The Snowball stat thread

[cmonurz]

If a team suffers injuries in say, November, a key player has personal problems that affects his form at the same time (maybe he gets married, or his laptop breaks or something), and they get two players suspended for a few games after an on-pitch brawl (one broke the other's laptop, say) – all this cumulates into the club losing a few games they otherwise might not. Why would it not be ‘the same’ to take this run out?

[Snowball]

So what happened from Nov 5th onwards, as up to that point were were only on 1.2PPG, a 55pt season.

2.40 ppg - 110 Points Season Equivalent - 20 Games Since - Nov-06 <<<
2.37 ppg - 109 Points Season Equivalent - 19 Games Since - Nov-19 <<<

2.50 ppg - 115 Points Season Equivalent - 18 Games Since - Nov-26 <<<
2.47 ppg - 114 Points Season Equivalent - 17 Games Since - Nov-29 <<<

2.44 ppg - 112 Points Season Equivalent - 16 Games Since - Dec-03 <<<
2.60 ppg - 120 Points Season Equivalent - 15 Games Since - Dec-10 <<<
2.57 ppg - 118 Points Season Equivalent - 14 Games Since - Dec-17 <<<
2.54 ppg - 117 Points Season Equivalent - 13 Games Since - Dec-26 <<<
2.50 ppg - 115 Points Season Equivalent - 12 Games Since - Dec-31 <<<

2.45 ppg - 113 Points Season Equivalent - 11 Games Since - Jan-02 <<<
2.70 ppg - 124 Points Season Equivalent - 10 Games Since - Jan-14 <<<
2.67 ppg - 123 Points Season Equivalent - 09 Games Since - Jan-21 <<<
3.00 ppg - 138 Points Season Equivalent - 08 Games Since - Jan-28 <<<

3.00 ppg - 138 Points Season Equivalent - 07 Games Since - Feb-11 <<<
3.00 ppg - 138 Points Season Equivalent - 06 Games Since - Feb-14 <<<
3.00 ppg - 138 Points Season Equivalent - 05 Games Since - Feb-17 <<<
3.00 ppg - 138 Points Season Equivalent - 04 Games Since - Feb-25 <<<

3.00 ppg - 138 Points Season Equivalent - 03 Games Since - Mar-03 <<<
3.00 ppg - 138 Points Season Equivalent - 02 Games Since - Mar-06 <<<
3.00 ppg - 138 Points Season Equivalent - 01 Games Since - Mar-10 <<<


It isn't complicated working out why that was the case.

We started the season a shambles, pretty much, with DWLLLL

and six, "not terrible" results individually, DWDDDL

the disappointing defeat at Forest neatly bookending the 15 games

H 0-0 Middlesbrough
A 1-0 Burnley
H 2-2 Derby
H 1-1 Southampton
A 0-0 Palace
A 0-1 Forest

Nov 6th was a win, followed by a defeat

H 1-0 Birmingham
H 1-2 Cardiff


But you make my point for me. It's quite clear that I could start the stats from the 3-2 win v Ipswich (18 Games)
and show RFC in a brilliant light. 2.5 points per game.

[Snowball]

[

Snowball actually provides an example of how this approach is flawed in the last two pages. Highlighting our good defensive run, he says ‘7 goals conceded in 15 games – but 3 were against Cardiff, so it’s 4 goals conceded in the other 14 games’. If Snowball wants to take the Cardiff aberration out of our goals conceded data (admittedly only to highlight our otherwise spectacular run in defence), and he has made other manipulations to illustrate other very valid points, then it is reasonable for some to question how valid it is to compare our run of games that eliminates our patch of poor form, with the same run of games for other teams that does not necessarily ignore their bad runs.

I am not "taking it out". I am merely highlighting the large percentage of our conceded goals occurred in a single game.

It's highlighting the difference between mean (average) and mode. The most common result is not-to-concede, second is conceding one goal only.

What about our poor run, as has been pointed out, from the start of the season to November 5th






The date chosen was not chosen because the earlier games were poor results.

It was chosen because we had begun to find a team.

Gorkss-Pearce a good CB pairing, Khumalo frozen out
two FBs in reasonable form (we had had three injuries to FBs in play in the first 6 games)

We STILL however, had not found a striking partnership, Karacan had lost form, was carrying an injury
or was in trouble for something "off the field', and Elwood was a shadow of his 2010-11 self.

We were very ordinary away to a poor Coventry, we were HORRIBLE at Bristol even tho we somehow won,
we only managed 0-0 at Home to Boro, sneaked a win in the 98th minute at Burnley, then had 3 consecutive
draws before losing 2 in 3 games.

We won just 4 from 11 taking us to 5 from 17 = 1.54 ppg

WDWDWDDDLWL

P11 W4 D5 L2 17 Points That hardly seems like a purple patch. Not exactly cherry-picking

[StroudRoyal]

However, whatever the reason, it is clear that something has changed from game 7 onwards.

