FIFA rankings - why they are important and how to beat the system

[Kadenge]

Crazy that our tie vs Mexico was considered to be a game played on neutral ground, when 98% of the fans were pro Mexico. It's an advantage for Mexico to play all their games in the USA,  Win/Loss.

[Olympique_de_Marseille]

3 minutes ago, maplebanana said:

That is certainly part of it. I tried to estimate NZ based on their FIFA ranking (95) being between Curacao (88) and T&T (101). But T&T's CONCACAF ranking is quite a bit higher than Curacao. But playing with the numbers in that range won't get me from -37 to -31 (more like -35).

What if, just like for FIFA rankings, the weighting is higher for friendlies played within the FIFA window than those played outside of it? Would that help explain part of the swing?

[maplebanana]

3 minutes ago, Olympique_de_Marseille said:

What if, just like for FIFA rankings, the weighting is higher for friendlies played within the FIFA window than those played outside of it? Would that help explain part of the swing?

Yes, that would help. But the USA vs PAN game was inside the window, no?

I can get the model to make sense but only if they assigned a value of 1700 for NZ, which would put them just ahead of CRC. Which seems awfully high.

[Olympique_de_Marseille]

39 minutes ago, maplebanana said:

That is certainly part of it.

If you've slightly underestimated friendly matches within the FIFA window, it still seems like you are overestimating the loss of points for a draw against New Zealand. The USA losing ~17 points for their draw against the Kiwis is a wild swing.

Like, what if what is at play is the not so much New Zealand's "CONCACAF" ranking but actually the coefficient used in your formula for a draw?

Sorry, just searching wildly for a reason.

Thanks for your work on this file

Edited Monday at 05:44 PM by Olympique_de_Marseille

[maplebanana]

1 hour ago, Olympique_de_Marseille said:

If you've slightly underestimated friendly matches within the FIFA window, it still seems like you are overestimating the loss of points for a draw against New Zealand. The USA losing ~17 points for their draw against the Kiwis is a wild swing.

Like, what if what is at play is the not so much New Zealand's "CONCACAF" ranking but actually the coefficient used in your formula for a draw?

Sorry, just searching wildly for a reason.

Thanks for your work on this file

I realized I actually forgot to include the home field advantage in the CAN vs USA game, which leads to CAN getting +23.7 instead of +19 (I fixed that in my original post). That means my window calculation is now +26.5 (vs actual +25). 

Based on my model the USA would have lost 23.7 points for losing to us and lost 13.6 for drawing NZ. That does seem like a lot but it was also a pretty big upset, so maybe?

If I adjust the coefficients (from 175 to 100 for home field adv) then I get exactly Canada +25 and USA -31 in the window (assume NZ is 1500, or Haiti-like). But then it's no longer consistent with the data set that I used to develop this model in the first place (March 2023).

I think bottom line: Beat Panama, and cheer for the USA to lose (even against PAN) and we end up 2nd

 

[archer21]

FYI don’t listen to me on this stuff, I basically just eyeballed based on the ES and GUA match. I don’t have a model, I was just guessing because I didn’t know someone had actually done the work to build a model. Not pretending to be an expert here lol

[archer21]

3 hours ago, maplebanana said:

I realized I actually forgot to include the home field advantage in the CAN vs USA game, which leads to CAN getting +23.7 instead of +19 (I fixed that in my original post). That means my window calculation is now +26.5 (vs actual +25). 

Based on my model the USA would have lost 23.7 points for losing to us and lost 13.6 for drawing NZ. That does seem like a lot but it was also a pretty big upset, so maybe?

If I adjust the coefficients (from 175 to 100 for home field adv) then I get exactly Canada +25 and USA -31 in the window (assume NZ is 1500, or Haiti-like). But then it's no longer consistent with the data set that I used to develop this model in the first place (March 2023).

I think bottom line: Beat Panama, and cheer for the USA to lose (even against PAN) and we end up 2nd

 

Any prediction on what the total points would be assuming Canada and USA both win all their October matches? Would be nice if we “controlled our own destiny”.

[maplebanana]

1 hour ago, archer21 said:

Any prediction on what the total points would be assuming Canada and USA both win all their October matches? Would be nice if we “controlled our own destiny”.

Based on my model using my original coefficients, a win @ MEX nets them +25 pts. Winning both is +37 which would put them at 1811, which just clips us at 1806 if we win. A mexico/usa draw though would put them at 1797 though. 

[Unnamed Trialist]

What @maplebanana and others are helping us conclude, thanks, is that we have to go all out for the win vs. Panama playing our very best side. 

So the whole business of a call-up with new caps and dual nationals, all very well to be part of the group and train in Montreal. But in Toronto we play with our very best, our starters, and we go for the jugular as our future tournament play in 2025 depends on the best ranking possible.