Yep. And League points per game that year were higher then (0.55) than they are now (0.44).

Exactly right. They don't really show the distribution across each range either. It's the boundaries between top4 / top8 and top8 / the rest that are important in that plot. 2003 and 2014 for instance, which were Halifax and London's respective years of total collapse. Take those away and for those years it's a much more competitive picture.

What we really need is something that takes into account who's beating who, not just how many wins they're getting. Step forward rlBradley Terryrl.

This is handy for things like, for the sake of argument, wine tasting. Which do you prefer, A or B. A beats B. B beats C. But then C beats A. No clear winner. Which one's the best if this situation and situations like it keep repeating over thousands of comparisons?

Bradley Terry gives us a score for a team / wine / whatever based on the relative strengths of who is beating who, and what that opponent had achieved up to that point. Winning games doesn't matter so much as who you are winning against. Beating Leeds last year would have got you a lot of "ability score" in this model. This year, not so much.

There is no unit for ability of course, the numbers it chucks out only make sense when you compare them against the ability scores of other teams. It's just a means of comparison rather than an absolute declaration. But the closer these scores all are then the more competitive the league is.

So if we find out what the standard deviation is (lower is better) for each year...



A steady improvement since the introduction of the salary cap (98? 99? I forget). Which is weird, because everyone knows the salary cap doesn't work and needs to be scrapped. (Unless we assume the league got more competitive when Bradford stopped battering everyone, which I'm willing to accept...)

That spike for 2014 is almost entirely down to how bad London were. It would have been a lot bigger if they hasn't beaten Leeds at Headingley that year.

FWIW, 2016's ability scores from this look like this atmhttp://i.imgur.com/0UMq5Sg.png" >



Still not mathematically certain, but psychologically it might be. At least 8 or 9 wins from the remaining 14 are required. Ten or eleven will see them in the S8s for sure. Any less and it's dangerous territory. Problem is, that's a win rate two and a bit times higher than that achieved so far.

the salary cap was introduced in 2001 for the 2002 season so theres not really that correlation. Its probably more likely (on that limited data) that the introduction of the play-offs in 1998 was the catalyst.

Id also ask how, if last year 2014 was down entirely to how bad London were, why we dont see a similar spike in 2005 with the terrible Leigh side?

as regards the middle 8s, i think there is something of an over-estimation here. Lets be honest, if, starting this week, Leeds won every other game, we would go in to the last game needing a win, but would anyone think we had gone in to it in outstanding form? in even good form? Theres no getting around it, 8 wins out of 14 isnt difficult in SL. We sneak 2 of those close games we have lost and its the form we are showing now. Which is terrible.

Every reference I can find, apart from one nonsensical academic paper which suggests '97, says it was '99.

I genuinely don't recall myself. Both Wikipedia and the RFL site are sadly lacking on this, but that does seem about right to me. Also, it might be true to say that the salary cap proper started when you say because there were those years when Wigan were allowed special dispensation to allow for their existing contracts. And nobody else needed that. Or was allowed it, depending on how that actually went down.

Besides, whilst it slowly moves from around 2000 the great leap forward happens after 2005.



Fair question. You might also ask the same thing about 2003 with Halifax.

The primary reason is that it was way more predictable then. This system is very much dependant on the relative strengths of who is being beaten by the team in question. Back then, top teams beat bottom teams. Top teams beat mid table teams. The old school Big 4 beat everyone, for the most part, and then took points off each other. Exceptions here and there, but largely not. Many fewer exceptions overall than now. I can prove this, and I will if you want me to.

This is exactly what the plot is showing. Despite that spike from London, it is lower than all seasons prior to and including the very year you're pointing out.

In 2014 London were disproportionally bad compared to everyone else that season, including Bradford.

London's sole win was from beating a 6th placed Leeds that year. Solid result. Second from bottom Bradford also beat Leeds. But they also had wins against Wakefield (twice), London (twice), Warrington, Wigan and Hull FC. Not a bad set of scalps overall, and from all over the table. Obviously a terrible season, but both in the table and by this method way better than London.

In 2005 Leigh beat a mid table team (London) and got a draw against another (Hull) and also beat Wakefield, and did so at a time when beating teams from outside your station was a little rarer than it is today. It's one of the worst runs in SL history but it's better than London's. The important point is that it happened in the context of a worse league overall.

In 2003 Halifax beat London, and in their first game. And then never won or drew again. That is easily the worst losing run in SL history. A more abject collapse than London 2014 for sure. London 2003, who Halifax beat, finished fifth. They were ten points behind fourth place Saints.

Leeds 2014, who London 2014 beat, finished the regular season in sixth. They were six points behind first placed Wigan.

All twelve team seasons too.

