Only 12 days to the World Cup, time to get your sophisticated predictions in!

https://futbolmetrix.wordpress.com/2022/11/08/world-cup-2022-sophisticated-prediction-contest/

On Tuesday, November 8, 2022 at 6:59:41 AM UTC, Futbolmetrix wrote:
> Only 12 days to the World Cup, time to get your sophisticated predictions in!
>
> https://futbolmetrix.wordpress.com/2022/11/08/world-cup-2022-sophisticated-prediction-contest/

How does this work? and why is it probabilities and not definite predictions? etc.

And what's the best way of doing it? I'm not sure if I can remember the advice I was given 4 years ago. Fill it in team by team I think? (I think I'd been trying to do it round by round and found it almost impossible.)

(And what's the best thing to do when, as usual, my computer won't open the spreadsheet?)

On Wednesday, November 16, 2022 at 2:10:51 PM UTC+1, Mark wrote:
> On Tuesday, November 8, 2022 at 6:59:41 AM UTC, Futbolmetrix wrote:
> > Only 12 days to the World Cup, time to get your sophisticated predictions in!
> >
> > https://futbolmetrix.wordpress.com/2022/11/08/world-cup-2022-sophisticated-prediction-contest/
>
> How does this work? and why is it probabilities and not definite predictions? etc.

Because that would be unsophisticated. This is the Sophisticated Prediction Contest.

Ciao,
Werner

On Thursday, November 17, 2022 at 9:32:31 AM UTC-5, Werner Pichler wrote:
> On Wednesday, November 16, 2022 at 2:10:51 PM UTC+1, Mark wrote:
> > On Tuesday, November 8, 2022 at 6:59:41 AM UTC, Futbolmetrix wrote:
> > > Only 12 days to the World Cup, time to get your sophisticated predictions in!
> > >
> > > https://futbolmetrix.wordpress.com/2022/11/08/world-cup-2022-sophisticated-prediction-contest/
> >
> > How does this work? and why is it probabilities and not definite predictions? etc.
> Because that would be unsophisticated. This is the Sophisticated Prediction Contest.

Mark's question is a good one, especially given the nature of a fairly complicated matrix such as the one for the World Cup. The strategy that I use goes something like this:

1) First fill in the group probabilities: This is relatively easy, you just need 4 numbers that add up to 2. So, for example, you could have, for the probability of *elimination* in Round 1, the following:

Qatar 0.5
Netherlands 0.1
Ecuador 0.6
Senegal 0.8

2) Then fill in each row separately and make it up to one. For example:

p_R1 p_R16 p_QF p_SF p_LF p_WF
Qatar 0.5 0.3 0.15 0.03 0.01 0.01

I use a couple of rules of thumb for this step:
a) The numbers in a row should be either monotonically decreasing (as in the example above) or hump shaped (*), as in:

p_R1 p_R16 p_QF p_SF p_LF p_WF
Netherlands 0.1 0.35 0.35 0.12 0.05 0.03

(*) True for most teams with the exception of heavy favorites, who are more likely to win the final than lose it.

b) If you have teams that you think are of similar strength, use the same row. For example, I may think that Portugal is about the same strength as the Netherlands, in which case I just copy paste the Netherlands row into the Portugal row.

3) Once you have completed all the rows, this is where it gets difficult, because the numbers in the bottom row don't add up to what they should be. At this point, just do a lot of manual tinkering until things you get things right. Often, this takes the following form:

- I have underestimate by quite a lot p_WF, and overestimated quite a lot p_QF. So start taking off some decimal points from each team's p_QF and add it to p_WF (or just for the teams for which it makes sense).

Usually, this whole process takes about 20 minutes.

There are of course other possibilities:
1. Look online for a source that makes a probabilistic forecast, and copy that as is, or with slight tinkering.
2. Run your own computer model. They are a dime a dozen, nowadays...

On Thursday, November 17, 2022 at 6:50:05 PM UTC+1, Futbolmetrix wrote:
> On Thursday, November 17, 2022 at 9:32:31 AM UTC-5, Werner Pichler wrote:
> > On Wednesday, November 16, 2022 at 2:10:51 PM UTC+1, Mark wrote:
> > > On Tuesday, November 8, 2022 at 6:59:41 AM UTC, Futbolmetrix wrote:
> > > > Only 12 days to the World Cup, time to get your sophisticated predictions in!
> > > >
> > > > https://futbolmetrix.wordpress.com/2022/11/08/world-cup-2022-sophisticated-prediction-contest/
> > >
> > > How does this work? and why is it probabilities and not definite predictions? etc.
> > Because that would be unsophisticated. This is the Sophisticated Prediction Contest.
> Mark's question is a good one, especially given the nature of a fairly complicated matrix such as the one for the World Cup. The strategy that I use goes something like this:
>
> 1) First fill in the group probabilities: This is relatively easy, you just need 4 numbers that add up to 2. So, for example, you could have, for the probability of *elimination* in Round 1, the following:
>
> Qatar 0.5
> Netherlands 0.1
> Ecuador 0.6
> Senegal 0.8
>
> 2) Then fill in each row separately and make it up to one. For example:
>
> p_R1 p_R16 p_QF p_SF p_LF p_WF
> Qatar 0.5 0.3 0.15 0.03 0.01 0.01
>
> I use a couple of rules of thumb for this step:
> a) The numbers in a row should be either monotonically decreasing (as in the example above) or hump shaped (*), as in:
>
> p_R1 p_R16 p_QF p_SF p_LF p_WF
> Netherlands 0.1 0.35 0.35 0.12 0.05 0.03
>
> (*) True for most teams with the exception of heavy favorites, who are more likely to win the final than lose it.

