Game Day Trends - Part Deux

Game Day Trends, The Follow up.

Hello Football Fans, you may remember an article I wrote all the way back in Game Week 9, on Game Day Trends, well this is the end of season follow up to that article. Back in GW9 after watching a handful of 'Bore Draws' I wondered if there was any pattern occurring relating to results on certain week days or if it was just my bad luck having just watched these games in the previous weeks.

After investigating the first 8 game weeks of the season I noted some surprising trends, at the time I wasn't sure if they were anomalies, so suggested that it would be appropriate to have a look at these details again, toward the back end of the season when the data set was larger.

From GW9 the 3 main points I picked out were:

i) Saturday, on average, means roughly a coin flip between a home win and any other result…ii) Sunday was the next best chance to guess a result: a draw is on the cards.iii) The swap in fortunes: A Saturday game means a home win is most likely. A non-Saturday, well the stars have re-aligned and now a home win is least likely!


Come Game week 37 and the surprising anomalies seem to have smoothed out. Saturday is still the best day to guess a result, Home win please. But the 'balance of expectation ' a Home Win, followed by  Draw and then Away win seems in line with 'expectation' across all days, (i.e. you would always expect a home win would to have great probability than and Draw followed lastly by an away Win. The large data set seems to make logical sense.) There is also a clear indication that there is greater chance of seeing a draw on a non-Saturday although this has come down from approx 20% to around 10% from GW9 to GW37, at least across all categories a chance of a draws come down (infer the season got better for viewing, not as many draws.)


Right so where do we go from here? I intend to go back at least one year & see if last season results were roughly similar.... I'm think it might be worth also collating teams in to Homogeneous Groups, i.e. Champions League Teams, Relegation Battler's etc and see if there are any nice grouping trends. I'm also beginning to wonder if there is a strategy for betting here (& of course a Fantasy angle, not that I'm a gambler,) but if it turn outs that the two season are similar in game day trends, then at points in the season such as GW9 we could hedge our selves predicting where the results will shift too.

Looking at the above table, the most obvious thing to do would be to start betting on Home Win's on a Sundays*, or structuring fantasy sides to have more player from team that play at Home on Sundays over the next few week. I need to give further consideration to the Balance of 'Expectation' assumption (as mentioned above) by checking this holds true for past seasons (watch this space.)


*Granted I have worked back word's here, i.e. GW37 data is not know before GW9 but the prior seasons data will help, hopefully there may be some consistency. 

@Mattistician

Shots on Target to Goal Ratio

Football Analytics.

Shots on Target to Goal Ratio 

After a couple of hours of looking through the mounds of data from the 2011/2012 season I decided to sneak a peak at the average goals to shot ratio, here's the low down of what the season told us.

Per Player per Game, 3.44 Shots on Target resulted in a goal. 

Per Player per Game (Considering players with 10 Goals & over only,) 2.64 Shots on Target resulted in a goal. 

The latter is probably the better stat from a fantasy football perspective, after all Goalkeepers and Defender's don't shoot too often.

The data was taken from the MCFC 'Premier league 2011-12 Match By Match' Data File, which shows that there were 3,522 Total Shots on target and 1,025 Total Goals (1,025 goals is a tad short of the actual goal scored for the 2011/12 season, although I don't think MCFC included Own Goals in the per player stats.) Overall according to their data, there were 10,369 Appearances by Premiership Players, this seems to make sense, 14 Player (11 & subs) * 38 Games in Season * 20 Teams = 10,640 Appearances. So I'm pretty happy these Stats are sound. 

Out of the 10,369 appearances, only 2,535 players actually bothered to shoot in a game, that's less than a quarter of all the players for the whole season (24.45%) I was surprised by this, I would have thought more players would like to take home a bonus in there weekly wage packet.

Looking at the data, the Number of Shots on Target seems close to the Normal Distribution, however goals clearly are not from a normal distribution, well least not for the 2011/12 seasons data.

