Newcastle and Wellington primed for the drop. If Blacktown manage a win that'd really help. 

Top work as ever!

CCM down to 17th, incredible.

The scenes if the Greens knock out another A-League team next round. They'd be about top 5 surely.

Whoa. Poor CCM, haha. But I believe it. Just need Gong to win on Wednesday.

Can't live without these rankings now! :)

The FFA Cup upsets through the week got me thinking about the biggest upsets that we've seen and where these rank on the list.

So I've gone back through all games in my database since the start of 2014 (7134 games) and listed the top 10 below.

There are two lists.  The first is probably the more accurate list in the sense that it's based purely on how many ELO points are transferred between the teams (i.e. accounts for difference in ELO points at the time of the game, home advantage and the score).  The second table takes the score out of the equation and looks purely at the result (home advantage still taken into account).

Interestingly, there were two results in the last week that made the list which I've highlighted.

As with all of these stats, take them with a grain of salt!

* Win Probability shows the probability of the underdog winning according to the difference in ELO points before the game (allowing for home advantage)

Green Gully beating Central Coast sits at #1976 on the first table and #720 on the second.
Adelaide City beating the Wanderers is #1061 on the first table and #101 on the second.

Interesting numbers. From that it'd appear the FFA Cup qualification spaces are completely out of whack.

Numbers of NPL teams in top 50 compared to qualification spots:

ACT 1 for 1 spot
NSW 7 for 5 spots
NNSW 2 for 2 spots
NT 1 for 1 spot
QLD 7 for 4 spots
SA 9 for 1 spot
TAS 2 for 1 spot
VIC 8 for 4 spots
WA 3 for 2 spots

Compared to top 25:
ACT 0 for 1
NSW 4 for 5
NNSW 1 for 2
NT 0 for 1
QLD 1 for 4
SA 4 for 1
TAS 0 for 1
VIC 5 for 4
WA 0 for 2

South Australia is getting seriously ripped off. I suspect the limited game time against other federations is artificially inflating their numbers, but still they seem like the federation getting the rawest deal whilst NSW, Queensland and WA seem comparatively over-serviced.

Interesting stats, thanks.

FIFA once sought to have the elite level Leagues (EPL, la Liga,Serie A Bundesliaga Lig une) limited to 18 teams. If we were to have a two tier league of 18 teams in each division the present rankings would give all confederations except ACT at least 1 more team and includes the Northern Territory.

I guess that shows the breadth of talent throughout the country and why there is the pressure to increase the "elite" level teams in the A League.

Just aiming at bringing in the requisite number into the HAL, we would have additional 4 Victorian, 2 NSW, 1 SA and 1 QLD.

I think the best determiner for extra Cup Spots would be a UEFA style coefficient rankings. All states and territories should still be guaranteed at least one spot but anything extra should be earned on merit. As far as I can tell, UEFA's system is something like Total points accumulated divided by number of participating teams and then '
averaged over a 5 year period. It means teams from states with seemingly inflated number of spots like Northern NSW would have to consistently ensure they are getting the results to keep their allocation. 

Awesome. Can you show top 10 or so for just FFA Cup upsets?

Yeah, although if you want to see only the Round of 32 onwards it'll take a bit longer.  I've included all qualifying games as 'FFA Cup' for now.

All good, no worries if it takes too much effort, you already do enough!

Here are the FFA Cup upsets only (from Round of 32 onwards), ignoring the score.  If I was to include the table that factors in the scores, it looks a bit weird because of the small number of games.  For example, Edgeworth were favourites against FNQ Heat a couple of weeks ago, but because they won easily (3-0) they make the top 10 list.

There have only been 69 games played in the FFA Cup that were included in my database.  Exclusions were because one of the teams is not in a league that I cover (e.g. South Springvale, Surfers Paradise).

Awesome stuff, it's so funny how far down the list the Green Gully win was!

I like how Green Gully beating CCM wasn't considered that big a Cupset.

Game Day 4 probabilities as per the club rankings.

Sydney is expected to beat Wolves easily.  An upset in that game will have nearly as big an impact as Redlands beating Adelaide did.

Updated to 14th August.

