Florida game

Only one game this weekend. Too bad - I was planning a road trip to Corvallis.

My assumption is that Garrett will play the strongest team he can. Earlier I said that this might be:

Davis; Tsao, Nicholson or McClusie, Olivier, Reed; Burk, Winters, Rapinoe, Enyeart: Schmidt, Foxhoven.

Do others agree - both with likelihood of strongest team, as opposed to not starting the three amigas, for example as last weekend; and with what the strongest line up is?

Sigh... gonna have to miss this one. Working that night. I'll be relying on all ya'll for the goods.

Martin Jol wrote:

My assumption is that Garrett will play the strongest team he can. Earlier I said that this might be:

Davis; Tsao, Nicholson or McClusie, Olivier, Reed; Burk, Winters, Rapinoe, Enyeart: Schmidt, Foxhoven.

Do others agree - both with likelihood of strongest team, as opposed to not starting the three amigas, for example as last weekend; and with what the strongest line up is?

Martin - I do agree that Garrett will play the strongest team he can. Florida should be a very difficult test for us. We have played four consecutive teams that were not quick enough and/or skilled enough to make a game of it.

Florida, as I understand from reading other posts, may not be a high scoring team, but their defense and goalkeeping are very strong. With a team like this, they may only need one goal to beat us, as was the case in the UCLA/USC games that we split a few weeks back.

They will be well aware of our explosiveness and will game plan accordingly. I am sure they are quick and skilled and will make us earn everything we can get from them.

I like your line-up. That's probably about it barring strange ideas Garrett might like to experiment with. But I think he'll be somewhat conservative unless we get to 3-0. Another issue is injuries. We have several players banged up. How that might affect the starting lineup we'll just have to wait until
Friday night to find out.

UF's defenders are not that solid, but their keepers are great. Watching our game against them was one of the most frustrating games I have watched in years. We had so many chances to score, 4 point blank. They have continued to get better with each game so look out, but you all will get plenty of chances to score. Abdullah, Elliot and Chapman all our very good for UF. On another note and I look forward to the next time we meet you guys on the pitch, always a great game. Good luck the rest of the year.

Welcome aboard, FSUfan. Thanks for the insights on UF from someone who's had experience with them lately. Fell free to join in the discussions if you have any more insights to East Coast teams. We don't get much insight into those teams on this coast.

I saw a Krikorian quote that he's looking to playing again at Merlo, and a rumor on our end that a game is in the works.. We'd love to have your team back.

The only thing I might add is Kelley might get in there due to possible lingering injuries to Tsao (sounded like she had a mild concussion) and Nicolson with the leg. Even so I don't think we will lose anything in the midfield and up top. Kelsey better bring her A game as it might just be a one goal difference and I expect her to see some SAs,

Albyn Jones just came out with updated prediction data. He gives UP a 2000 rating, and UF an 1836, or a difference of 164.

If you add in a 60 point home field advantage, that means a difference of 204. Ordinarily, if I understand AJ correctly, a 200 point difference means the higher rated team has a 95% chance of winning.

They still have to play the game...

Well lets keepour fingers crossed for AJ's system and make sure the outcome is never in doubt. Hear that Danielle/Michelle/Ellie? First goal within the first five please!

It appears I don't quite understand AJ. I was assuming 200 pts. was 2 st. Dev's. on a bell curve, but apparently there's a bit more to it. (the 2/3 ratio for 100 points was the clue)

Using the data in the predictor (12/05/2007) UP has a 82% chance of winning, Fla has about a 10% chance.b ut that was from last year, when the spread was about 70 points higher.

If I pick a team with the same spread as this years UP-FLA spread, I get ratios more like 73%:15% (the rest tie)

Any way you figure it, we should win... but...

They STILL have to play.

Geezer - a telling stat to me would be how often he is correct, and thusly, how often he is not. A simple accuracy percentage would help me a lot.

