League Table 2015/16

woosh

Go home Leggy, youre drunk.

Central Coast Mariners 0 Newcastle Jets 1

		P	W	D	L	GD	Pts
	Western Sydney	21	11	5	5	8	38
	Adelaide United	21	10	6	5	8	36
	Brisbane Roar	21	10	5	6	3	35
	Melbourne City	21	9	5	7	13	32
	Melbourne Victory	21	8	6	7	5	30
	Sydney FC	21	7	9	5	4	30
	Perth Glory	21	8	4	9	2	28
	Newcastle Jets	21	7	5	9	-10	26
	Wellington Phoenix	21	6	4	11	-9	22
	Central Coast	21	3	3	15	-24	12

With 7 games still to go am I delusional to think we still may have a chance to slip into the 6?

22 + 21 = 43

It's hard to believe we gave up so many points to the Jets and the Mariners. Over half the Mariners points come from results against one team!

There's only 6 more games isn't there?

Probably.  Not only we would need to win our remaining games, but also not one, but two (or possibly even three) teams above us, would have to have suddenly start consistently losing.their remaining games.

If we were to win all our games I would say we would make it. 

We just need some salary cap violations from a few of the top 6 and we'd be right back in the running

Well, it won't hurt if we win every game from here on out..

Ernie was talking as if Krishna's recovery is going quicker than expected and we might get him back, maybe in a couple of weeks?

The odds of us winning all the remaining games are very remote. If we were to get 50% I would be surprised.

I still have faith we can make the top 6

Let's see. We're playing: Adelaide (h), Sydney (a), Perth (h), Melb. City (a), Melb. Victory (a), WSW (h).

That is a maximum of 18 points. If we win all of them, we will have 40 points (we have 22 now).

We would need to win two more games than the Jets, and three more games than Perth, Victory or Sydney FC, to get to 6th place.

So the Jets need to lose four games from now on, and 1) Perth, and 2) either Victory or Syd FC need to lose three games from now on. While it is possible mathematically, the chances are so low that I would rather sit back and enjoy our remaining games with "nothing to lose"  rather than expecting miracles.

At least we're playing top teams wouldn't have a chance if there were any at our end of the table.

In the unlikely event that we win all our games, the above is more than possible, I'd say it's probable. 

40 points has been more than enough to make the A-League finals in every single season so far. 

erm... some of this maths is very wrong too

If we won 6, and the jets won 4, drew 1, and lost 1, we would overtake them. That's very different than them needing to lose four times!

Some people seem to forget that all the teams will be playing each other - someone has to drop points, not everyone can take 3 every round!

Obviously we're not going to go 6 from 6, but it will be interesting to re-jig the table at the end as if we had, and see if we would've made it. I think you'll be surprised. 

I'm writing a script that will randomly generate results for all the other matches and see where we would be on the table if we win 3, 4, 5 games etc. Repeat that a million times and total the results and it should give some sort of indication as to what our chances of making the playoffs are given how many games we win.

I'm going to hope we make it until we cant.  It will make it hurt more if we don't but that's the risk I'll have to take.

Same...

It's not a conscious decision to hope that we make it, rather than watch games with a "nothing to lose" attitude - my brain don't work that way!

There's always something to lose. I'm always over-invested emotionally. It can't be stopped.

Correct, my mistake - Jets do not need to lose all their games (but I would be happy if they did).

However, as per your maths, if we win 6 and they win 4, we would win two more games than Jets, which is what I said above.

We would need to overtake Perth and one of Victory or Sydney FC, which is a tad harder.

So, I wrote a little script to simulate the rest of the season to work out the chance that the Nix will make the playoffs if they reach a certain number of points. I used the distribution of goals scored by the home and away team from 2000-2013 in the top 4 Euro leagues to generate a score for each team in each match, and worked out what it would do to the table. Reason I used that data is that it was easily accessible on the internet.

You can have a look at my code here: https://gist.github.com/42ae44a0e9d119e25190

Here's the results after I ran the simulation 1,000,000 times:

22 points: 0.00%

23 points: 0.00%

24 points: 0.00%

25 points: 0.00%

26 points: 0.00%

27 points: 0.00%

28 points: 0.00%

29 points: 0.00%

30 points: 0.08%

31 points: 0.95%

32 points: 4.57%

33 points: 15.89%

34 points: 36.16%

35 points: 63.26%

36 points: 87.53%

37 points: 96.16%

38 points: 99.78%

39 points: 0.00%

40 points: 100.00%

Note: the reason why 39 points is at 0% is that it's impossible for the Nix to finish on that number of points. 5 wins 1 draw = 38 points, 6 wins = 40 points.

Interesting... So if we get to 35 points we are more likely than not to get in? 13 points from 6 games then. 4 wins and a draw...

Pretty much. 5 wins basically confirms a playoff spot.

What I was actually trying to say, was my original statement: If we were to win all our games, I think we will almost certainly make it. 

The Jets are hardly the hurdle, they will never win 5 games from their remaining draw. And I think focussing on Perth, Victory and Sydney is misguided - if all three of them win most of their games, then Man City, or even the team above them, will drop to within our grasp. Your mathematical approach doesn't work in a small league - even with 6 rounds left most teams are going to be playing eachother - at least one team has to drop points in every game. 40 points should always be enough to make the playoffs. 

Patrick's simulation backs up my point beautifully. Looks like we may not even have to win them all... 

That model works with the assumption that all teams are equal which is obviously not the case. Still interesting.

Does running the simulation 1,000,000 times cover anywhere near all of the possible combinations of results for the remainder of the season?

If it does, then there are no assumptions in the model at all, so should be a very accurate representation of our chances of making the play-offs...

However if it's miles away from all possible scenarios, which it may well be, then perhaps we're back to square one!

I've run the simulation about 10 times now and the results haven't ever diverged by > 0.5%.

Awesome stuff Patrick.  Love your work.

#metrics?

My picks:

Adelaide 2-1 WIN - Siggy with last minute header

Sydney - 1-1 DRAW

Perth - 2-0 LOSS

City - 2-0 LOSS

Victory - 3-2 WIN

Wanderers aka bunnies - 2-1 WIN

So a 4.57% chance of progressing. Better than 0%

The only game I think we have absolutely no chance in is the City game. (because I will be there)

Sydney and Victory are in a bit of a downward spiral and have ACL commitments to focus on.

Fair predictions imo

But if we could win at home this weekend, then steal 3 points at Sydney, we would start to get a roll on, and anything would be possible. 

Whatever Adelaide can do, we can do better (well the song lyrics encouraged me to say that)

Models used for simulations can be (almost) endlessly tweaked to try to improve the accuracy of the results, and if you have sufficient resources or the outcomes are being used for something critical, then improving the model should be done.

But one can spend their time tweaking the model to get a redundant level of precision, or one can just run the simulations and get 'good enough' results. The way A-League matches have surprised this season (eg WSW conceding 5 at home to Nix), I've no qualms with the way the results have been predicted per simulation.

Well done patrick478.

Hope you are right, but  I doubt we will win 40% of those.

I agree. Because we'd have to win 2.4 games to win 40% and you can't win 0.4 of a game. 

That game away to the Jets -at the final whistle it felt just like we won 0.4 of the game.

Hare up Krishna

Realistically, is it better to leave him out for the rest of the season?