Colorado - Postmatch

[mgarbowski]

Look above the second "C" covering the lower level seats. There are 2 people alone in the front row on the aisle. That's my daughter and me. And the guys in my story I posted earlier who came out for the second half had their choice of every empty seat you see in this image and they picked ours.

 

[Schwallacus]

He’s playing the position like Lothar Mathaus played for Germany.

Schwallacus spider sense just got triggered with the mention of LM.

That's why I felt a little ringing sound in my ear yesterday morning (Matthäus*)
I like Ring there as a CB, between Callens and Chanot/Ibeagha. Want to see more of it, as we saw a pinch of it during the second half of the slaughter at the hands of RBNJ. How does he fare a whole game vs. a side like Atlanta, NJ, or Toronto.

 

[sammy3469]

With Ring able to play CB, a 442 diamond morphs into a 3412 very easily with him shifting back and the outside backs holding a higher line with free reign to push forward.

Berget today split the sdefenders so many times and he didn’t get the throughball service. Players on the ball need to look for his runs because hitting him like that gives a completely different variable to the team’s attack.

Having watched from behind the goal they were attacking in the first half, it's pretty apparent that he just knows how to make dangerous runs and more importantly how to make them at angles and depths that force defenders to drag with him opening up space for Villa (I don't think that movement is as readily apparent from the standard TV view). He's really good at dragging that CB just out of position enough to destabilize the whole setup. Granted Colorado was putrid, but moving him inside a tad allowed Villa to play off of him more effectively.

The other part of it is it also allows them to overload a side with an extra man more effectively. They pretty much continually had Berget making the run to open space, Villa moving into and then receiving the ball in that space and then initiating the attack with the width of Tajouri and Matarrita providing additional space (ie pulling a defender out wide) and outlet.

 

[Mario]

If you’re going to insult Herrera at least get your cultural references right. Herrera is from Venezuela, not Guatemala and pupusas are from El Salvador.

ok, he should go back to mexico then.

 

[Fantazma]

I think I am kind of over expected goals. It's a good idea in the abstract, but I think the application is very flawed and the results meaningless.

i was never into it to begin with.

 

[Keith Putnam]

i was never into it to begin with.

Did you have a better model in mind for assessing the quality of shots and building a framework for whether teams or players are under or over-performing?

 

[Fantazma]

Did you have a better model in mind for assessing the quality of shots and building a framework for whether teams or players are under or over-performing?

no...just that this is MLS...time and time again stats say X teams should do something and then complete opposite happens...the way you check is how teams play and see how teams play week in and week out and see how results are showing if they are really doing as good or bad as their record shows (eye test, i guess) . There are are stats that help in different areas but to me xG is not one of them to me. outside of penalties how do you really judge a quality of a shot? its so subjective, especially when there are cases where shots get deflected or bounce of defenders and lead to goals

im sure the xG said something about us having more chances and better game vs RB and we saw how that went. I still think we are in the start of the era of statistics so there is so much out there right now that we will eventually get to the point where we know which ones are useful which are not and even modify them as we go on.

based on that tweet it seems the game should of been closer vs colorado....we saw that we were never really in danger of losing that game at all and was not truly close at all despite the shot count being close

 

[SoupInNYC]

no...just that this is MLS...time and time again stats say X teams should do something and then complete opposite happens...the way you check is how teams play and see how teams play week in and week out and see how results are showing if they are really doing as good or bad as their record shows (eye test, i guess) . There are are stats that help in different areas but to me xG is not one of them to me. outside of penalties how do you really judge a quality of a shot? its so subjective, especially when there are cases where shots get deflected or bounce of defenders and lead to goals

im sure the xG said something about us having more chances and better game vs RB and we saw how that went. I still think we are in the start of the era of statistics so there is so much out there right now that we will eventually get to the point where we know which ones are useful which are not and even modify them as we go on.

based on that tweet it seems the game should of been closer vs colorado....we saw that we were never really in danger of losing that game at all and was not truly close at all despite the shot count being close

I don't think that xG should be completely dismissed.

It's a stat, in which just like many others, has its flaws. And as such, should be treated that way. There will be instances in which xG will agree with the story of a match and instances in which it doesn't line up. It just needs to be taken with a grain of salt.

 

[Keith Putnam]

how do you really judge a quality of a shot?

I mean, all due respect, and acknowledging that xG is (like all of them) an incomplete statistical model, but xG is *entirely about* judging the quality of a shot, and a lot of time and thought has been put into how much weight to give to all of the considered factors. Your question makes it sound like "magnets, how do they work?".

 

[Fantazma]

I mean, all due respect, and acknowledging that xG is (like all of them) an incomplete statistical model, but xG is *entirely about* judging the quality of a shot, and a lot of time and thought has been put into how much weight to give to all of the considered factors. Your question makes it sound like "magnets, how do they work?".

so how does xG judge the shot leading to second goal vs colorado? and the fourth goal where howard just completely moved out of the way ....are they both judged the same? it just feels too subjective to give it a set weight ( outside of penalties and perhaps set pieces) no matter how many similar goals you see.

 

[Fantazma]

I don't think that xG should be completely dismissed.

It's a stat, in which just like many others, has its flaws. And as such, should be treated that way. There will be instances in which xG will agree with the story of a match and instances in which it doesn't line up. It just needs to be taken with a grain of salt.

right .....maybe i just judge it too much....feels too detailed and unnecessarily complicated for something that can pretty much be subjective when someone can say yes this shot should have more weight than other. this is why i just disregard it, i feel like i have other stats that may tell me similar story already.

its like when i see possession stats...i need to see where this possession occurs on the field to see if it really made a difference. so many times we have the ball much more but all we are really doing is passing it between callens and johnson. and from johnson to chanot or something like that.

 

[Keith Putnam]

are they both judged the same?

No. Every shot has a large list of data points which are run through the vendor's particular model and return a percentage chance of a goal being scored relative to their data set of analyzed goals scored. Every shot is going to return a different percentage chance of scoring, either by a lot or a little.

The fact that we overperformed our xG by 2.46 goals just means that those 2.46 goals were statistically less likely to go in than the other 1.54 which did go in, so either we got lucky or had excellent finishing, or a bit of both. It doesn't mean they were impossible goals; they were simply lower-percentage chances which fell our way.

I don't pay for access to OPTA or Statsbomb, so I can't dig super deep, but there are tons of explanatory articles on the web about how xG is derived.

 

[Gotham Gator]

so either we got lucky or had excellent finishing, or a bit of both.

Well, that's the issue, right? Most people look at expected goals and figure the difference between it and actual goals is due to luck. Some will also realize that better finishing is an important factor - it does matter whether David Villa or Brek Shea is the guy taking the shot.

But there is also the noise in the system. All the things the different models don't, or can't, take into account, and all the ways in which the models don't deal correctly with the data they do have. The longer I've looked at expected goals, the more I've come to believe that noise has as much or more to do with variance against results as anything else.