(You may have mentioned this last season, but ain't no way in blazes I'm remembering a detail like that from so long ago. Thanks!)

what Geezer said

I think I noted that the Albyn Jones Java predictor has predicted the winner of every UP tournament game the last two years. I haven't checked before that, and I don't know if the predictor works in reverse, but it would be interesting to know how it did before that.


For the record, AJ also predicted that UCLA would beat UP this year, and that UP would beat USC. In both cases, however, the prediction was inside the error range.

Albyn also gave a strong prediction we would beat the other teams, and we did.Go Pilots! ...

For those of you who want to see Jones' latest ratings, covering games through September 14, use the following link and click on Women, Division I: http://soccerratings.com/index.php?title=Main_Page. His "prediction" device also is on that page. If I understand his "prediction" device correctly, it takes into account the expected statistical accuracy of his system and, based on that, tells you the likelihood of each team winning and also of a tie.

UPSoccerFanatic wrote:For those of you who want to see Jones' latest ratings, covering games through September 14, use the following link and click on Women, Division I: http://soccerratings.com/index.php?title=Main_Page. His "prediction" device also is on that page. If I understand his "prediction" device correctly, it takes into account the expected statistical accuracy of his system and, based on that, tells you the likelihood of each team winning and also of a tie.

One caveat. The data in the Prediction device is NOT CURRENT.

According to the flash notice that comes up when the page loads, the data was entered 12/05/07, which means it is data from 11/28/07.

The tables will give you the current data, but don't tell you how to compute odds with them (other than to say that 100 points is equivalent to approximately a .67 chance of winning)

But by using both the table and the predictor, you can get a good feel for the odds.

Just to confuse things a little more about rating systems, including Jones' system:

Jones' system assigns a team a rating. In addition, it discloses for each team a Standard Error. What you can do is add the standard error to the team's rating and also subtract the standard error from the team's rating, to get a range. Using Portland's current Jones rating of 2000 (having that nice round number is just a fluke), the SE for Portland is 123. So, the range for Portland's rating using one SE is 2123-1877. What this means is that from a statistical perspective you can expect that 68% of the time, game outcomes will be consistent with Portland having an actual rating within that range. If you were to use two SEs, increasing the amounts on either side of 2000, your expectation percentage would go up to 90%. Jones very nicely explained this to me last year.

Now, turning to Florida, Jones' system gives it a rating of 1836, with a SE of 106. So, you can expect that 68% of the time, Florida's game outcomes will be consistent with Florida having an actual rating in the range of 1942 to 1730.

You have noted, I'm sure, that there is an overlap between Portland's range and Florida's. I assume this means that the rating system is inaccurate enough that Florida actually might be better than Portland. Probably isn't, but might be.

I don't know exactly how Jones' predictive calculator works, but my guess is that it says that taking into account the two teams' ratings and their SEs, in a game on Portland's soil, the statistical likelihood of Portland winning is X, of Florida winning is Y, and of a tie is Z. It can do similar calculations for a game on Florida's soil and for a game on neutral soil. The likelihoods are different based on where games are played because of the known value of home field advantage, which is part of what goes into the calculator's calculation. I believe, but do not know, that the calculator's assigned value to home field is the same for all teams -- meaning that Portland's home field advantage is treated as the same as any other team's home field advantage, which I'm sure we all think is not actually the case.

As the season progresses and more data from this year enter into the system, and reliance on past experience is extruded from the system, the SEs for teams will decline. What this means is that the calculator, in stating the likelihoods of possible results, takes into account the reliability of the data at the time it is doing a calculation. That being the case, the calculator should be as accurate now as it will be later in the season.

There will be a test on this at some time in the future, yet to be decided.

Nice job.

to further add to the confusion, not only do we think we have a better home field than most schools, I ran the data since Merlo opened, and out Neutral field winning percentage is nearly identical to our Merlo percentage. I didn't run SOS data, though.

Maybe we play because it's easier than doing the calcs....