Yeah, I usually regard any final match as basically a toss-up, so generally assign the same probabilities to both winning and losing the final.

I don't know how common that is, and whether or not that has hurt my chances in these competitions, but it has made filling out the matrix a tad easier.

Ciao,
Werner

> b) If you have teams that you think are of similar strength, use the same row. For example, I may think that Portugal is about the same strength as the Netherlands, in which case I just copy paste the Netherlands row into the Portugal row.
>
>
> 3) Once you have completed all the rows, this is where it gets difficult, because the numbers in the bottom row don't add up to what they should be. At this point, just do a lot of manual tinkering until things you get things right. Often, this takes the following form:
>
> - I have underestimate by quite a lot p_WF, and overestimated quite a lot p_QF. So start taking off some decimal points from each team's p_QF and add it to p_WF (or just for the teams for which it makes sense).
>
> Usually, this whole process takes about 20 minutes.
>
> There are of course other possibilities:
> 1. Look online for a source that makes a probabilistic forecast, and copy that as is, or with slight tinkering.
> 2. Run your own computer model. They are a dime a dozen, nowadays...

On Thursday, November 17, 2022 at 2:26:21 PM UTC-5, Werner Pichler wrote:

> Yeah, I usually regard any final match as basically a toss-up, so generally assign the same probabilities to both winning and losing the final.

I used to do that too, but I have changed that in more recent SophCons. I have no idea if this has affected my relative standings.

On 2022-11-17 12:26, Werner Pichler wrote:
> On Thursday, November 17, 2022 at 6:50:05 PM UTC+1, Futbolmetrix wrote:
>> On Thursday, November 17, 2022 at 9:32:31 AM UTC-5, Werner Pichler wrote:
>>> On Wednesday, November 16, 2022 at 2:10:51 PM UTC+1, Mark wrote:
>>>> On Tuesday, November 8, 2022 at 6:59:41 AM UTC, Futbolmetrix wrote:
>>>>> Only 12 days to the World Cup, time to get your sophisticated predictions in!
>>>>>
>>>>> https://futbolmetrix.wordpress.com/2022/11/08/world-cup-2022-sophisticated-prediction-contest/
>>>>
>>>> How does this work? and why is it probabilities and not definite predictions? etc.
>>> Because that would be unsophisticated. This is the Sophisticated Prediction Contest.
>> Mark's question is a good one, especially given the nature of a fairly complicated matrix such as the one for the World Cup. The strategy that I use goes something like this:
>>
>> 1) First fill in the group probabilities: This is relatively easy, you just need 4 numbers that add up to 2. So, for example, you could have, for the probability of *elimination* in Round 1, the following:
>>
>> Qatar 0.5
>> Netherlands 0.1
>> Ecuador 0.6
>> Senegal 0.8
>>
>> 2) Then fill in each row separately and make it up to one. For example:
>>
>> p_R1 p_R16 p_QF p_SF p_LF p_WF
>> Qatar 0.5 0.3 0.15 0.03 0.01 0.01
>>
>> I use a couple of rules of thumb for this step:
>> a) The numbers in a row should be either monotonically decreasing (as in the example above) or hump shaped (*), as in:
>>
>> p_R1 p_R16 p_QF p_SF p_LF p_WF
>> Netherlands 0.1 0.35 0.35 0.12 0.05 0.03
>>
>> (*) True for most teams with the exception of heavy favorites, who are more likely to win the final than lose it.
>
> Yeah, I usually regard any final match as basically a toss-up, so generally assign the same probabilities to both winning and losing the final.
>
> I don't know how common that is, and whether or not that has hurt my chances in these competitions, but it has made filling out the matrix a tad easier.
>
> Ciao,
> Werner
>
>
I do the same. Did not always do so in the past. While I would be able
to predict different probabilities for the finalists if I knew they were
Brazil and England, for example, but in the absence of any idea who the
opponents are going to be, I find it easier to consider them a toss-up.
Especially given the chance it might go to PKS.

