By Tony Attwood
Just how bad are we in November? Not just this November but each and every November. Year after Wengerian year after Wengerian year…
Curiously, given that this year this has been the topic there hasn’t been any real analysis of this. Indeed I have seen commentators writing that this is the worst November ever for Arsenal, or at least the worst in the current manager’s reign but without quoting any figures.
Mind you, it was ever thus. So I thought I’d better waste a Sunday morning doing the work no one else wants to do.
This November we’ve played two and drawn two, so depending on your outlook we’ve either dropped four points or are unbeaten. Two 1-1 draws then, against Man U and the Tinies. Now let’s go back in time (you can make whoooooooo type noises if you like as they would if this were on TV) and see what the last 10 years have been like.
And then for fun let’s compare with the last three times we won the league.
Two draws and a defeat. 1.5 points a game – final league position 2nd.
- We drew with Tottenham and Norwich, and lost to West Brom.
Two wins and two defeats. 1.5 points a game – final league position 3rd.
- We beat Burnley and WBA and lost to Swansea and Man U.
Three wins and one defeat. 2.25 points per game – final league position 4th.
- We beat Liverpool, Southampton, and Cardiff, and lost to Man U.
One win, two draws, one defeat. 1.25 points per game- final league position 4th.
- We beat the Tiny Totts, drew with Aston Villa and Fulham and lost to Man U.
Two wins and one draw. 2.33 points per game – final league position 3rd
- We beat WBA and Norwich, and drew with Fulham.
Three wins and two defeats. 1.8 points per game – final league position 3rd
- We beat Wolverhampton, Everton and Aston Villa, and lost to Newcastle and Tottenham.
One win and two defeats. 1 point per game – final league position 4th.
- We beat Wolverhampton and lost to Sunderland and Chelsea.
Two wins and three defeats. 1.2 points per game. Final league position 3rd.
- We beat Man U and Chelsea and lost to Stoke, Aston Villa and Man City.
Two wins and one draw. 2.33 points per game. Final league position 4th.
- We beat Reading and Wigan and drew with Man U.
So that is the last nine years and two of the games in this one. The best year in terms of points per game in November were 2011 and 2007 when we got 2.33 points per game and came third and fourth in the league. The worst was 2009, when we got one point per game and came fourth in the league. So not much of an indicator there.
Last season when we came second we got 1.5 points per game. In five seasons out of the ten we did better than this in November and yet came lower in the league at the end.
But what about the years when we won the league? How did we do in November?
One win and three defeats. 0.75 points per game. Final league position 1st.
- We beat Man U, and lost to Derby, Sheffield Wednesday and Liverpool.
One win, one draw and one defeat. 1.33 points per game. Final position 1st.
- We beat Man U, lost to Charlton and drew with Tottenham
Three wins and one draw. 2.25 points per game. Final position 1st (unbeaten)
- We beat Leeds, Tottenham and Birmingham, and drew with Fulham.
Our best ever Novembers in these years (the last ten and the three most recent championship years) were 2011 and 2007, when we came third and fourth. The three seasons in which we won the league we did worse than this in November.
So does this mean you can prove anything with statistics?
No not at all, but it does mean that you can prove pretty much what you want to prove by selecting very few statistics. What it shows is that just taking one month and trying to extrapolate from that to the whole season is meaningless. It shows what must be blindingly obvious to everyone, that you cannot judge 38 games on the basis of between three and five games.
And yet the media and the bloggettas go on and on and on and on and on and on and on (plus a few more) talking about Arsenal’s disaster in November and how this is a perfect example of Arsenal’s inability to win anything under Mr Wenger.
In 2001/2 we had an average points per game ratio of 2.29. In November it dropped slightly below that. But so what? Getting above that level in November doesn’t mean anything at all and most certainly doesn’t mean we will win the league.
In fact whoever told you that November was some sort of indicator to the whole season, and that Arsenal always fail because of November was talking rubbish. To know how well Arsenal do in the total season you need to look at the league table at the end of the season.
But here’s a funny thing. In the 13 Novembers we have looked at, in six of them we have played Man U and five of them Tottenham. However if the games across the season were drawn out randomly we would expect to play each team either two or three times.
Yet we played Man U six times and Tottenham five times. Is that just a quirk, or after the computer has drawn out all of the league fixtures, is someone mucking around with the fixtures, giving Arsenal two of the games that bring an extra intensity.
