 9.61 points, 3.17 correct results, and 0.72 exact scores.

Here is a list of exact scores. If you are not listed below, you didn't have any exact scores:

ioppolo: Dominica 0:2 Canada and Curaçao 0:0 Cuba
jagum: Curaçao 0:0 Cuba and Nicaragua 1:0 Suriname
dinamo_zagreb: Dominica 0:2 Canada
ChaChaFut: Dominica 0:2 Canada
bartecky: Aruba 0:2 Barbados
Sebsasour: Aruba 0:2 Barbados
IASocFan: St. Vincent and the Grenadines 2:2 Guyana
PaulieJay: Curacao 0:0 Cuba
chef: Dominica 0:2 Canada
zenryan: Dominica 0:2 Canada
barnabyjonez: Dominica 0:2 Canada

 

You predicted Grenada to win both games (so specifying you predicted Grenada to win the aggregate wasn't necessary), but you sound like you meant to predict each team to win 2-1. Puerto Rico won the first leg, so if you meant to predict them to win the first leg and typed your prediction wrong, it hurt your score. The "2" you typed was next to Grenada for both games.

 

I've updated https://docs.google.com/spreadsheets/d/1RGsJdCklTrh12WWERZ3YOUMkOyfqJfU4NgJ0LsMs3x0/edit#gid=0 with the Round 2 results and I'm posting them here also. Here is everybody's points from score predictions for both legs (out of 80), adjusted points (out of 4 with the top person automatically getting a perfect score), correct results, and exact scores in that order:

1. bartecky: 38, 4.000, 13, 3
2. dinamo_zagreb: 34, 3.579, 12, 1
3. ChaChaFut: 30, 3.158, 10, 2
4. IASocFan: 28, 2.947, 9, 4
Tied 5. blacksun: 27, 2.842, 9, 2
Tied 5. SetPeace: 27, 2.842, 10, 0
Tied 5. jagum: 27, 2.842, 9, 4
Tied 8. EvanJ: 26, 2.737, 9, 2
Tied 8. ioppolo: 26, 2.737, 9, 2
Tied 10. drt2k3: 25, 2.632, 9, 0
Tied 10. chec: 25, 2.632, 8, 4
Tied 12. Sebsasour: 23, 2.421, 7, 3
Tied 12. zenryan: 23, 2.421, 8, 2
14. Dr. Gamera: 22, 2.316, 8, 1
Tied 15. PaulieJay: 21, 2.211, 7, 2
Tied 15. italiancbr: 21, 2.211, 8, 1
17. Outburstz: 20, 2.105, 9, 0
18. barnabyjonez: 18, 1.895 6, 2 (only predicted half the games in time)
Mean: 25.61, 2.696, 8.89, 1.94

Half the people scored on each side of the mean.

There were 22 exact score predictions for second legs. If you are not listed, you didn't have any second leg exact score predictions. This does not include first legs.

EvanJ: Bermuda 0:1 Guatemala and Barbados 1:0 Aruba
blacksun: Canada 4:0 Dominica and Barbados 1:0 Aruba
IASocFan: Grenada 2:0 Puerto Rico, Canada 4:0 Dominica, and Barbados 1:0 Aruba
Dr. Gamera: Canada 4:0 Dominica
ChaChaFut: Grenada 2:0 Puerto Rico
Sebsasour: Canada 4:0 Dominica and Cuba 1:1 Curacao
bartecky: Canada 4:0 Dominica and Cuba 1:1 Curacao
PaulieJay: Canada 4:0 Dominica
jagum: Antigua and Barbuda 4:1 St. Lucia and Canada 4:0 Dominica
chec: Grenada 2:0 Puerto Rico, Canada 4:0 Dominica, and Bermuda 0:1 Guatemala
italiancbr: Canada 4:0 Dominica
zenryan: Canada 4:0 Dominica
barnabyjonez: Belize 3:0 Dominica Republic

Canada-Dominica and El Salvador-St. Kitts and Nevis were the only two games where everybody predicted the correct result. Antigua and Barbuda-St. Lucia (Leg 1) and Guatemala-Bermuda were the only two games where everybody predicted the wrong result. Here is how many points each game's score predictions resulted in:

Canada-Dominica: 60
Dominica-Canada: 52
El Salvador-St. Kitts and Nevis: 48
Barbados-Aruba: 38
Antigua and Barbuda-St. Lucia (Leg 2); 37
Grenada-Puerto Rico: 36
Bermuda-Guatemala: 34
St. Vincent and the Grenadines-Guyana: 25
Puerto Rico-Grenada: 25
Nicaragua-Suriname: 25
Cuaracao-Cuba: 21
Belize-Dominican Republic: 19
Aruba-Barbados: 14
St. Kitts and Nevis-El Salvador: 8
Cuba-Curacao: 8
Guyana-St. Vincent and the Grenadines: 6
Dominican Republic-Belize: 3
Suriname-Nicaragua: 2
Antigua and Barbuda-St. Lucia (Leg 1): 0
Guatemala-Bermuda: 0

The average game resulted in 23.05 points, with half the games on each side of the mean.

Here is how many people were correct and wrong about the winner of each aggregate:

St. Vincent and the Grenadines: 7 out of 17
Antigua and Barbuda: All 17
Grenada: 10 out of 18
Canada: All 18
Belize: 6 out of 18
Guatemala: All 18
Barbados: 14 out of 17
El Salvador: All 18
Curacao: 3 out of 17
Nicaragua: 8 out of 17

I was the only person to be correct about the three most even splits and predict St. Vincent and the Grenadines, Grenada, and Nicaragua to advance.