Em terça-feira, 8 de novembro de 2022 às 03:59:41 UTC-3, Futbolmetrix escreveu:
> Only 12 days to the World Cup, time to get your sophisticated predictions in!
>
> https://futbolmetrix.wordpress.com/2022/11/08/world-cup-2022-sophisticated-prediction-contest/

538 have published their predictions:

https://projects.fivethirtyeight.com/2022-world-cup-predictions/

Tchau!
Jesus Petry

On Friday, November 18, 2022 at 2:56:36 PM UTC-5, jesus...@gmail.com wrote:
> Em terça-feira, 8 de novembro de 2022 às 03:59:41 UTC-3, Futbolmetrix escreveu:
> > Only 12 days to the World Cup, time to get your sophisticated predictions in!
> >
> > https://futbolmetrix.wordpress.com/2022/11/08/world-cup-2022-sophisticated-prediction-contest/
> 538 have published their predictions:
>
> https://projects.fivethirtyeight.com/2022-world-cup-predictions/

Yes, included it already. List of participants so far:

Spreadsheet number Name
S01 Agnostic
S02 KU Leuven
S03 Swiss Football Data
S04 goalprojection
S05 KonakaLab
S06 InPlayFootballTips
S07 Jesus Petry
S08 Luke Benz
S09 FiveThirtyEight
S10 Futbolmetrix
S11 Betfair
s12 AIrgentina
s13 MH
s14 Zander (Tournament Structure)

I have done the calculation for all my usual bots and more. Haven't uploaded them yet. But expect to see the following:
Elo
Transfermarkt
ClubElo/ClubTM <-- defending champion
econ0
econ1
-----------------------

Then I will probably also add the following

scrabble (based on each team's official name scrabble score, in strong competition with Aggie and the econs for the dumbest prediction)
gk <-- best goalkeeper (suggested by milivella)
def <-- best four defenders (suggested by milivella)
mid <-- best four midfielders (suggested by milivella)
fwd <-- best 2 forwards (suggested by milivella)

And maybe these ones too:

adjusted tm: adjusts tm value for age and position (a 50m 35-year old Messi is probably better than a 50m 25 year old Richarlison)
age
experience (number of caps)
goal experience (number of goals)
q1bias (the share of players born in quarter 1 of the year out of the total number of players born in q1 or q4 - I think this is an interesting measure of a nation's scouting ability at youth level: are you really able to pick the best talents, or do you just invest in the slightly older 12 and 13 year olds that have a physical advantage over their peers in their same age category?)

On Friday, November 18, 2022 at 10:35:45 PM UTC, Futbolmetrix wrote:
> On Friday, November 18, 2022 at 2:56:36 PM UTC-5, jesus...@gmail.com wrote:
> > Em terça-feira, 8 de novembro de 2022 às 03:59:41 UTC-3, Futbolmetrix escreveu:
> > > Only 12 days to the World Cup, time to get your sophisticated predictions in!
> > >
> > > https://futbolmetrix.wordpress.com/2022/11/08/world-cup-2022-sophisticated-prediction-contest/
> > 538 have published their predictions:
> >
> > https://projects.fivethirtyeight.com/2022-world-cup-predictions/
> Yes, included it already. List of participants so far:
>
> Spreadsheet number Name
> S01 Agnostic
> S02 KU Leuven
> S03 Swiss Football Data
> S04 goalprojection
> S05 KonakaLab
> S06 InPlayFootballTips
> S07 Jesus Petry
> S08 Luke Benz
> S09 FiveThirtyEight
> S10 Futbolmetrix
> S11 Betfair
> s12 AIrgentina
> s13 MH
> s14 Zander (Tournament Structure)
>
> I have done the calculation for all my usual bots and more. Haven't uploaded them yet. But expect to see the following:
> Elo
> Transfermarkt
> ClubElo/ClubTM <-- defending champion
> econ0
> econ1
> -----------------------
>
> Then I will probably also add the following
>
> scrabble (based on each team's official name scrabble score, in strong competition with Aggie and the econs for the dumbest prediction)
> gk <-- best goalkeeper (suggested by milivella)
> def <-- best four defenders (suggested by milivella)
> mid <-- best four midfielders (suggested by milivella)
> fwd <-- best 2 forwards (suggested by milivella)
>
> And maybe these ones too:
>
> adjusted tm: adjusts tm value for age and position (a 50m 35-year old Messi is probably better than a 50m 25 year old Richarlison)
> age
> experience (number of caps)
> goal experience (number of goals)
> q1bias (the share of players born in quarter 1 of the year out of the total number of players >born in q1 or q4 - I think this is an interesting measure of a nation's scouting ability at youth >level: are you really able to pick the best talents, or do you just invest in the slightly older 12 >and 13 year olds that have a physical advantage over their peers in their same age category?)

Working through my first SophnCon entry.

The group stage probabilities all add up to 2.

Each team’s total probability adds up to 1.

So far so good.

But some of the columns below say 1.9 (for example) when the spreadsheet says they should add up to 2.

Is this a problem? Out of interest what is the reasoning for those columns needing to be 2?