From the Players who scored more than 10 goals in the season we begin to see a clear difference in the way they got there goals. As if we didn't know already RVP is truly an immense player, he was nearly top of every aspect considered as can be seen in the table below. Another player that sticks out at the other end is Newcastle's Cisse for the opposite reasons. Cisse who scored the lowest in all categories expect for Heading, where he was also one of the lowest players. The stats show Cisse definitely clinical, across all the metric's he lags behind all of the other players, he has to make the most of all the chances he gets.

Whilst on the topic of biting, or well maybe not.... Suarez is well know for all his Shots off Target, however RVP and Rooney weren't far behind his shots off total though, it's just they both have a lot more Shots on Target than Suarez so don't get tarred with the same brush.

Suarez, Dzeko, Ba, Dempsey, Aguero, Rooney & Van Persie are the top performers. This can be seen by the Total Column = Shots on Target inc Goals + Shots off Target inc Woodwork + Blocked Shots +Head Shots on Target +Headed Shots off Target. If a manager wanted to build a team around a player then this could be a pretty good Stat to investigate, (perhaps I should invest in Opta and see if I can pick any gem's out from the rest of the leagues.) Sunderland's Fletcher (Ex-Wolves,) Stand's out for his heading ability which has to be reconised, or it may imply that Wolves were playing lot of crosses and long balls that season, either way he's good with his head.

At the bottom of the above table, I included three players that have done well this year, Bale, Walcott & Mata, none of them scored more than 10 goals in 2011/12 hence didn't make the first table. Looking carefully, Bale stats from last year indicated he was a Top Performer without the goals to boot, Walcott and Mata however are average at best. The Illustrations below empirically show the players performance, and if MCFC had made the data avaliable earlier, perhaps I would have predicted Bale would be as good as he is today.

@Mattistician

GW36: Punt

GW36: A Double Punt - Tevez & Lampard

Thoughts from the Gut! 

With the Double Game Week looming, most people will be prepping there team with a player or two from the six teams that play twice in GW36, who could argue with that logic. Chelsea were bitten twice last week by the hungry Suarez, a last minute header earned most of us 11 points, with a suspension looming, most now have £11m to spend on a replacement. After a good few hours of looking through strikers, no-one seems to be great form. With Giroud’s slip and suspension and 40% of us already owning RVP, it makes sense to look through the DGW player options.

Man City,

The next 5 games are mouth-watering, WEST HAM (H), Swansea (a), WEST BROM (H), Reading (a), NORWICH (H) nearly everyone will want taster! Kun Aguero hasn't played 90 minutes since February and seems to be a tad overpriced at 10.9m, although I do fancy him to get a few goals before the seasons over. Tevez is £2.1m cheaper and that could be where the wise money lays; the cash difference could help in drag another DGWer into your team. We've all suffered from the Man City rotation nightmare, Tevez could even find himself warming the bench come the weekend. So making sure your bench has players who start looks like a smart move. With the money saved, regulars such as Kompany or Hart could be snapped up, after all there’s a potential of 20-30 points in clean sheets alone here.

Tottenham

Wigan (a), SOUTHAMPTON (H), Chelsea (A), Stoke (a), SUNDERLAND (H) every manager will want to sink their teeth into Tottenham’s Fixtures. Bale back and fit is just too obvious, with Defo and Adebayor battling it out for the top spot a move for regular starter could be key, Vertonghen and Walker will be encouraged to get forward to help AVB keep his job, the fixtures favour a top 4 finish. AVB’s return to bridge on the 8^th of May will be a cracker.  

Wigan

TOTTENHAM (H), West Brom (a), Swansea (H) Arsenal (A), ASTON VILLA (H)

Chewing the fat over Wigan's Fixtures and League Position, they have only scored 36 goals this season and currently seem to be the only team with any hope of getting out of the relegation zone. With only Kone in double figures, they have not have not generated great fantasy points so far. Wigan’s unpredictability made me look at Boyce and Maloney, but even if Martinez’s magic works and they manage to stay up again, I don’t think they are going to generate great fantasy points; I will be staying away from the Latics. Aston Villa on the last day of the season could be a premiership knockout decider.