Here is how many aggregate winners out of 10 everybody predicted and how many points they got from them at 0.4 points each:

1. bartecky: 3.6, 9
Tied 2. EvanJ: 3.2, 8
Tied 2. IASocFan: 3.2, 8
Tied 2. PaulieJay: 3.2, 8
Tied 5. ChaChaFut: 2.8, 7
Tied 5. drt2k3: 2.8, 7
Tied 5. Outburstz: 2.8, 7
Tied 5. jagum: 2.8, 7
Tied 9. blacksun: 2.4, 6
Tied 9. SetPeace: 2.4, 6
Tied 9. Dr. Gamera: 2.4, 6
Tied 9. dinamo_zagreb: 2.4, 6
Tied 9. Sebsasour: 2.4, 6
Tied 9. chec: 2.4, 6
Tied 9. italiancbr: 2.4, 6
Tied 9. zenryan: 2.4, 6
Tied 17. ioppolo: 2.0, 5
Tied 17. barnabyjonez: 2.0, 5 (got all 5 correct and predicted 5 of them too late to count)

The average was 6.61 correct aggregate winners and 2.64 points.

 

I just decided to change how I score the aggregate winners component, which is the last list in my previous post. I'm retroactively applying the new method to Round 1, but since many people got every aggregate winner in Round 1 correct, it doesn't change anything for Round 1. For Round 2 nobody got all 10 winners correct. Here are the adjusted scores out of 4 points for Round 2 aggregate winners:

1. bartecky: 4.000
Tied 2. EvanJ: 3.556
Tied 2. IASocFan: 3.556
Tied 2. PaulieJay: 3.556
Tied 5. ChaChaFut: 3.111
Tied 5. drt2k3: 3.111
Tied 5. Outburstz: 3.111
Tied 5. jagum: 3.111
Tied 9. blacksun: 2.667
Tied 9. SetPeace: 2.667
Tied 9. Dr. Gamera: 2.667
Tied 9. dinamo_zagreb: 2.667
Tied 9. Sebsasour: 2.667
Tied 9. chec: 2.667
Tied 9. italiancbr: 2.667
Tied 9. zenryan: 2.667
Tied 17. ioppolo: 2.222
Tied 17. barnabyjonez: 2.222 (predicted some of them too late to count)

Here are the standings for Round 2 score predictions and aggregate winner predictions combined out of a possible 8 points (barnabyjonez predicted half of Round 2):

1. bartecky: 8.000
2. IASocFan: 6.503
3. EvanJ: 6.293
4. ChaChaFut: 6.269
5. dinamo_zagreb: 6.246
6. jagum: 5.953
7. PaulieJay: 5.767
8. drt2k3: 5.743
Tied 9. blacksun: 5.509
Tied 9. SetPeace: 5.509
11. chec: 5.299
12. Outburtsz: 5.216
Tied 13, Sebsasour: 5.088
Tied 13. zenryan: 5.088
14. Dr. Gamera: 4.983
16. ioppolo: 4.959
17. italiancbr: 4.878
18. barnabyjonez: 4.117 (only predicted some of the round)

Here are the cumulative standings out of a possible 12.2 points:

1. EvanJ: 10.175
2. bartecky: 9.836*
3. ChaChaFut: 9.769
4. blacksun: 9.709
5. jagum: 9.580
6. drt2k3: 9.434
7. dinamo_zagreb: 9.191
8. Sebsasour: 8.906
9. PaulieJay: 8.822
10. Dr. Gamera: 8.738
11. chec: 8.735
12. ioppolo: 8.586
13. italiancbr: 8.442
14. Outburstz: 7.671
15. IASocFan: 6.503*
16. SetPeace: 5.509*
17. zenryan: 5.088*
18. barnabyjonez: 4.117*

A * means the person did not predict the whole competition.

I've also decided to put everybody's predictions into Google Docs spreadsheets, but to only put people's scores here. Whenever I made a change to the spreadsheet on my computer, I had to change it on Google Docs also, and it was getting annoying, so I stopped.

 

I'm not letting people other than me edit the Google spreadsheet. For one thing, I don't want somebody to be able to edit somebody else's predictions or post predictions under somebody else's username.

 

Thank you. Even with an online form somebody could lie in a username field. Typing the predictions into my spreadsheet is manageable. If 100 people predicted each round then I would want an alternate method. Excel and my calculator do the math for me. Here's what's most complicated and/or annoying:

1. Weighting the different parts of the competition to add up to 100, with different rounds having different amounts of games and some rounds being group stages. I also want each part to be worth something that doesn't require too many decimal places and is not a repeating decimal.

2. People who post predictions for each game of a Group Stage but don't post standings and make me add it up myself, or people who post standings that don't match the results of their game predictions.

3. Working with points off in the standings, because this is a number where the lowest is the best.

4. Being careful about how I evaluate score predictions.

5. Checking what's wrong when two numbers don't match. For example, the sum of the points from score predictions for all 18 people combined equals the sum of the points from each game's score predictions for all 20 games combined.

I think yesterday I came up with a good method because Part X being 10% of the contest and Part Y being 20% doesn't mean that Part Y will provide twice as many points as Part X because the percent of possible points that were earned could vary from part to part. To put it another way, let's say the same amount of students take two tests, one with 10 questions and one with 20 questions, and each answer is right or wrong (no partial credit). There is no guarantee that the test with 20 questions will have about twice as many correct answers as the test with 10 questions.

I have attached a file about what percent of the contest each round is worth.

 

For Rounds 4 and 5, I add up how many points everybody was off for all the teams in the group combined in addition to score predictions and predicting what order the teams finish the standings in. Rounds 4 and 5 have double the games than you posted. I'm not changing how I evaluate the predictions.