I won’t have a chance to amend this tomorrow so could I please send mine over with the above errors subject to being tinkered with to get the columns to balance after Qatar v Ecuador?

RM

..

On Saturday, November 19, 2022 at 2:10:10 PM UTC-5, Real Mardin wrote:
> Working through my first SophnCon entry.
>
> The group stage probabilities all add up to 2.

Good!

>
> Each team’s total probability adds up to 1.
>
> So far so good.

Even better!

> But some of the columns below say 1.9 (for example) when the spreadsheet says they should add up to 2.
>
> Is this a problem? Out of interest what is the reasoning for those columns needing to be 2?

Yes, it is a bit of a problem. Because the round of 16 is crossed (A1 plays B2 and B1 plays A2), there have to be exactly 2 teams that qualify to the QFs out of each pair of groups (A-B, C-D, E-F, and G-H).

> I won’t have a chance to amend this tomorrow so could I please send mine over with the above errors subject to being tinkered with to get the columns to balance after Qatar v Ecuador?

Hmm...I'll see what I can do. But I'm OK if you send in your predictions a bit late, as long as you pinky swear that you did not
let the results of the first match influence your choices.

On Saturday, November 19, 2022 at 8:44:20 PM UTC, Futbolmetrix wrote:
> On Saturday, November 19, 2022 at 2:10:10 PM UTC-5, Real Mardin wrote:
> > Working through my first SophnCon entry.
> >
> > The group stage probabilities all add up to 2.
> Good!
> >
> > Each team’s total probability adds up to 1.
> >
> > So far so good.
> Even better!
> > But some of the columns below say 1.9 (for example) when the spreadsheet says they should add up to 2.
> >
> > Is this a problem? Out of interest what is the reasoning for those columns needing to be 2?

> Yes, it is a bit of a problem. Because the round of 16 is crossed (A1 plays B2 and B1 plays A2), there have to be exactly 2 teams that qualify to the QFs out of each pair of groups (A-B, C-D, E-F, and G-H).

Right, that makes sense now. Thanks for explaining.

> > I won’t have a chance to amend this tomorrow so could I please send mine over with the above errors subject to being tinkered with to get the columns to balance after Qatar v Ecuador?
> Hmm...I'll see what I can do. But I'm OK if you send in your predictions a bit late, as long as you pinky swear that you did not
> let the results of the first match influence your choices.

Thanks, I'll see what I can do!

If nothing else working out these probabilities has made me realise (albeit it too late) what a folly some of my World Cup bets have been!

RM

RM

On Saturday, November 19, 2022 at 9:22:53 PM UTC, Real Mardin wrote:
> On Saturday, November 19, 2022 at 8:44:20 PM UTC, Futbolmetrix wrote:
> > On Saturday, November 19, 2022 at 2:10:10 PM UTC-5, Real Mardin wrote:
> > > Working through my first SophnCon entry.
> > >
> > > The group stage probabilities all add up to 2.
> > Good!
> > >
> > > Each team’s total probability adds up to 1.
> > >
> > > So far so good.
> > Even better!
> > > But some of the columns below say 1.9 (for example) when the spreadsheet says they should add up to 2.
> > >
> > > Is this a problem? Out of interest what is the reasoning for those columns needing to be 2?
>
> > Yes, it is a bit of a problem. Because the round of 16 is crossed (A1 plays B2 and B1 plays A2), there have to be exactly 2 teams that qualify to the QFs out of each pair of groups (A-B, C-D, E-F, and G-H).
> Right, that makes sense now. Thanks for explaining.
> > > I won’t have a chance to amend this tomorrow so could I please send mine over with the above errors subject to being tinkered with to get the columns to balance after Qatar v Ecuador?
> > Hmm...I'll see what I can do. But I'm OK if you send in your predictions a bit late, as long as you pinky swear that you did not
> > let the results of the first match influence your choices.
> Thanks, I'll see what I can do!

Unfortunately not factoring in that the low columns are the total probability for each bracket means I've set about this the wrong way correcting the entry is now going to take too long even taking into account a possible post opening game extension.

For now I'm going to have to concede that I'm only two thirds sophisticated :(

It's great that the Soph Con honours former rss'er Ruud aka John De Lange, for my part I shall honour him by being outspoken and by speaking up when it's right to do so, which are my abiding memories of him as a rss'er.

RM

Futbolmetrix:

I HATE disagreeing with you, but...

> scrabble (based on each team's official name scrabble score, in strong competition with Aggie and the econs for the dumbest prediction)
> gk <-- best goalkeeper (suggested by milivella)
> def <-- best four defenders (suggested by milivella)
> mid <-- best four midfielders (suggested by milivella)
> fwd <-- best 2 forwards (suggested by milivella)

I suggested just one rating! (best XI = best gk + best 4 defs + best 4
mids + best 2 fwds)

Your interpretation is more interesting, though.

> adjusted tm: adjusts tm value for age and position

I also suggested this... (and don't reply that you added "and position",
as I suggested that years ago, in the context of FS predictions!)