West Brom

Southampton (a), WIGAN (H), Man City (a) Norwich (a), MAN UTD (H)

Quite simply WBA have Fixtures to Munch on, speckled with Fixtures to avoid. Clake has proven to love his subs this season, Long or Lukaku could have been viable options as a third striker, but the constant cycling of players has meant that no forward at the Baggie is close to 2000 minutes on the pitch, McAuley and Morrison are viable choices, but I will be staying away from West Brom, the two Manchester games mean there are better options elsewhere.

Chelsea

SWANSEA (H), Man Utd (a), TOTTENHAM (a), Aston Villa (a), EVERTON (H)

Bentiez will be chomping at the bit, a last minute goal from Suarez means Chelsea can’t afford to drop any points if they are to make the Champions League, with Ba and Zorres up top miss firing, Hazard or Mata might be the best bet, not to forget Luiz is always good for a free kick and a clean sheet. Surprising even myself, I’m going to pick out Lampard, I know he didn’t play against Liverpool, but I think he will be back with a bang over the next few weeks, to break the Chelsea top goal scoring record and earn himself the offer of a new contract, even if in the ends he opts to try fill Beckham's shoes at LA. 

Swansea

Chelsea (a) MAN CITY (H), Wigan (A) Man Utd (a), FULHAM (h)

Whilst the swans fixtures are the worst of the six, I fancy a nibble. I can’t see the Laudrups men going 5 games without a goal, I’ve stuck with Michu all year (a snip a £6.5m,) he will be involved in whatever action goes on for the swans. I also have Ben Davies and he’s proved to be a bargain I don’t see any reason to sell before the DGW. 

Overall thanks to Suarez’s appetite, I get the chance to pick out a couple of punts. A couple of punts definitely has ring to it & that seems to fits nicely with Tevez & Lampard. I would have gone with Zabaleta but with Mica Richard's return from injury I think the Argentinian will find him self benched. Let's just hope the DGW and End of season is a Jaw dropper. Okay the last pun was a bit(e) too much.

GW9: No football like Saturday at 3pm - Football Analytics

No football like Saturday at 3pm – A Matt-istician Update

Hello football fans, now I know this article goes against the requests for further positional analysis, but I just couldn’t resist. So hold back your boos, after all I’m not wearing black and I haven’t pointed to the spot. Further positions will be available in the near future!

Now if your love for football comes from the same place as mine, you most probably have someone else to blame, I can most definitely lay this blame at my Dad’s door! He like every man from his generation loves flares, wore shorts 3 times too small for him, but most of all he likes football to be played at 3pm on a Saturday (hear, hear). Hence this article is dedicated to picking apart ‘Game Day Trends.’ That being Saturday’s, Sunday’s, midweek gaming, and the mighty Champions League.

Premier League

After forcing my missus to watch Gameweek 8′s Super Sunday 1-a-piece bore draws, between both Sunderland and Newcastle, and QPR and Everton, I was thoroughly depressed at the thought of grueling ‘X-factor / Strictly’ combo sessions to be thrust upon me in future weeks, what a waste! But did this horrible lesson teach me anything? Reluctantly I started to mull over the relating stats and surprisingly the numbers started to speak to me.

Game Days	Games	Home Wins	Away Wins	Draws
All Games	79	44.3%	24.1%	31.6%
Saturday’s	56	51.8%	21.4%	26.8%
Non-Saturday’s	23	26.1%	30.4%	43.5%
Sunday’s	16	21.1%	31.6%	47.4%

Firstly if we consider all games so far this season, most people would guess that on average more teams win at home than away and there should also probably be more draws than away wins.

From the first 8 Gameweeks they’d be spot on, and these stats seem reasonable to common logic. If we break this down, the stats start to skew: Saturday games have resulted in almost a coin flip for home wins (just over 50% for home wins vs an away win or a draw). Although an away win drops from roughly just under 1 in 4, to just over 1 in 5 games depending if played on a Saturday or not.