Biggest upsets of the week:
Lions FC (38) 2 - 3 UQ FC (187) - This is the biggest upset in my database, even bigger than Redlands beating Adelaide United last week. UQ were given a 2.9% chance of winning.
Cooma FC (58) 0 - 1 Tuggeranong United (179)
Northcote City (72) 2 - 1 Bentleigh Greens (8)
West Torrens Birkalla (27) 1 - 2 Adelaide Raiders (113)

Updated to 21st August:

Biggest upsets of the week.  All of these cracked the top 100 upsets out of over 7000 games since the start of 2014.
- North Pine 2 - 3 Lions FC [QLD BPL]
- FNQ Heat 0 - 2 SWQ Thunder [QLD NPL]
- Ballarat Red Devils 1 - 2 Bendigo City [VIC NPL1 West]
- Hume City 0 - 2 Bullen Lions [VIC NPL]
- UQ FC 1 - 0 Capalaba [QLD BPL]

Odds for this week's FFA Cup games according to the rankings:

I wonder if Vick's win over Hume is enough to get them back to #1. We shall find out next week!

I am no ELO expert but my guess is Victory would be likely to lose points given the 1-0 margin when the model above predicted would be a 39% chance of Hume winning, a draw or a 1-0 margin. ie 2-0 to Victory would be more or less a par score with a slight increase to Victory's ranking points, 3-0 or higher would see Victory's ranking points increased more.

As you said though, we will see.

That is some confusing stuff. I've tried to read the Elo formula many times but just can't get my head around it.

no...team don't lose points on any win....

This is right.  Very simply, the ELO points are used to calculate the probability of each team winning.  This is then compared to the actual result (1 for a win, 0 for a loss, 0.5 for a draw) and points are transfered between the two teams based on whether the teams under or over-performed.

So for Hume, the probability of winning was around 8%, but they actually lost.  So going by the ELO calculations they under-performed (0.08 prediction compared to 0 actual).  But because they had only a small chance of winning, the points transfered was only small.

Updated to 28th August 2016.

Central Coast Mariners have crept back up to 15th with the two Melbourne A-League clubs winning their FFA Cup games.

Some competitions are into the Finals now, so there will be less games each week.  It also means that for clubs like Bentleigh Greens, they could finish their season in the top 10, and it'll be up to the A-League clubs to catch them during their season.

Not enough games to apply this method. It's a pretty simple calculation of points per wins and draws divided by the number of participating clubs per country although they get bonus points for reaching certain rounds which is fine over there as they have group stages, etc. We don't have draws so it's simply 2 points per win, totalled up and divided by clubs per federation. You could apply different points for when they play an A-League club. Below is what it looks like here. Considered adding in the NPL finals but it skews the data too much. 

 

Fed.	2014	2015	2016	TOTAL	2017 Allocation
VIC	2.5	3.5	3.0	9.0	4
ACT	2.0	0.0	4.0	6.0	4
SA	4.0	1.0	0.0	5.0	3
QLD	2.0	1.0	1.0	4.0	3
NSW	0.6	1.2	1.0	2.8	2
TAS	0.0	0.0	2.0	2.0	2
NNSW	0.0	0.0	1.0	1.0	1
WA	0.0	0.0	0.0	0.0	1
NT	0.0	0.0	0.0	0.0	1
			Available Cup Spaces:		21

Going on the average alone after only 3 years would be a crazy but i'm sure it would balance out over time.

Interesting that so far the data shows that, on raw results, NNSW and WA are definitely undeserving of their 2nd spots.

Time for another update.  This table includes all games up to and including the 18th September, which is all Finals games and the Quarter Finals of the NPL Championship.  The only games still to play with non-A-League clubs are FFA Cup and NPL Championship games.

Bentleigh Greens have fallen to 12th after being knocked out of the NPLV finals by Oakleigh Cannons and also losing 3-1 to Edgeworth FC in the NPL Championship.  Blacktown City have moved above Newcastle Jets into 9th and South Melbourne into 11th.  Central Coast Mariners have caught up a little ground but are still in 14th.

I am tipping that Blacktown fall a bit based on their 3-0 FFA cup loss

I also think that CCM could be down as low as 16th given Edgeworth and Sydney United 58 both won their NPL Playoff Semi Finals

You're not far off at all.