> (a 50m 35-year old Messi is probably better than a 50m 25 year old Richarlison)

Maybe I shouldn't claim the autorship of this idea, after all...

Nitpicking aside, thanks for the credit. Above all, your multi-decade
job as organizer of SPCs and designer of very interesting models
deserves a standing ovation!

(I hope someone else rises and applauds, as I am contracutally
prohibited from doing so.)

--
Cheers
milivella

On Friday, November 25, 2022 at 8:14:22 AM UTC-5, milivella wrote:
> Futbolmetrix:
>
> I suggested just one rating! (best XI = best gk + best 4 defs + best 4
> mids + best 2 fwds)
>
> Your interpretation is more interesting, though.

To be honest, I'm not so sure the "best n" is the right approach. There will be subs, tactical changes, etc. Maybe it should have been just the "best DF", "best MF" etc.

> > (a 50m 35-year old Messi is probably better than a 50m 25 year old Richarlison)
> Maybe I shouldn't claim the autorship of this idea, after all...

Yeah, yeah, enjoy your 15 minutes of glory. Let's not forget that for about 70 minutes yesterday Richarlison looked like a cross between Serginho 1982, Fred 2014 and Gabriel Jesus 2018 in terms of sheer ineptitude.

On Tuesday, November 8, 2022 at 1:59:41 AM UTC-5, Futbolmetrix wrote:
> Only 12 days to the World Cup, time to get your sophisticated predictions in!
>
> https://futbolmetrix.wordpress.com/2022/11/08/world-cup-2022-sophisticated-prediction-contest/

Standings after MD1: https://futbolmetrix.wordpress.com/2022/11/25/wc2022-sophcon-standings-after-day-1/
Link to Master spreadsheet: https://futbolmetrix.wordpress.com/2022/11/25/wc2022-sophcon-the-master-spreadsheet/

If you don't want to open the link, here are the standings:

Standings after the first match day:
1 AIrgentina -20.432
2 MH -20.496
3 Silver Aurora -20.952
4 FiveThirtyEight -21.128
5 Bookmaker Consensus Model (by AZ) -21.181
6 El Pais -21.303
7 Midfielders (by FM) -21.319
8 Jesper -21.328
9 goalprojection -21.457
10 Collective Consciousness -21.467
11 tm (by FM) -21.469
12 Fantomen -21.496
13 Alessandro Riolo -21.539
14 Stephen K -21.715
15 Elo (by FM) -21.862
16 Patrick Flynn -21.954
17 Luke Benz -21.961
18 Opta Analyst -21.994
19 Betfair -22.022
20 Scrabble (by FM) -22.133
21 Agnostic -22.181
22 FoG -22.181
23 FIFA (by FM) -22.252
24 KU Leuven -22.376
25 Footballgenius -22.399
26 KonakaLab -22.527
27 TransferMarkt - Adj (by FM) -22.535
28 Strongest link (by FM) -22.574
29 Econ1 (by FM) -22.581
30 Hybrid Machine Learning (by AZ) -22.591
31 Middle link (by FM) -22.602
32 ClubElo/ClubTM (by FM) -22.661
33 TransferMarkt (by FM) -22.758
34 Defenders (by FM) -22.789
35 cuca -22.889
36 Werner Pichler -22.898
37 Zander (Tournament Structure) -22.909
38 NT-ELO predictions (by Ed) -23.073
39 Jesus Petry -23.132
40 Volkan Riolo -23.220
41 InPlayFootballTips -23.309
42 dopemodels.info -23.476
43 Econ0 (by FM) -23.546
44 Weakest link (by FM) -23.673
45 Lléo -24.116
46 Dan Levy -24.255
47 Goalkeepers (by FM) -24.485
48 Historic Match Model (by AZ) -24.559
49 Futbolmetrix -25.318
50 Q1_Bias (by FM) -25.489
51 Keanu J -25.690
52 Swiss Football Data -25.818
53 Forwards (by FM) -27.036
54 Goal Experience (by FM) -27.339
55 Experience (by FM) -27.885
56 Mark -32.106
57 Anders T -33.615

Quoting Futbolmetrix in rec.sport.soccer:

>57 Anders T -33.615

https://youtu.be/FXO0GIIy3xU !

--
Manchester United FC - CHAMPIONS
Latest: England '13 (20th) Europa '17, UEFA '08, World '08

Futbolmetrix:

> On Friday, November 25, 2022 at 8:14:22 AM UTC-5, milivella wrote:
>> Futbolmetrix:
>>
>> I suggested just one rating! (best XI = best gk + best 4 defs + best 4
>> mids + best 2 fwds)
>>
>> Your interpretation is more interesting, though.
>
> To be honest, I'm not so sure the "best n" is the right approach. There will be subs, tactical changes, etc. Maybe it should have been just the "best DF", "best MF" etc.

It may be!

I guess the "best n" approach settles the question (so to speak) about
"defense wins trophies" vs. "midfield wins trophies" vs. etc.