Man United – Arsenal : 1-1 Our beach saves the day
Man U v Arsenal – the teams, the pizza, the telepathy, the result
José Mourinho vs Arsène Wenger. As far as the media is concerned the teams are irrelevant
Wenger v “M”. Tactical naivety, blaming the medics,
Manchester United v Arsenal 19/11/2016 – The Match Officials
Previews: A (Possibly) Thrilling End to a Great Year
It’s November so it must be injury crisis time (and ticket price analysis time)
“Making the Arsenal” and “Woolwich Arsenal the club that changed history” are available both via Kindle and in paperback. For the Kindle editions please go straight to the Kindle shop, for paperback editions and more information about the books please click here.
27 Replies to “Arsenal in November. Exactly how bad are we year after year? You might be surprised.”
Interesting analysis, and didn’t realise that about the fixtures. Have often thought we tend to get some difficult games coming very soon after ECL games in years gone by, but hadn’t looked at the EPL games we faced.
So the Nov thing could be yet another myth, though it wouldn’t surprise me at all if we did actually struggle in this month.
The reason, our style of play vs the cumulative disruption of the awful interlulls…..and the injuries they can bring. Yes, I know it’s a squad game, and that others have to face the same issues. But, some teams have managers that coach them within an inch of their lives, they are coached and controlled on all aspects of a particular game. And yes, this can be a successful policy. But Wenger encourages a much more freestyle pattern of play. This needs speed of foot, of thought, familiarity, momentum and confidence. Constant interlulls in the autumn followed by early kick offs disrupt this wengers teams are at their best when fresh, and in a long run of togetherness. Hopefully, that is coming.
In all novembers many players come and go.But that man in charge is always there and is stinking.
In years gone by, Jose was a Special one and United were the best team in the league. Old trafford isnt difficult anymore, we just made it look it.
Welcome November with a new managerza
Well, whatever be the case, Le Prof himself has agreed in one of his press conferences that the month of November has been a difficult month in our League campaign for Arsenal to overcome without suffered a glitch.
And so far in this month of Nov this season, we’ve not lost in the League and in the Ucl after playing 3 games in the 2 competitions. And we look to beat PSG, Bournemouth and Southampton all at the Ems in all competitions in this very month of November to make it 4 wins & 2 draws. God’s willing.
And this will be great for us if we achieve these good results this November. Which will be a departure from the past bad results we’ve had before in 3 competition campaigns in the past months of November. But sorry, I’ve no facts to substantiate my claims in respect of the Cup games we had played in the past months of November.
As so often, alleged truths don’t turn out to be quite so true when examined in detail. But the main point Tony is making – that how you perform in November doesn’t decide where you end up in May – is important. The pundits and media are always obsessed with the moment and trying to make it as dramatic as possible – as well as all saying much the same thing at the same time.
very kind of you to debunk this useless november hoodoo going around,
i always thought february was our sinkhole, i’d really love to see how our february games have panned out over the years.
Apart from that strange quirk of playing Spurs and Manure so often in November there are other common features of the month. We always play two Champion League games (both of them usually vital) – and the vast majority of our EPL opponents don’t. We nearly always play a League Cup game which will include (these days) at least a few ‘first-teamers’. Again our opponents within the month may not.
November also contains an international break (following hot on the heels of another!) which will involve a disproportionately high number of our players travelling long distances and risking injury. Other teams may have the same problem but, again, some that we might play in the month won’t.
Arsenal therefore have all three factors against them (being victims of their own success) while opponents may have two or less.
In a game where a split second of indecision or half-a-yard of slowness can make all the difference we will always carry the greatest burden and run the greatest risk. The facts that Tony presents seem to indicate that we actually could be seen to over-perform in the month given the problems we face and actually the fact that we tend to improve as the season wears on shows great (and consistent) inner strength. A tribute to both players and manager.
90% sure that the people behind the fixture list admitted they often use the previous year’s fixtures as a template of sorts.
That was a year or two back when one of the main guys responsible for the fixtures did a bit of media stuff discussing it.
I can’t remember what the justification was. It might have simply been that compiling the fixtures is an extremely complex business and that if the current year’s fixtures satisfy the criteria then they will generally do so for the next year. Safe to say, i din’t buy it as an explanation.
I’ve had a look but not been able to find the article I read which directly addressed why teams can end up playing each other at the same times year after year. Quite a few articles about explaining some of the complexities, though
Glenn Thompson is the fella responsible, for an IT company called Atos.
