The J(uro) Curve

[Juro]

Now that we have a few games under our belt for the season, I think it's time for me to start looking at where we are, similar to what I did last year (but with a snazzier title).

5 weeks in, 2 wins, 3 losses, -49 F/A, sitting 12th on the ladder. But that is only looking backwards. What are our chances of making it to the pointy end of the season?

Well, according to my model, we still have a 39% chance of making the top 8\. This is our 3rd worst effort after 5 rounds, being only better than 2012 (25%) and 2007 (27%), both being 1 win and 4 losses.

Here is the graph of our progress to date, compared against a few of the more recent and/or interesting years. Obviously, it is early days and anything could happen from here.


Other key results:
- 2% chance of minor premiership (Storm are 25%, Souths are 22%)
- 14% chance of finishing in top 4
- 8% chance of wooden spoon (Warriors are 15%, Panthers are 14%, Bulldogs are 13%)

Just a few details on my model. It is a stochastic model, simulating each game so that each team has a 50% chance of winning. Scores are a randomised distribution so the majority of wins are by narrower margins, with the chance of a blow-out still being present. I run 10,000 simulations and record the final position. Percentages are determined by the number of simulations giving a particular result.

I have tried having more complicated models, taking into account a higher chance of winning at home, but it doesn't have a significant impact on the results because most teams have a relatively even distribution of home and away games throughout the year. I've also taken into account the relative position on the table, giving an indication of relative strength, but this gives extreme results.

 

[happy_tiger]

Juro so going off you can tell us who will run 1st to 16th ,is that correct ??

 

[Gary_Bakerloo]

Just a few details on my model. It is a stochastic model, simulating each game so that each team has a 50% chance of winning. Scores are a randomised distribution so the majority of wins are by narrower margins, with the chance of a blow-out still being present. I run 10,000 simulations and record the final position. Percentages are determined by the number of simulations giving a particular result.

Did not know you modeled this stochastically.

Given we are now 5 rounds into the competition would it be possible to have the data exhibit some serial correlation? This would take into account teams hitting form (and maintaining it) from period to period.

 

[Juro]

Juro so going off you can tell us who will run 1st to 16th ,is that correct ??

In any given simulation, all teams are ranked from 1st to 16th after 26 rounds. From that, results are compiled and probabilities are determined.

 

[Juro]

Just a few details on my model. It is a stochastic model, simulating each game so that each team has a 50% chance of winning. Scores are a randomised distribution so the majority of wins are by narrower margins, with the chance of a blow-out still being present. I run 10,000 simulations and record the final position. Percentages are determined by the number of simulations giving a particular result.

Did not know you modeled this stochastically.

Given we are now 5 rounds into the competition would it be possible to have the data exhibit some serial correlation? This would take into account teams hitting form (and maintaining it) from period to period.

I take the view that a simple model is sometimes best. Trying to build in extra complexities does not always help. I take the results simply as a guide, not gospel.

In terms of teams hitting form, each game is simulated independent of the previous results. So if it is a 50% chance of winning a game, it is a 25% chance of winning 2 games in a row, …, 0.1% chance of winning 10 games in a row, etc.

 

[Shooshmeister]

UMmmmmmmm….......You are so smart, S M R T ! All i want to know is will we maqke the top 8 in 2013?

 

[LaT]


 

[Flippedy]

Juro, are you David Middleton? :laughing:

 

[Juro]

Juro, are you David Middleton? :laughing:

I'll take that. Just don't call me Tom Waterhouse!

 

[barra]

Great work Juro!

One pattern this indicates is that the team under Sheens always suffered a 4-5 game losing streak each season (sometimes more than one). Generally they got back on track after, but we know the sort of pressure that can put on the end of season result. It would be interesting to know if this is a pattern for other teams?

But far more interesting to see if a Potter coached side can avoid it!

 

[Juro]

Another week, another loss, and the curve continues to dip. Now at 30% chance of making the top 8\. This is our 3rd worst effort after 6 rounds, being only better than 2012 (19%, 1 win and 5 losses) and 2007 (28%, 1 win, 1 bye and 4 losses).


Other key results:
- 1% chance of minor premiership (Storm are 30%, Manly are 18%)
- 9% chance of finishing in top 4
- 9% chance of wooden spoon (Bulldogs are 18%, Warriors and Panthers are 17%)

Still plenty of time to get this graph pointing up. Bring on the Broncos!

 

[Swordy]

78% of statistics are made up on the spot!

 

[Juro]

Hey, I don't make up my stats! My model may but me? Never!

 

[pHyR3]

How are we the same chance of spooning as getting in the top 4??

 

[LCA]

78% of statistics are made up on the spot!

Like that one!

 

[LCA]

How are we the same chance of spooning as getting in the top 4??

Why can't the probabilities be the same?? It's not a prediction of an outcome, just the relative likelihood of each outcome in the distribution.

 

[tigergirlz]

Is there an injury adjustment, and does it change when the injury is to a marquee player?

_Posted using RoarFEED 2013_

 

[Juro]

Is there an injury adjustment, and does it change when the injury is to a marquee player?

_Posted using RoarFEED 2013_

No. 50% chance of winning each game, regardless of injuries, form, players lost to rep duties, etc.

 

[Juro]

Here we go again. Another week, another loss, and the curve continues to dip. Now at 23% chance of making the top 8\. This is now our worst effort after 7 rounds, with other similar years being 2003 (25%) and 2012 (26%). Both those years were also 2 wins and 5 losses. However, 2003 only had 15 teams so it was easier to make the finals. And our F/A in 2012 after 7 rounds was -16, compared with -60 this year.


Other key results:
- 0.4% chance of minor premiership (Storm are 36%, Souths are 18%)
- 5% chance of finishing in top 4
- 12% chance of wooden spoon (Warriors are 23%, Eels and Bulldogs are 12%)

If we keep losing, I might need to start showing graphs of getting the spoon.

Come on, J curve, turn, damn you!!

 

[Juro]

Another week, another loss, but at least the injury curve has turned the corner. Yes, I am grasping for positives.

We are now at 15% chance of making the top 8\. This is now our worst effort after 8 rounds, with the only other similar year being 2003 (16%). We are now more likely to get the spoon (18%) than play the finals, something that has not happened since Round 16, 2009\. This is also the earliest in the season for this to happen to us. The previous worst was Round 10, in 2003.

So we have 2 graphs this week. First we look up.


Then we look down.


Please note the scale on the spoon graph only goes to 40%. Hopefully I won't need to extend that!

Key results:
- 0.1% chance of minor premiership (Storm are 27%, Souths are 24%)
- 3% chance of finishing in top 4
- 15% chance of finishing in top 8
- 18% chance of wooden spoon (Warriors are 17%, Eels are 16%, and Panthers are 15%)