>>> (a 50m 35-year old Messi is probably better than a 50m 25 year old Richarlison)
>> Maybe I shouldn't claim the autorship of this idea, after all...
>
> Yeah, yeah, enjoy your 15 minutes of glory.

Still more than Kane+Mueller's 0 minutes of glory...

> Let's not forget that for about 70 minutes yesterday Richarlison looked like a cross between Serginho 1982, Fred 2014 and Gabriel Jesus 2018 in terms of sheer ineptitude.

Before scoring, a player hasn't scored. More news at 11.

--
Cheers
milivella

On Thursday, November 17, 2022 at 7:50:05 PM UTC+2, Futbolmetrix wrote:
> On Thursday, November 17, 2022 at 9:32:31 AM UTC-5, Werner Pichler wrote:
> > On Wednesday, November 16, 2022 at 2:10:51 PM UTC+1, Mark wrote:
> > > On Tuesday, November 8, 2022 at 6:59:41 AM UTC, Futbolmetrix wrote:
> > > > Only 12 days to the World Cup, time to get your sophisticated predictions in!
> > > >
> > > > https://futbolmetrix.wordpress.com/2022/11/08/world-cup-2022-sophisticated-prediction-contest/
> > >
> > > How does this work? and why is it probabilities and not definite predictions? etc.
> > Because that would be unsophisticated. This is the Sophisticated Prediction Contest.
> Mark's question is a good one, especially given the nature of a fairly complicated matrix such as the one for the World Cup. The strategy that I use goes something like this:
>
> 1) First fill in the group probabilities: This is relatively easy, you just need 4 numbers that add up to 2. So, for example, you could have, for the probability of *elimination* in Round 1, the following:
>
> Qatar 0.5
> Netherlands 0.1
> Ecuador 0.6
> Senegal 0.8
>
> 2) Then fill in each row separately and make it up to one. For example:
>
> p_R1 p_R16 p_QF p_SF p_LF p_WF
> Qatar 0.5 0.3 0.15 0.03 0.01 0.01
>
> I use a couple of rules of thumb for this step:
> a) The numbers in a row should be either monotonically decreasing (as in the example above) or hump shaped (*), as in:
>
> p_R1 p_R16 p_QF p_SF p_LF p_WF
> Netherlands 0.1 0.35 0.35 0.12 0.05 0.03
>
> (*) True for most teams with the exception of heavy favorites, who are more likely to win the final than lose it.
>
> b) If you have teams that you think are of similar strength, use the same row. For example, I may think that Portugal is about the same strength as the Netherlands, in which case I just copy paste the Netherlands row into the Portugal row.
>
>
> 3) Once you have completed all the rows, this is where it gets difficult, because the numbers in the bottom row don't add up to what they should be. At this point, just do a lot of manual tinkering until things you get things right. Often, this takes the following form:
>
> - I have underestimate by quite a lot p_WF, and overestimated quite a lot p_QF. So start taking off some decimal points from each team's p_QF and add it to p_WF (or just for the teams for which it makes sense).
>
> Usually, this whole process takes about 20 minutes.
>
> There are of course other possibilities:
> 1. Look online for a source that makes a probabilistic forecast, and copy that as is, or with slight tinkering.
> 2. Run your own computer model. They are a dime a dozen, nowadays...

“1) First fill in the group probabilities: This is relatively easy, you just need 4 numbers that add up to 2. So, for example, you could have, for the probability of *elimination* in Round 1, the following:

Qatar 0.5
Netherlands 0.1
Ecuador 0.6
Senegal 0.8”

Futitser:

Assiomatico, the hardest time I’ve had: Convincing people that there ain’t NO absolute certainty. As I put it: “Absolute certainty is absolute absurdity”. The Fundamental Formula of Gambling (FFG) proves it beyond every shred of reasonable doubt.

Still, you step one pace over the fundamental absurdity. You “calculate” a “probability” of 0.5+0.1+0.6+0.8. “Result”: 2.0. Probability never reaches 1.0, amico! If it would reach 1.0, then there would be infinity, or perfection, … or absolute certainty! See it? You go 2.0 times beyond ABSURDITY!

• https://saliu.com/Saliu2.htm
• Fundamental Formula of Gambling (FFG): Theory of Probability, Mathematics, Degree of Certainty, Chance.

On Saturday, November 26, 2022 at 6:30:17 PM UTC, ions...@gmail.com wrote:

> Assiomatico, the hardest time I’ve had: Convincing people that there ain’t NO absolute certainty. As I put it: “Absolute certainty is absolute absurdity”. The Fundamental Formula of Gambling (FFG) proves it beyond every shred of reasonable doubt.
>
> Still, you step one pace over the fundamental absurdity. You “calculate” a “probability” of 0.5+0.1+0.6+0.8. “Result”: 2.0. Probability never reaches 1.0, amico! If it would reach 1.0, then there would be infinity, or perfection, … or absolute certainty! See it? You go 2.0 times beyond ABSURDITY!
>
> • https://saliu.com/Saliu2.htm
> • Fundamental Formula of Gambling (FFG): Theory of Probability, Mathematics, Degree of Certainty, Chance.