As I was reading the article I was thinking to myself how often certain teams where turning up.
Also how often we we get two difficult games one after the other.
@Pat, Mandy, Samuel and Tony,
the plain truth regarding our November issue is if you compare the average points earned per game in each month of the season.
The fact is that Arsenal earns less points per game (on average) in November than in any other month. It’s a fact. An undisputed fact. There might be reasons for it but the fact remains that we get less points per game in November.
Arsenal’s objective is to get as many points as possible so as to WIN the league title (or at least get to the Top 4).
If we get on average 1.6 points on November in Year 201X but finish 3, and then get on average 1.8 points per game in year 201Y but finish 4th, it doesn’t mean that the fact that we are getting less points in November doesn’t matter.
If we had got on average 2.3 points or 2.75 points in November we might have moved up the final league table by one or two places.
Every point matters. It just so happens that in November we are getting less points per game than in other months.
Tony, asked about what we take of the current season. The stats for the current year are plain. We have played two league games and drew both of them. The average points per game is 1. This average is even lower than the long term average.
We may say that we had tough opponents in Spuds and Manure. But really? Really? Is that a fair excuse?
I think both Manure and the Spuds were there for the taking this year. We simply didn’t turn up. We benefited by playing both these teams when they were weak and we were going through our strong patch. We still failed to win either game.
Even Arsene Wenger has confirmed that in November we have historically earned less points per game than in other months.
Every points counts, we cannot explicitly or implicitly say that the fact that we earn less points per month in November doesn’t matter.
The only time I would accept such an argument is if we always win the league despite us not performing well in November.
I find it incomprehensible the reason Le Prof had decided to start Ramsey at right wing and started Walcott at left wing in the Man U game we had yesterday.
Le Prof’s experimentation of Ramsey at RW and he’s shifting of Walcott to the left wing had not worked in the Man U game. And it may not work in subsequent games because Ramsey is not a specialist RW player but a good central midfielder. While Walcott is a specialist RW and not a specialist LW player.
Apparently, Le Prof decided to thinker with what has become his standard 1st team starters by eliminating Iwobi to the bench and shifted Walcott to the left so that he can play Ramsey. Good and fine if he plays Ramsey. But at RW? And the Ramsey who is yet to be at the top of his game in a big game for that matter?
Ramsey has no width. He can’t operate efficiently from the wide because he’s not a natural wide man. To continue to force him to play there is like trying to bend a dry fish. I think since Cazorla is not available, the natural player that should step into his deep lying position is Ramsey alongside LeCoq as DM.
I think it’s this singular error in starting Ramsey and Walcott in wrong positions in the team that has led to the lost of the traded fluency in the Gunners game. Which has resulted in their lack of ability to creat good goals scoring chances in the Man U game.
We had no width in the Man U game due to this mis positional start of 2 players. But when a natural width playing man, Oxchambo was brought on, we all witnessed the positive result of that action taken by Le Prof.
Agree Zuruvi, both were tricky games, but completely there for the taking, the team just weren’t themselves really in either. We just didn’t get enough shots in in either game, and would add the Middlesboro game .
Know many will disagree and have good reasons for doing so, but convinced the loss of Santi is a big issue, not so much that other quality players cannot step up, perhaps more that our creatives seem very reliant on the way this special player is able to control games.
Sadly, Santi seems to have underlying Achilles issues, notoriously difficult to treat and predict recovery, Mr Wenger is going to need to find a way around this. A shame Jack is not around, I believe when fully fit, he is the closest to Santi, but no point crying over spilled milk I guess
2015. 2 points from 3 games surely equates to 0.66 points per game .
Gord did some analyses looking at trends through the season when this topic was raised a couple of weeks ago. If I recall right he found that there was little to be concluded and indeed it was rather silly to do so.
I have been mentioning the somewhat fatuous view of our “performance” this year which do not reflect the fuller picture.
Your Sunday stats study emphasises that whilst the data is there, attaching undue significance it can be misleading and unjustified. Indeed stats themselves should be evaluated to see if they do represent anything and if anything worthwhile can be concluded or be worthy of further investigation.
Generalisations arising from bad use of selected stats are all around us and often are used disingenuously, lazily, or unwittingly.
Testing theories and understanding the background data is a good excercise and may itself lead to more relevant conclusions, not least about the person or organisation posting them.