So from this we see that if teams play on a Saturday there is more likelihood of a home win, in fact by over 50%! But why… as I sit here scratching my head, I think this must be down to more fans attending Saturday games, getting behind their teams and driving them on to a win, as well as the general crowd’s sway and its effects on the ref.

Moving away from the classic Saturday game, and considering non-Saturday games, the stats go haywire. A home win’s likelihood is almost halved and is also the least expected of the three out comes! Draws jump up by just under 17% and an away win looks like a good punt at 30.4%. Justification? I think this may be as simple as the opposing reasons provided previously, not as many home fans attending, teams going flat without the support. When the opposition scores perhaps the opposite effect takes place, where away players are driven on by the smaller support going crazy, without the home fans to drown them out.

Looking at the fixtures, Sky Sports and ESPN might also have to take some blame here. The games that are not on Saturday are usually between competitive teams, to make good television. Let’s have a look at the next few non-Saturday games….

Non Saturday Dates	Home	Away
GW 09 – 28 Oct 13:30	Everton	Liverpool
GW 09 – 28 Oct 15:00	Newcastle	West Brom
GW 09 – 28 Oct 15:00	Southampton	Tottenham
GW 09 – 28 Oct 16:00	Chelsea	Man Utd
GW 10 – 04 Nov 13:30	QPR	Reading
GW 10 – 04 Nov 16:00	Liverpool	Newcastle
GW 10 – 05 Nov 20:00	West Brom	Southampton
GW 11 – 11 Nov 13:30	Man City	Tottenham
GW 11 – 11 Nov 15:00	Newcastle	West Ham
GW 11 – 11 Nov 16:00	Chelsea	Liverpool
GW 12 – 18 Nov 16:00	Fulham	Sunderland
GW 12 – 19 Nov 20:00	West Ham	Stoke

All of these games are ideal for TV; they all seem to be at least competitive games, from teams that are closely matched (except for the Southampton-Spurs game. Good luck Saints fans! I have Lambert in my team so haters beware). Hence it is no wonder a draw is worth 43.5% of the previous Gameweek data.

If we look at just Sunday games, we see a massive 47.4% for draws (followed by an increase in away wins). This win percentage is the most likely of outcomes for all the away ‘Game Day’ types. Looking at this weekend’s games, the stats show there’s about 50% chance of at least 2 draws, this will no doubt be interesting to compare over the next 8 games: Gameweek 9 – 16.

So, from this what are the Matt-istician Game Day Stats to remember?

i) Saturday, on average, means roughly a coin flip between a home win and any other result…

ii) Sunday is your next best chance to guess a result: a draw is on the cards…

iii) The swap in fortunes: A Saturday game means a home win is most likely. A non-Saturday, well the stars have re-aligned and now a home win is least likely!

Champions League

We’re having a laugh… but what effect does the Champions League have on the 4 English teams. I do not doubt that most readers will have at least one, if not two players from the Champions League in their fantasy side.

With the endless perils of squad rotations, especially apparent in the Manchester clubs, the Champions League forces managers to look at their players and question their ability or stamina to play 3 games in week or so.

If your players manage to avoid the Champions League axe and actually start after the Champions League games, then you’re doing well, reward yourself with a 250 calorie ‘Apple Pie’ next time you visit McDonald’s, you’ve earned it.

Right so where were we…. after the Champions League games this season so far, all teams have won their next games except for Man City vs Arsenal which ended up a draw. So instead of looking at results, which may prove less than useful, let’s look at the players who were involved.

It’s a little too early in the season to draw any solid conclusions, but we can begin to see which players the Champions League teams like to use either side of their Champions League exploits, hence this is good for fantasy points and gives us all an insight into manager’s minds.

Below I have only considered games around the Champions League i.e. Champions League games excluded. It’s Monday the 22nd as I write this, prior to this week’s Champions League games.

There have only been 2 Champions League games for each team hence there is a maximum of 4 games in which players could have appeared in either side of a Champions League game.

I have listed specifically the players that appeared in the Premier League either side of the Champions League games (the before and after games). I have also noted the number of Champions League games inside the brackets for reference. If a player is on the pitch for less than 25 minutes, i.e. a sub appearance, I have not counted this as an appearance, hence the super-sub Dzeko misses out.