No it isn't.

What changed between Game 6 and Game 7? Nothing. No new players joined that I can recall. Same manager. Same ground. Same pitch.

The reason the former is excluded and the latter is not is for one simple reason - we won one and not the other. If you eliminate unfavourable results based on nothing more than the fact they're unfavourable then of course you'll paint a better picture.

Something did change from game 6 onwards - our form improved dramatically and this trend as been maintained. I really wish I could show the graph as it does illustrate the point I'm making. I'm not saying what caused the change simply that it did. All I'm saying is that for the first 6 games our results were poor and for the following 29 games our form has significantly improved and that has been maintained - that's what changed. If we only get 4 points from the next 6 games then It would be equally valid for me to say that there has been another change from that point i.e. that our form has taken a dip compared with the previous 29 games.

[Vision]

Surely all Form Guides are cherry picking to some degree aren't they?

If not they'd be no need for them.

[Snowball]

However, whatever the reason, it is clear that something has changed from game 7 onwards.

No it isn't.

What changed between Game 6 and Game 7? Nothing.

WRONG. It was the first start that season for Cummings, and only the second for Mills.

That same full-back partnership, a new FB pairing, was Game 7.

Game 7 was the first game that partnership had played,
and it then went on an 8-Game undefeated run.

Also Karacan was dropped after the Watford Game (Game 6) and Tabb's first game was Game 7.

And Game 7 was only the third time Gorkss had partnered Pearce (unless you expect a new CB pairing to be instantly good?)

It was also Church's first start of the season, and only Le Fondre's second game for Reading.

No new players joined that I can recall. Same manager. Same ground. Same pitch.

First Start Cummings (Clean Sheet)
First Start Mills (New Signing) (Clean Sheet)
First Start Tabb (Clean Sheet)
First Start Church (Scored)
Only second game for new signing Adam Le Fondre

but apart from that, and apart from it being only the third start for Gorkss, it was just the same.

[Snowball]

If a team suffers injuries in say, November, a key player has personal problems that affects his form at the same time (maybe he gets married, or his laptop breaks or something), and they get two players suspended for a few games after an on-pitch brawl (one broke the other's laptop, say) – all this cumulates into the club losing a few games they otherwise might not. Why would it not be ‘the same’ to take this run out?

We ALSO had a middle run that was poor. Not taken out.

The difference here is we were building a new side, replacing a top player (Long)
discovering Khumalo was a disaster, struggling with the fitness/form of BOTH CBs

So apart from three of the back four being hopeless and one, by all counts, a disaster,
and having two midfielders in temporary trouble, and having lost a 27-Goal player,
there were really no problems, were there?

[Blue Hooped Moose]

What was harsh?

Sorry, maybe I should have used a word like "unfriendly" - all your post was missing was a "...and don't let the door hit you on the ass on the way out". That was your first interaction with me on this thread so it's not surprising that I took your response as harsh.

I don't know why you have such an issue with people querying your cut-off date for this table. You mention that managers / the club / whoever have agreed that our form without those games is impressive but I'd imagine their comments would have been framed along these lines:
"Well Brian, you've hit some impressive form recently and are riding high in the table"
"Yes Tim, it's great - it's a shame we couldn't start the season in such fine form as if we had we'd be right up there in the automatic promotion spots."

To then publish reams of figures excluding a range of dates is not what statistics are about - you paint a full picture, as unbiased as possible, and then allow people to comment on it afterwards. We all know we're doing impressively well this season and hiding the times when we weren't so good just devalues the whole thing.

You mention how form is used throughout the media and it's acceptable to use a window of x-games but it is just that; a rolling window of games to show how teams are performing. There is a reason it only shows a few games and that's because any more shows a distorted picture or becomes worthless, even more so when you're being selective.

It seems to me your background as a writer seeps into your stats and that is not a good thing - as I said, by all means apply commentary after the event but at least portray the whole picture first. Your post about conceding 7 goals in 15 games is a great example; you stick a comment in about the how Leicester's goal was in the 90th - how does this add anything to the stat that we've only conceded 7 in 15? A goal is a goal is a goal, regardless of how or when it was scored.

[Snowball]

Something did change from game 6 onwards - our form improved dramatically and this trend as been maintained. I really wish I could show the graph as it does illustrate the point I'm making. I'm not saying what caused the change simply that it did. All I'm saying is that for the first 6 games our results were poor and for the following 29 games our form has significantly improved and that has been maintained - that's what changed. If we only get 4 points from the next 6 games then It would be equally valid for me to say that there has been another change from that point i.e. that our form has taken a dip compared with the previous 29 games.

Yes. Not only that, the step up in form has been improve/plateau/improve

Games 7-19 were not exactly "brilliant..."