Halifax and Leigh may well have been terrible. But the league overall was terrible, hence the higher overall scoring in the output of this system. The league was not terrible in 2014. Bradford weren't that great but London happened to equal the worst performance in the league's history, and that made the gap between them and next worst huge. It stands out when determining standard deviation in a sample size of twelve, albeit one built from around knocking on for 180 games overall.

FWIW, I don't deny at all that the play offs are a factor here. They absolutely are. I have always been in favour of a grand final and associated play off system for precisely this reason. But the cap is a huge factor too, and whilst I can't particularly prove it I would certainly argue that it was the primary one.



Not impossible by any means. As you say, it's every other game at this point, though I do think 8 wins is at the lower end of what you're going to need. Doable though. I believe more likely today for a bottom of the table side than at any other point in SL history, and certainly more probable for you than for Huddersfield.

FWIW, this system only measures wins and losses. For draws I mark both teams as a win. It doesn't take points difference into account at all. I prefer it for pairwise comparisons, cos a win is a win is a win.

There are plenty of other systems that are used in US college sports for figuring out which of these teams in a league of over a hundred, some of whom have never even heard of each other never mind played, make it to the play-offs. These do take points difference into account. rlKen Massey'srl is popular, and whilst I've got a method for it I've never tried doing something like this with it.

Leeds will make top 4 easy, can't see Batley, Leigh, London, Bradford getting above you.

2001 Wigan were spending 3.2m on wages. That would be nearly 5m today. They were given 1 years grace at 2.3m in 2002. by 2003 everyone was spending 1.8m. Which as much as anything simply highlights just how far we have gone backwards.

there is another correlation you may have missed which correlates far better and seems a far more obvious reason that passes occams razor.

from 1997, the bottom team in SL scored 9, 4, 10, 8, 11, 0, 12, 5, and then in 2005 we relegated not only the terrible Leigh, but also the 2nd worst side in SL Widnes and replaced them with a side who had money and spent it and built. Les Catalans entered the league and stopped the perpetual merry-go-round of clubs not big enough for SL replacing each other and swapping players.

in 2006 Les Catalans finished bottom with a record points total of 16 which was more points than Widnes got in 11th the previous year.

'"

£900k off our wage bill from one season to the next and we still remained competitive. How much £ in salaries did Leeds let go of last year?

Agree with your points on Catalan though, they've been a great addition to Super League.

Was there an earlier date when the 50% rule first came in? That might explain the difference.

There are certainly references to a cap of some kind existing prior to the dates you mention, and not speculating about its coming. rlTalking about its effectsrl. rlHererl as well. And rlthisrl, which Google picked up '99, has Caisley talking about it too (para 19) .

I don't doubt you, what you're saying is starting to ring a bell (my initial comment you picked up on was an aside really, and my interest in the analysis in the first place was comparing with the NRL rather than looking at the effects of the cap). It's just odd that there are references to it before then.

yeah, im not sure when exactly it came in but the 50% rule came in first, not really a hard cap. Wakefield of all clubs were the first to break it in 2001, spending 60%.

One thing that does tell us though is that in the early 2000's Wigan were turning over at least 6.4m, which despite growth in their crowds and a huge cash increase in TV money (if not really in real terms) Wigans turnover hasnt grown an awful lot in 15 years. Which lets face it is the major problem the game faces. Income has stagnated at most clubs and we just arent doing a huge amount to move forward.

With regards to the NRL and the comparison there, the major difference between us is player numbers. NRL has a surplus, we have a deficit. They have tons of players busting to get to the next level, we have too many clubs scraping the same pool. And i would say that rather than make it better (which it should have done) the recent changes have made that worse. The cut to 12 should have seen a concentration of talent, except instead of us going from 14 clubs trying to make 14 SL sides out our player pool to 12 clubs trying to make SL sides out of the same pool, we have gone to 14 plus 3 or 4 kinda trying.

To be honest, I don't give a monkey's about players. I care about competitive games. It is far and away my primary interest. I want close games, more distribution of wins and doubt over the victor at kick off time. It's all I've ever cared about. On that, we are not the far off the NRL at all.

edit: Interesting point on player distribution between the leagues here though. I hadn't considered that, but you're right there.

FWIW, these are the Massey numbers as of this morning for relative strengths of attack and defence.

Attackhttp://i.imgur.com/G8ACSsF.png" >

Defencehttp://i.imgur.com/lAB0M4N.png" >

This system takes into account the relative strengths of who was played, the points difference and whether the results happened at home or away.

Defence isn't great but far from the worst. But on results so far your attack, or rather your ability to score, is very much the worst.

Interesting.

Do you have a combined attack and defence Massey ratings table?

I'd be interested how that matches up to the SL table as it currently stands.