@Samir Khan I didn't think anyone would come up with something so fast! 

**added... ok I see now, pretty simple the theoretical low is just on the psychrometric chart, now just have to test the theory if or when I have time.

That 5 degree C minimum I mentioned at 0%humidity could only occur if the outside air temperature was at most about 18 degrees C. 

... using maple, if the lowest possible temperature someone measured was 5 degrees celcius (wet bulb temp) using a zeer we can calculate the temperature it might have been at 0% humidity at a standard pressure of 101.325 kPa

with(ThermophysicalData):
Property(Tdb, HumidAir, R = 0.00001, P = 101325, Twb = 278.15)   #R=0 (humidity @ 0%)produces strange result so I have chosen a value close to zero instead.

                                               291.567042014157437

convert(%,temperature,kelvin,Celsius)

                                               18.4170420

Currently where I am P=102.2 kPa, T=26.5 degrees Celsius with humidity at 36%.  The theoretical lowest temperature I could achieve right now with a zeer is ..

Property(Twb,HumidAir,R=.36,P=102200,Tdb=26.5+273.15)
 
                                        289.800660462402220

convert(%, temperature, kelvin, Celsius)

                                        16.6506605

Thanks for the great application Samir!

On the topic of thermal applications, I would like to bring to light a primitive refrigerator called the Zeer - one smaller clay pot inside a larger clay pot filled in between with damp sand.  It would be interesting to see this "Zeer" explored in Maple.  Based on the principles of evaporation it is said to get as cool as 5 degrees celcius at 0% humidity. 

It would be interesting to plot the theoretical inside temperature vs. outside tempurature at various humidity levels. 

I don't know if anyone would find interest in this topic but I would certainly enjoy seeing the results that anyone could come up with.

I've attached an updated round of 16 knockout simulation with up to date Elo ratings of the teams in the round.  For anyones interest.

Eurosimulation2016-round_of_16.mw

@acer This works great!  I wonder if the Laplace interpolation produces a better end result? 

I'm actually quite amazed at how well both methods work.  Thanks for looking into this Acer.

@vv Great work!  We can apply this to the world cup in two years.  Although I'm not a strong coder I can do some of the grunt work and add more countries flags later.

Fine tuning the scoring : 
I was looking into how a teams scores were related to their elo ratings.  I found initially (haven't been able to look further lately) although it is more likely for a team with a higher elo rating to win, it is not unlikely a team with a lower elo rating to have a high score like 3-1 or even 4-0.  We might be able to come up with some distribution probability for the scoring.  ie teams with elo ratings within 100 points of each other are fairly evenly matched, teams separated by more than 300 points are more likely to score 1 more goal than their opponent .. or something like that. 

.. and keywords like bodybuilding, viagra, and therapy.  I have been deleting many as well in the last while. 

That's interesting, keep that generator.  I was meaning the first round of 16 who is knocked out.  Is your knockout from the group stage?

@vv Yes it seems to work correctly with your qualification algorithm. 

An interesting stat to include in the first round results so far list would be how many times who is knocked out in the first round. 

@vv Very nice start!  A couple of fixes we'll have to take care of.  I had 2 Croatias and 2 Italy's in the knockout round.  Another time 2 Swedens and 3 Croatia's.  Then we can look at the qualification a little more closely

Standings for 3rd place

Additionally, below the first table is what I produced above and the second table determines third place matches.

Ideally it should.  The score also plays a part in the rating as well, meaning a higher goal differential will result in more elo points.  We also have to predict the scores in able to rank the 3rd place teams in case of ties to get through and who they play.  I think this is part of the difficulty of the group stage.

Yes the first 2 get through the group stage (each team plays the other teams in their group once - 3 games).  The placing for 1st round knockout goes like this

Group A runner up vs. Group C runner up

Group D winner vs. Group B/E/F 3rd place

Group B winner vs. Group A/C/D 3rd place

Group F winner vs. Group E runner up

Group C winner vs. Group A/B/F 3rd place

Group E winner vs. Group D runner up

Group A winner vs. Group C/D/E 3rd place

Group B runner up vs. Group F runner up

Yes.  So the expected outcome of team x is the formula you show. 

And the new rating adjustment is calculated as

Elo_new=Elo_old + K * (actual outcome - expected outcome)  where expected outcome is the calculation you show, (actual outcome - win =1 draw =0.5 and loss = 0) and K is a weight constant factor 60 for world cup finals, 50 for continental championships and major intercontinental tournaments, 40 for World cup and continental qualifiers, 30 for all other tournaments and 20 for friendly matches. 

In our case I think it would be 50.

It's been a while since I looked at that and I created an excel sheet for it but can't locate it.  Wikipedia is a good place to see how it's calculated https://en.wikipedia.org/wiki/Elo_rating_system

Basically if a weaker team beats a stronger team, points gained by the weaker team will be larger than if the weaker team lost to the stronger team. 

Everything is updated, I've updated the file above and everything works with 24 teams in the tournament at the 16 knockout stage round (the 16 teams I chose were the most probable based on ELO ratings only).  Now to work the group stages into the simulation which should take far too long for me to complete before the Eurocup group stage is over.

I invite anyone to figure the group stage out.