It's not as if dropping Games 1-6 is dumping merely "dodgy results"

Long, Manset, Griffin, Khumalo were starters Game 1. 3 have left and one is a fourth choice now.

[cmonurz]

That doesn’t answer my question, to be fair. There are ‘bad runs’ for all sides that supporters of those sides might reasonably equate to the ‘problems’ we suffered in our first six games of the season, just their runs happen to be in the middle of the last 29 games, not before them.

[Maguire]

What changed between Game 6 and Game 7? Nothing

WRONG. It was only the second [start[ for Mills.

only the third time Gorkss had partnered Pearce

only Le Fondre's second game for Reading.

only the third start for Gorkss

So you say i'm WRONG then list all these things which, guess what, HAD ALL HAPPENED BEFORE

How stupid do you have to be?

[Snowball]

"Well Brian, you've hit some impressive form recently and are riding high in the table"
"Yes Tim, it's great - it's a shame we couldn't start the season in such fine form as if we had we'd be right up there in the automatic promotion spots."

Actually he has NOT said that.

he has (and so has the OS, and so has the programme)
simply said "We are top of the table over the last XX games"

To then publish reams of figures excluding a range of dates is not what statistics are about - you paint a full picture, as unbiased as possible, and then allow people to comment on it afterwards. We all know we're doing impressively well this season and hiding the times when we weren't so good just devalues the whole thing.

That is simply WRONG.

There is an extremely sound reason for ignoring the first six games. The reasons have been given dozens of times.

Factoring in the first six games IS A DISTORTION OF THE SIDE'S TRUE VALUE.

In the same way as the signing of Roberts and Connolly has made a fantastic difference to the CURRENT side's "value"
so did the changes made after game 4/5/6.

And that is why for at least three months I have been saying we are BETTER THAN OUR LEAGUE POSITION would otherwise indicate.

And only know are the bookies waking up.

We have out-pointed every team in this league now for a 29 consecutive game table, no gaps. That is a real, true, totally honest statistic.

It doesn't mean that the first 6 games won't count. We all know that.

You mention how form is used throughout the media and it's acceptable to use a window of x-games but it is just that; a rolling window of games to show how teams are performing. There is a reason it only shows a few games and that's because any more shows a distorted picture or becomes worthless, even more so when you're being selective.

I disagree. I made the statement 3-4 months ago and have stuck to my guns
DESPITE THE FACT that more recent form-guides make us look even better.

Your post about conceding 7 goals in 15 games is a great example; you stick a comment in about the how Leicester's goal was in the 90th -
how does this add anything to the stat that we've only conceded 7 in 15? A goal is a goal is a goal, regardless of how or when it was scored.

Did I expand on it? No, I made a simple, factual statement.

And the word and his mother knows that a side winning 3-0 often concedes a dumb late goal.

There is even a term for it... "a consolation goal"

[Snowball]

That doesn’t answer my question, to be fair. There are ‘bad runs’ for all sides that supporters of those sides might reasonably equate to the ‘problems’ we suffered in our first six games of the season, just their runs happen to be in the middle of the last 29 games, not before them.

OIC. You mean "building a new team in December", that sort of thing?

[Snowball]

What changed between Game 6 and Game 7? Nothing

WRONG. It was only the second [start[ for Mills.

only the third time Gorkss had partnered Pearce

only Le Fondre's second game for Reading.

only the third start for Gorkss

So you say i'm WRONG then list all these things which, guess what, HAD ALL HAPPENED BEFORE

How stupid do you have to be?

Dear, Esteemed Debater,

You seem to have forgotten, overlooked or deliberately ignored these "Firsts".

Game 7. First Start. (Clean Sheet) Cummings
Game 7. First Start. (Clean Sheet) Tabb
Game 7. First Start. (Clean Sheet) Brand New Full-back Pairing
Game 7. First Start. Church (scored)

[cmonurz]

That doesn’t answer my question, to be fair. There are ‘bad runs’ for all sides that supporters of those sides might reasonably equate to the ‘problems’ we suffered in our first six games of the season, just their runs happen to be in the middle of the last 29 games, not before them.

OIC. You mean "building a new team in December", that sort of thing?

No, I meant what I said. Injuries, suspensions, players getting married, new players coming back into the team or enjoying a run of games they didn’t get previously. All relevant to others clubs but may have happened at a different point in the season.

[MmmMonsterMunch]

What was harsh?

Sorry, maybe I should have used a word like "unfriendly" - all your post was missing was a "...and don't let the door hit you on the ass on the way out". That was your first interaction with me on this thread so it's not surprising that I took your response as harsh.