Is it not an absolute certainty that Qatar won't win the World Cup?

On Friday, November 25, 2022 at 8:31:20 AM UTC-5, Futbolmetrix wrote:
> On Tuesday, November 8, 2022 at 1:59:41 AM UTC-5, Futbolmetrix wrote:
> > Only 12 days to the World Cup, time to get your sophisticated predictions in!
> >
> > https://futbolmetrix.wordpress.com/2022/11/08/world-cup-2022-sophisticated-prediction-contest/
> Standings after MD1: https://futbolmetrix.wordpress.com/2022/11/25/wc2022-sophcon-standings-after-day-1/
> Link to Master spreadsheet: https://futbolmetrix.wordpress.com/2022/11/25/wc2022-sophcon-the-master-spreadsheet/
>

Standings after Match Day 2:

1 Jesus Petry -19.243 <--- Up RSS!
2 Defenders (by FM) -19.547 <-- Defense wins championships! (or at least does well after 2 match days!)
3 Simon G -20.096
4 Volkan Riolo -20.273
5 Luke Benz -20.300
6 FiveThirtyEight -20.326
7 Collective Consciousness -20.349
8 Bookmaker Consensus Model (by AZ) -20.379
9 Dan Levy -20.399
10 Midfielders (by FM) -20.448
11 cuca -20.473
12 KU Leuven -20.490
13 Patrick Flynn -20.523
14 tm (by FM) -20.568
15 El Pais -20.602
16 Zander (Tournament Structure) -20.647
17 Middle link (by FM) -20.718
18 Stephen K -20.721
19 TransferMarkt - Adj (by FM) -20.820
20 goalprojection -20.832
21 Silver Aurora -20.835
22 Jesper -20.915
23 TransferMarkt (by FM) -20.948
24 Futbolmetrix -20.977
25 Betfair -21.042
26 Elo (by FM) -21.251
27 ClubElo/ClubTM (by FM) -21.319
28 Opta Analyst -21.325
29 Strongest link (by FM) -21.337
30 Alessandro Riolo -21.539
31 Goalkeepers (by FM) -21.548
32 Hybrid Machine Learning (by AZ) -21.645
33 Econ1 (by FM) -21.787
34 Q1_Bias (by FM) -21.908
35 KonakaLab -21.936
36 AIrgentina -21.996
37 FoG -22.181
38 Agnostic -22.181
39 Weakest link (by FM) -22.211
40 InPlayFootballTips -22.255
41 Anders T -22.344
42 NT-ELO predictions (by Ed) -22.516
43 dopemodels.info -22.560
44 Historic Match Model (by AZ) -22.786
45 Keanu J -22.875
46 Lléo -22.906
47 MH -22.965
48 Econ0 (by FM) -23.078
49 Footballgenius -23.299
50 FIFA (by FM) -23.371
51 Fantomen -23.403
52 Werner Pichler -23.602
53 Mark -23.983
54 Forwards (by FM) -24.257
55 Swiss Football Data -25.209
56 Scrabble (by FM) -27.891
57 Goal Experience (by FM) -30.514
58 Experience (by FM) -36.811

Also available here: https://futbolmetrix.wordpress.com/2022/11/28/wc2022-sophcon-standings-after-match-day-2/