Man City
Played in 4 Premier League games – Hart (2), Yaya Toure (2), Silva (2), Tevez (1)

Played in 3 Premier League games – Kompany (2), Lescott (0), Clichy (2), Garcia (2), Barry (1), Aguero (1), Balotelli (0)

Man United
Played in 4 Premier League games – Rafael (2), Van Persie (2), Ferdinand (1), Carrick (1)

Played in 3 Premier League games – Evans (2), Evra (2), Lindegaard (0)

Arsenal
Played in 4 Premier League games – Jenkinson (2), Gibbs (2), Arteta (2), Podolski (2), Cazorla (2), Gervinho (2)

Played in 3 Premier League games – Vermaelen (2), Mertesacker (2), Mannone (2), Ramsey (1)

Chelsea
Played in 4 Premier League games – Cech (2), Ivanovic (2), Luiz (2), Cole (1), Hazard (1)

Played in 3 Premier League games – Terry (1), Mata (1), Oscar (1)

So what does this show? Although this is early on in the Champions League, we are looking at players that managers are willing to play 3 times in one week (or only has one player for a particular position due to other injuries).

So either these players are dead certs i.e. Arteta, Yaya Toure, Van Persie, Cech etc, or the manager recognises them as players who are not quite certs but are good for Premier League games when the squad is tired i.e. Balotelli, Lescott, Oscar, Lindegaard, Ramsey, Barry. It’s also useful to consider who’s not there, think about your own team, and your Champions League FPL players should be listed within the above or it may be time to get your thinking cap on and find out why they are not.

Looking at the appearances, Arsenal have the most number of players who appear in 4 games, hence Wenger is being more consistent with his players compared to the other 3 managers. Man United have the least consistency overall, indicating Fergy is dabbling with the most rotation.

Obviously the key here is to stick to players that will get good Premier League game time rather than the Champions League time. However this data shows players who the managers think are key to the team, or literally there isn’t anyone else in that position to play, i.e. Rafael and Jenkinson.

Now, Tevez, Luiz and Silva are the biggest surprises for me. Although there is true fantasy pedigree within this group, lately player rotation is a topic that has been associated with each of these. The stats show that the managers don’t have a problem playing these guys 3 times in a week, so in theory these players should get good Premier League game time in games to come.

I haven’t picked out any Man United or Gooner players here, as even though the stats are on Carrick, Ferdinand, Ramsey and Mertesacker’s side, my better logic tells me to question this! Players like Gervinho, Hazard and Van Persie etc seem more than obvious picks to me, so I haven’t bothered to go into detail here. But overall I will be definitely keeping an eye on Tevez, Luiz and Silva over the next few weeks.

GW7: Jumpers for Goal Posts

GW7: Jumpers for Goal Posts – A Matt-istician Update

In the back corner of FPL Dugout, with one too many members of staff sent off to Azkaban Prison, the Mathematician finally gets to deputise and write an article. The stats in this article were taken prior to Gameweek 7, but I can hear you cry “What are the stats telling us?”. Music to my ears, and without further ado, let’s start with my first positional analysis.

The Lads between the sticks

The top 5 selected goalkeepers combined percentage selection totals a whopping 74.6% (although the FPL total for all goalies doesn’t quite sum to 100%, not helpful).

This indicates a large percentage of people have chosen between the following 5 shot stoppers: 1. Vorm (19.6%; 5.6m), 2. Jaaskelainen (16.5%; 4.8m), 3. Cech (14.5%; 6.6m), 4. Ruddy (12.7%; 4.5m) and in 5th place Foster (11.5%; 5.2m). This shows that people in general have given up on the top 6 teams goalkeepers (other than Cech). Are there any stats out there that can help us analyse this?

Last season Manchester United had the most clean sheets with a total of 20 which was over 50% of their games. This season with rotations between de Gea and Lindegaard and what seems to be an even more injury (and leaky) prone defence, United have already let in 9 goals this season, this being 27% of their quota from last season’s tally of 33 conceded in total.