I don't know why you have such an issue with people querying your cut-off date for this table. You mention that managers / the club / whoever have agreed that our form without those games is impressive but I'd imagine their comments would have been framed along these lines:
"Well Brian, you've hit some impressive form recently and are riding high in the table"
"Yes Tim, it's great - it's a shame we couldn't start the season in such fine form as if we had we'd be right up there in the automatic promotion spots."

To then publish reams of figures excluding a range of dates is not what statistics are about - you paint a full picture, as unbiased as possible, and then allow people to comment on it afterwards. We all know we're doing impressively well this season and hiding the times when we weren't so good just devalues the whole thing.

You mention how form is used throughout the media and it's acceptable to use a window of x-games but it is just that; a rolling window of games to show how teams are performing. There is a reason it only shows a few games and that's because any more shows a distorted picture or becomes worthless, even more so when you're being selective.

It seems to me your background as a writer seeps into your stats and that is not a good thing - as I said, by all means apply commentary after the event but at least portray the whole picture first. Your post about conceding 7 goals in 15 games is a great example; you stick a comment in about the how Leicester's goal was in the 90th - how does this add anything to the stat that we've only conceded 7 in 15? A goal is a goal is a goal, regardless of how or when it was scored.

Brilliant post & far more eloquent than I could have written. Hit the nail firmly on the head there.

[Maguire]

WRONG. It was only the second [start[ for Mills.

only the third time Gorkss had partnered Pearce

only Le Fondre's second game for Reading.

only the third start for Gorkss

So you say i'm WRONG then list all these things which, guess what, HAD ALL HAPPENED BEFORE

How stupid do you have to be?

Dear, Esteemed Debater,

You seem to have forgotten, overlooked or deliberately ignored these "Firsts".

Game 7. First Start. (Clean Sheet) Cummings
Game 7. First Start. (Clean Sheet) Tabb
Game 7. First Start. (Clean Sheet) Brand New Full-back Pairing
Game 7. First Start. Church (scored)

So you mean "oh yeah, shit, so i did"?

Good to know that the addition of Tabb and Church to the starting line up is what helped us turn the corner though

[Ian Royal]

However, whatever the reason, it is clear that something has changed from game 7 onwards.

No it isn't.

What changed between Game 6 and Game 7? Nothing. No new players joined that I can recall. Same manager. Same ground. Same pitch.

The reason the former is excluded and the latter is not is for one simple reason - we won one and not the other. If you eliminate unfavourable results based on nothing more than the fact they're unfavourable then of course you'll paint a better picture.

Something did change from game 6 onwards - our form improved dramatically and this trend as been maintained. I really wish I could show the graph as it does illustrate the point I'm making. I'm not saying what caused the change simply that it did. All I'm saying is that for the first 6 games our results were poor and for the following 29 games our form has significantly improved and that has been maintained - that's what changed. If we only get 4 points from the next 6 games then It would be equally valid for me to say that there has been another change from that point i.e. that our form has taken a dip compared with the previous 29 games.

Saying something happened is not entirely accurate. Results improved yes, that isn't necessarily caused by a significant change that occurred at that time. A better way to pick a more selective period than the entire season, would be to find an event first and use that point and compare results from then. Not find the point in the results where things started to get better ... that's going at it backwards. You then look at your data and see if there is evidence that event affected things.

If you want to use Gorkss as a trigger, do a table for all games since Gorkss has joined. If you want to use the transfer window, use all games since the transfer window. If you want to say since we finsihed making transfers + a period of time to allow them all to settle in, pick the transfer window date + 2 weeks or a month for example, of go for Rodgers method and include 5 games of the season for an extra pre-season piss about. ALthough obviously most of these will include too small a sample pre-event to be reliable as a comparison set.

Don't just look at the data and find the point where it starts to look good and use that. Assuming one or several things did happen to change our season, who's to say that it happened by 17 Sept? It could have been November and first few results of the period chosen by snowball were an aberration in form, or temporary blip made permanent by a later event..

[Snowball]

How about Cummings as a trigger?

How about "A new FB pairing" as a trigger?


Go back and read how much happier people were with Cummings in for Griffin.

And Griffin broke down in the 72nd minute of Game 3 (our first defeat, at Portsmouth)

He was out in the next game and we played Gunnarson who went off in the
74th minute and we brought on MANSET and put a midfielder at RB

Griffin came back in for the following game and broke down again in the 79th,
replaced by CHURCH... did McAnuff go RB that time?

So in 7 games we played

Griffin, Gunnarson, HRK?, McAnuff, Cummings at RB

and when we stuck with Cummings (and Mills at LB) we went 8 games undefeated.

[Ian Royal]

How about Cummings as a trigger?

How about "A new FB pairing" as a trigger?

Easily rejected seeing as we have continued to be great with Cummings out of the side and Left Back personel have been even more changeable in that period. As have Tabb & Church being in the side.

Don't look at the data to find a theory. Find a theory and check it with data. You're doing things backwards.