2014 FIFA World Cup |OT5| Que Jara, Jara. Whatever Will Be, Will Be.

Status
Not open for further replies.
Man, Netherlands vs Mexico is going to be great tomorrow. I wish Layun was playing better though. Chapo Montes would've been great for this match :( Mexico's Defense is going to get worked tomorrow. Hopefully we pull through!!!

Hopefully prediction:
Mexico wins 2-1.
 
cc7.gif


leaked fifa 15 physics engine
 
videogamer said:
Um... 20% is a fairly large number. An adult Coloradan has about the same chance of being obese. Top schools have acceptance rates below 10%.
Click to expand...

But to have that happen numerous times in different groups? Not to mention what's the point when the forecasts are so reactive. It almost states the obvious.
 
videogamer said:
Um... 20% is a fairly large number. An adult Coloradan has about the same chance of being obese. Top schools have acceptance rates below 10%.
Click to expand...

I don't even know what you are trying to argue, of course there are probabilities involved when a team faces another, if the result was set in stone then why even play the match, what you are saying is that teams win or lose according to the probabilities, and even if it was a minor probability of something to happen and it happens your response was "of course, it had a mi or probability" which is, well obvious.

I'm going with Mexico winning tomorrow, thank you
 
Bam Bam Baklava said:
But to have that happen numerous times in different groups? Not to mention what's the point when the forecasts are so reactive. It almost states the obvious.
Click to expand...
It's unfortunate that the forecasting isn't good enough to entirely predict outcomes with absolute certainty.
f0lken said:
I don't even know what you are trying to argue, of course there are probabilities involved when a team faces another, if the result was set in stone then why even play the match, what you are saying is that teams win or lose according to the probabilities, and even if it was a minor probability of something to happen and it happens your response was "of course, it had a mi or probability" which is, well obvious.

I'm going with Mexico winning tomorrow, thank you
Click to expand...
ummm... not trying to argue as much as have a reasoned discussion. Probabilities are what they are, and I think it should be ok to bring up organizations' forecasts. No one needs to "believe" the forecasts.
 
f0lken said:
I don't even know what you are trying to argue, of course there are probabilities involved when a team faces another, if the result was set in stone then why even play the match, what you are saying is that teams win or lose according to the probabilities, and even if it was a minor probability of something to happen and it happens your response was "of course, it had a mi or probability" which is, well obvious.

I'm going with Mexico winning tomorrow, thank you
Click to expand...

Lol! Great response.
 
videogamer said:
ummm... not trying to argue as much as have a reasoned discussion. Probabilities are what they are, and I think it should be ok to bring up organizations' forecasts. No one needs to "believe" the forecasts.
Click to expand...

Probability and forecasts are actually different things... Probability only make sense when you're dealing with big numbers and random sampling. Team A having a 70% probability of beating Team B means that if they meet 1000 times, you should expect around 700 Team A victories. It also means that if you always predict team A to win in the next 1000 matches, you'll be wrong 3 times out of 10.
But it doesn't tell much for the next match, the only thing that we can take from it is that it is possible for any of the two teams to win.
 
Alx said:
Probability and forecasts are actually different things... Probability only make sense when you're dealing with big numbers and random sampling. Team A having a 70% probability of beating Team B means that if they meet 1000 times, you should expect around 700 Team A victories. It also means that if you always predict team A to win in the next 1000 matches, you'll be wrong 3 times out of 10.
But it doesn't tell much for the next match, the only thing that we can take from it is that it is possible for any of the two teams to win.
Click to expand...

Thanks for the clarification. I think I should have stuck with the word probability instead of using both words.
 
videogamer said:
Do you have a STEM degree?
Click to expand...
I don't think the model 538 is using takes into account enough factors into its post-match updates to accurately predict match results better than a random choice would (if you can come up with the exact hit-miss ratio for the model so far and prove its significantly better than what a random predictor would come up with in matches where the probability difference between teams were sufficiently similar then great). IMO a model that started out as blank or was evenly distributed at the start of the tournament and updated as the games went on, or at least one which attributed a good part of each team's "weight" from that would be way more effective than the model 538 uses.

You could say then that people who think Costa Rica has a better chance of winning are using a much more short-sighted model to predict the result of tomorrow's game, but IMO that might actually be the better model.

Even Goldman Sachs predictive model and analysis document states their own model is only barely better than a random choice (having applied their methodology to the 2010 WC), and that it becomes increasingly imaccurate in each round, because of the obviously very random nature of soccer.
 