Em segunda-feira, 28 de novembro de 2022 às 19:52:39 UTC-3, Futbolmetrix escreveu:
> On Friday, November 25, 2022 at 8:31:20 AM UTC-5, Futbolmetrix wrote:
> > On Tuesday, November 8, 2022 at 1:59:41 AM UTC-5, Futbolmetrix wrote:
> > > Only 12 days to the World Cup, time to get your sophisticated predictions in!
> > >
> > > https://futbolmetrix.wordpress.com/2022/11/08/world-cup-2022-sophisticated-prediction-contest/
> > Standings after MD1: https://futbolmetrix.wordpress.com/2022/11/25/wc2022-sophcon-standings-after-day-1/
> > Link to Master spreadsheet: https://futbolmetrix.wordpress.com/2022/11/25/wc2022-sophcon-the-master-spreadsheet/
> >
> Standings after Match Day 2:
>
> 1 Jesus Petry -19.243 <--- Up RSS!
> 2 Defenders (by FM) -19.547 <-- Defense wins championships! (or at least does well after 2 match days!)
> 3 Simon G -20.096
> 4 Volkan Riolo -20.273
> 5 Luke Benz -20.300
> 6 FiveThirtyEight -20.326
> 7 Collective Consciousness -20.349
> 8 Bookmaker Consensus Model (by AZ) -20.379
> 9 Dan Levy -20.399
> 10 Midfielders (by FM) -20.448
> 11 cuca -20.473
> 12 KU Leuven -20.490
> 13 Patrick Flynn -20.523
> 14 tm (by FM) -20.568
> 15 El Pais -20.602
> 16 Zander (Tournament Structure) -20.647
> 17 Middle link (by FM) -20.718
> 18 Stephen K -20.721
> 19 TransferMarkt - Adj (by FM) -20.820
> 20 goalprojection -20.832
> 21 Silver Aurora -20.835
> 22 Jesper -20.915
> 23 TransferMarkt (by FM) -20.948
> 24 Futbolmetrix -20.977
> 25 Betfair -21.042
> 26 Elo (by FM) -21.251
> 27 ClubElo/ClubTM (by FM) -21.319
> 28 Opta Analyst -21.325
> 29 Strongest link (by FM) -21.337
> 30 Alessandro Riolo -21.539
> 31 Goalkeepers (by FM) -21.548
> 32 Hybrid Machine Learning (by AZ) -21.645
> 33 Econ1 (by FM) -21.787
> 34 Q1_Bias (by FM) -21.908
> 35 KonakaLab -21.936
> 36 AIrgentina -21.996
> 37 FoG -22.181
> 38 Agnostic -22.181
> 39 Weakest link (by FM) -22.211
> 40 InPlayFootballTips -22.255
> 41 Anders T -22.344
> 42 NT-ELO predictions (by Ed) -22.516
> 43 dopemodels.info -22.560
> 44 Historic Match Model (by AZ) -22.786
> 45 Keanu J -22.875
> 46 Lléo -22.906
> 47 MH -22.965
> 48 Econ0 (by FM) -23.078
> 49 Footballgenius -23.299
> 50 FIFA (by FM) -23.371
> 51 Fantomen -23.403
> 52 Werner Pichler -23.602
> 53 Mark -23.983
> 54 Forwards (by FM) -24.257
> 55 Swiss Football Data -25.209
> 56 Scrabble (by FM) -27.891
> 57 Goal Experience (by FM) -30.514
> 58 Experience (by FM) -36.811
>
> Also available here: https://futbolmetrix.wordpress.com/2022/11/28/wc2022-sophcon-standings-after-match-day-2/

Senegal going through instead of Ecuador put Goalkeepers past me.

Tchau!
Jesus Petry

On Tuesday, November 29, 2022 at 1:02:40 PM UTC-5, jesus... wrote:
> Em segunda-feira, 28 de novembro de 2022 às 19:52:39 UTC-3, Futbolmetrix escreveu:
> > Also available here: https://futbolmetrix.wordpress.com/2022/11/28/wc2022-sophcon-standings-after-match-day-2/
> Senegal going through instead of Ecuador put Goalkeepers past me.

What's even worse is that milivella's cuca is in third place. We'll never hear the end of it from him...

On Tuesday, November 29, 2022 at 1:47:06 PM UTC-5, Futbolmetrix wrote:
> On Tuesday, November 29, 2022 at 1:02:40 PM UTC-5, jesus... wrote:
> > Em segunda-feira, 28 de novembro de 2022 às 19:52:39 UTC-3, Futbolmetrix escreveu:
> > > Also available here: https://futbolmetrix.wordpress.com/2022/11/28/wc2022-sophcon-standings-after-match-day-2/
> > Senegal going through instead of Ecuador put Goalkeepers past me.
> What's even worse is that milivella's cuca is in third place. We'll never hear the end of it from him...

After the end of Group B, my army of TransferMarkt based predictions is now crowding the top 10, but cuca in effective third place is just bad. Jesper has snuck into the top 10, Jesus has dropped out (MatchDay2 is an hors-de concours entry that assumes perfect knowledge of everything that happened in MDs 1 and 2, and then uses Elo)

1 MatchDay 2 -15.404
2 Goalkeepers (by FM) -17.042
3 Defenders (by FM) -17.568
4 cuca -17.828
5 Middle link (by FM) -18.007
6 TransferMarkt (by FM) -18.029
7 Jesper -18.143
8 Midfielders (by FM) -18.292
9 TransferMarkt - Adj (by FM) -18.400
10 Dan Levy -18.571
11 ClubElo/ClubTM (by FM) -18.663

Futbolmetrix:
> On Tuesday, November 29, 2022 at 1:02:40 PM UTC-5, jesus... wrote:
>> Em segunda-feira, 28 de novembro de 2022 às 19:52:39 UTC-3, Futbolmetrix escreveu:
>>> Also available here: https://futbolmetrix.wordpress.com/2022/11/28/wc2022-sophcon-standings-after-match-day-2/
>> Senegal going through instead of Ecuador put Goalkeepers past me.
>
> What's even worse is that milivella's cuca is in third place. We'll never hear the end of it from him...
Now I am not even allowed to enjoy my 15 minutes of glory?! Tomorrow
I'll be 35th again.
Also: what else I am supposed to do with my 200-page dissertation on why
CuCa is the best possible model?
--
Cheers
milivella