Cech and Chelsea find themselves sitting pretty top of the Premier League with only 3 goals conceded, so Cech and Co seem to have their defensive qualities back in order. Last season saw Chelsea finish a surprising 12th for least goals conceded overall, only managing clean sheets in 26.3% of their total league games, shocking considering Chelsea’s defensive mentality throughout the Champions League.

Cech has 4 clean sheets (of the politically correct variety) to his name this season, the most of any keeper, already nearly half of last season’s total of 10. Clearly Di Matteo has been working on their defensive attributes and the below graph shows Cech is in the top 2 keepers for ‘Value for £m-honey$’.

View a larger version of this graph

What does this graph show us? I hear you say…. Firstly as titled we are looking at a ‘Value vs. Points’ Graph – we have points on the left hand side and value on the right. All the keepers that have had any playing time are listed across the horizontal axis (ordered by price, value left to right).

The Blue Bar Charts show the keepers price increasing right to left, the Red Line Chart shows the keepers and the points they have collected so far.

Lastly the Green Line is the Line of Best Fit for points. By this we can see that players who are above the Green Line are ‘Value for Money’ (V4M), i.e. Howard, Schwarzer, etc. Players below the line just aren’t V4M, i.e. Reina, Hart, etc. Players close to the line can be seen as par value, i.e. just below the line Guzan (although in general there are more outfield players which are par value).

From this Graph, Jaaskelainen is the major V4M keeper, although West Ham’s next run of games are not overly easy: SOT wig MNC new STK, but as he’s still a cheap option and with a good pairing for rotation, the stats show he has paid dividends so far (presuming most readers subscribe to a goalkeeper rotation policy or theory).

As anyone experienced with looking at stats will know, you have to look at the detail and question what the stats aren’t telling you.

Here the ‘GW7 – GK Trend, Value vs. Points’ Graph doesn’t take into account minutes played, and perhaps there is some underlying information here that can justify the trend and help us identify / confirm other players that are V4M or help identify potential up and coming options.

View a larger version of this graph

Looking at this second graph ‘Minutes Played vs. Points’, we can clearly see that the goalkeepers who have played the most minutes have in general ended up with the most points and vice versa (Rocket Science – obviously not).

Please note goalies have been listed left to right by number of minutes played, again Cech and Jaaskelainen majorly peak as the best two butterfingers. At the lower end of the scale (right hand side), I find the peaks and troughs most interesting.

Looking at this I would offer two pieces of advice i) avoid the troughs, and ii) there’s room for speculation on the peaks, by this Begovic, Mignolet, Mannone, Davis, Cesar and Szczesny are peaking and indicate they are doing better than their counter-parting troughs – Lindegaard, Guzan, Krul, Green, de Gea, Given, Federici and Gazzaniga. I would avoid these like the plague for now. There are also a host of good middle-ranging keepers like Foster and Al-Hasbi, so there is spectrum on offer to choose from.

Overall who sticks out considering everything?

1. Obviously – Cech and Jaaskelainen, no reiteration needed.
2. Stats indicate the following goalies are over-rated: Hart, Reina, Friedel, etc, both Man United keepers and the host of troughs.
3. Cheap other options – Ruddy / Mignolet (plus due a double gameweek at some point).
4. Matt-istician’s unobvious picks? – Begovic and Cesar.

Late joiner, Soares Julio Cesar is peaking on the minutes chart in a poor form team, and whilst he seems like an expensive ‘no no’, with some easy fixtures round the corner / defensive injury’s due to come back, and Hughes at the helm, the stats indicate Cesar could pick up some points. Next 5 are EVE, ars RDG stk SOT.

With Stoke’s hard start to the season so far, Begovic looks like he could prove to be a true bargain with future easy games. The next two games for Stoke are hard but after that things start to get better. Next 5 are mnu, SND, nor, QPR, whu.

Wrapping up, it would be good to get readers feedback on their current goalkeeper selection policy. Let me know if you have any other mathematical or stat based questions, as it will be good to understand if you guys are grappling with anything out there.