Fireblend said:
I don't think the model 538 is using takes into account enough factors into its post-match updates to accurately predict match results better than a random choice would (if you can come up with the exact hit-miss ratio for the model so far and prove its significantly better than what a random predictor would come up with in matches where the probability difference between teams were sufficiently similar then great). IMO a model that started out as blank or was evenly distributed at the start of the tournament and updated as the games went on, or at least one which attributed a good part of each team's "weight" from that would be way more effective than the model 538 uses.

You could say then that people who think Costa Rica has a better chance of winning are using a much more short-sighted model to predict the result of tomorrow's game, but IMO that might actually be the better model.

Even Goldman Sachs predictive model and analysis document states their own model is only barely better than a random choice (having applied their methodology to the 2010 WC), and that it becomes increasingly imaccurate in each round, because of the obviously very random nature of soccer.
Click to expand...
Interesting criticisms of their model. Hopefully as more games are played, accuracy will improve in the future.

However, if soccer is very random---or approaches totally random results, then why would anyone bet on soccer or even say that "x team will win"? Would skill amount to anything if the results are random?
 
meanspartan said:
I am no stats expert, but if the game ends and you have at least one more goal than the other team, there is a 100% chance you win.
Click to expand...

Technically not 100%, you can beat the other team and still have your victory taken away for fielding an ineligible player, for example.
 
videogamer said:
Interesting criticisms of their model. Hopefully as more games are played, accuracy will improve in the future.

However, if soccer is very random---or approaches totally random results, then why would anyone bet on soccer or even say that "x team will win"? Would skill amount to anything if the results are random?
Click to expand...

First, I'm not saying modeling is useless. You can infer possible results from trends. Of course any good model will take a game like Germany-Laos and assign a high chance of winning to Germany, but for games that are closer (even something like 60-40), as I said, a random predictor would probably be as good as the model, and the more recent data you have, the more reliable it is in this case.

My point isn't that models don't work, it's that you have to establish one as useful by showing it can predict significantly better for at least a subset of the events you want it to work on and understand its limitations before waving around its results. People's predictions, even those guided by mostly hope are still results of mental models, which also work far from how a random predictor would, so they're not inherently worse than a formal model or should be ignored as useless. Far from it, making a predictive model analyze samples as a brain does is a major part of machine learning and data intelligence.
 
Fireblend said:
First, I'm not saying modeling is useless. You can infer possible results from trends. Of course any good model will take a game like Germany-Laos and assign a high chance of winning to Germany, but for games that are closer (even something like 60-40), as I said, a random predictor would probably be as good as the model, and the more recent data you have, the more reliable it is.

My point isn't that models don't work, it's that you have to establish it as useful by showing it can predict significantly better for at least a subset of the events you want it to work on and understand its limitations before waving around its results. People's predictions, even those guided by mostly hope are still results of mental models, which also work far from how a random predictor would, so they're not inherently worse than a formal model or should be ignored as useless.
Click to expand...
Said the random dude on the internet!!!!
 
FeD.nL said:
Fucking nerves. Got a test tomorrow but can't get myself to study.
Click to expand...

I'm taking the California Bar at the end of July and this world cup is fucking my shit up, can't focus at all. Constantly either watching matches or reading articles/comments. Even if Greece gets eliminated tomorrow, I still have team USA. If they are out too, then I guess it will be easier to get back on track lol. I'll still follow, but I'll be far less invested.
 
Fireblend said:
First, I'm not saying modeling is useless. You can infer possible results from trends. Of course any good model will take a game like Germany-Laos and assign a high chance of winning to Germany, but for games that are closer (even something like 60-40), as I said, a random predictor would probably be as good as the model, and the more recent data you have, the more reliable it is in this case.

My point isn't that models don't work, it's that you have to establish one as useful by showing it can predict significantly better for at least a subset of the events you want it to work on and understand its limitations before waving around its results. People's predictions, even those guided by mostly hope are still results of mental models, which also work far from how a random predictor would, so they're not inherently worse than a formal model or should be ignored as useless. Far from it, making a predictive model analyze samples as a brain does is a major part of machine learning and data intelligence.
Click to expand...
Fret not, I wasn't saying that you are saying that "modeling is useless". I was just questioning the point of prediction/skill in a game if the results approximated random. I question if an individual's predictions are that helpful, especially if humans are prone to decision making biases.
 
Status
Not open for further replies.

Similar threads

  • Locked
Trump will be dictator like Putin
30
2K
Korranator replied
New Jurassic World movie will be Jurassic World: Rebirth. Official details released.
2 3
122
7K
jason10mm replied
DC Studios: Pattinson will not be the DCU Batman
2
54
3K
Doom85 replied
Christopher Nolan’s ‘The Odyssey’ will be the first blockbuster shot ENTIRELY on IMAX cameras
25
505
jason10mm replied
NHL Stanley Cup playoffs
2 3 4
162
2K
Trogdor1123 replied
Top Bottom