One of the reasons I support this site is that the source is generally stated or that assertions such as the ref reviews include underpinning evidence. We are entitled to disagree but provided it is based on evidence it doesn’t normally descend into schoolyard namecalling as seen on some sites ( I find contradictory or questioning views to my own to help my understanding and sometimes (obviously) might be valid or even correct.)
Seriously as of me I don’t believe in fate, so all those just pass. Back on yesterday’s game how could we expect to win when we just had one shot on target!!! For 98minutes??I’m not a coach but in our days defender or striker we could train to shoot at goal more than 200 times in training. Title contenders one shot in the entire 98 Minutes and 50 something possession?? What is football if I may ask??
Just imagine if we had taken two shots on target.
I am still working at the data.
There appears to be some difference with a region before the end of the year, and both the beginning of the season and the end of the season. The problem is that the amount of difference is probably not statistically significant.
There are a bunch of questions. Is it a month that is different, or some other length of time? How is it different? Some Novembers have fewer EPL games than others (breaks in play), is this a cause?
Not every season has a break in November, or at least what I call or recognize as a break. If there is a break, the first game after the break is about Nov 20, but it can be as early as Nov 15.
Looking at years (sinece Wenger came in 1996) where I think there is a break, our average points per game in that first game after the break is 1.58. Wonderful, I proved everyone’s point. Hold it. I never calculated the variance in that average.
The second game back, our average is 2, and the third game back it is 2.42. Again, no variance for any of those averages. But it is highly unlikely that any November effect would be noticed by the medja, if it comprised a single game after a break in November, if the season had a break in November.
What about before the break? The game immediately before the break was 1.67, the game before that was 2.08 and the game before that is 2.0.
Just over half the years since Wenger arrived have a break in November (12 of 20). For this 6 game interval associated with a break, 1/3 of our points per game are less than 2, a 1/3 are at 2, and a third are greater than 2.
Assuming EPL standard scheduling, the 3rd game post break is 2 weeks after, the break is probably 10-14 days, and the 3rd game pre-break is 2 weeks. So this 6 game interval is (in general) over 1 month long.
Oh well, back to numbers.
Ahahaha the last few sentences of the article cracked me up.. No other Arsenal blog comes close to this i’m sorry.
I won’t entertain the insuniation because the old adage, you can bring a horse to water but you can’t make it drink, comes to mind.
20/11/2016 at 3:07 pm
Just imagine if we had taken two shots on target. <—–LOL again, great comment mate, funny but in a way true!
Just checking because my math skills aren’t great – wasn’t 2015 our lowest for average points per game at 0.66? Backs up the argument as we finished in a relatively high position! So confused by information this year!
Interesting article, i too had noticed the frequency of which we played Tottenham and Utd from your analysis, i might now look back at how many Novembers also include short trips up north after international breaks, or big matches and quick turn arounds between conclusive CL matches…
Just a thought though but did anyone notice if the media considered how everyone else’s November’s compare to their respective other months?
Alex – you can smell chicken shit from a chickens Arse but you wont find any at Arsenal. Long live Wenger & dissenters are welcome to go looking for chickens.
Gord – One thing the media have not picked on yet is that every season Arsenal start at the top because of a alpha quirk.
Think you will find that the honour of starting the season at the top of the table now falls to AFC Bournemouth
0.66 points per game times 38 games is 25 (.08) points for the season. So yes, your math is wrong. We finished with 71 points, which when we divide by 38 games is 1.87 points per game.
Relegation is typically close to an average of 1 point per game (tie every game on average): 38 points. Winning the league can be as low as averaging 2 points per game (alternating between winning and tying): 76 points.
Gord – sorry, I should have articulated that I was referring to results in ‘November 2015’, in keeping with the theme of the article!
2 draws and a defeat. 2pts ÷ 3 games = 0.6666666667pts.
It was. Last November was rough on us!
I have been working with “moving averages” of this point per game data, most of which are over a 6 game period. Which is a bit more than 1 month (expected to be 6 weeks, but it varies). The standard deviation for values at any time of year is about 1.2. And one normally looks for more than one standard deviation of difference, before one considers something is different. Well, 0.67 plus 1.2 is about 1.87, which is close to the year long average points per game. Not even some multiple of 1.2. There is no statistical basis to say that November is different.
Or, there is no way without introducing some kind of model. I don’t know yet if a person can find a model (and explain how this model should be considered) that allows for “November” being different.