Thursday, 2 June 2016

Mark Lawrenson - Smarter than he looks

For a few seasons now, I’ve been amazed that despite the fact that everyone considers his predictions something of a joke, if you’d actually bet on them, Mark Lawrenson’s predictions would be profitable.

This started back in 2012/13 when I was putting together my first ‘Football and Data’ event and wanted to see how accurate a supposed ‘expert’ was and I was staggered to find that Lawro’s predictions would make money.

Considering this season’s predictions would have placed Leicester in 12th (with relegated Newcastle predicted to finish 11th) surely this season would finally show him up as a know-nothing?
Lawro's predictions over the season (via BBC Sport)
Well, actually 2015/16 would have been another profitable season if you’d followed Lawro’s tips. With a £10 stake on each of his 376 predictions, (4 matches were rearranged and not predicted) then you would have made £271 profit.

Odds were taking from Pinnacle's quoted market from the fantastic Football-Data site and Lawro's predictions from MyFootballFacts.  Pinnacle's margin was 2% so picking at random and placing £10 on each of 376 selections an average punter would expect to lose around £75.

As seen in previous seasons this profit was not made by daring longshots but by being willing to sit on the fence and profit from draws which are often overpriced, in relation to the ‘true’ probability more money goes on a win as the typical punter fancies a positive result.

Sometimes a question will be phrased (either to someone or even within their own thoughts as) ‘Who do you think will win on Saturday’ already suggesting you ignore the draw but even if it’s considered as ‘What do you think the result will be’ then sitting on the fence is considered a pretty boring option even if it makes good betting sense.
Yet again it's the draws that bring in the profit for Lawro
Fans of a lot of teams (especially West Ham this season but also Swansea most seasons), feel that Lawro has a grudge and in West Ham's case it does seem as if Lawro had a blind spot to their form, as a Swans fan it's a bit weird to see Lawro predicting us to finish 18th but still making an £80 profit on matches involving us.
Profit/Loss by Team: Total of  this would double count overall profit as counting each match twice
Where some may change their opinion as more games are played, you could probably predict Lawro's predictions would a fair degree of accuracy at the start of the season. It's not the optimal strategy but it does make some sense, rather than chasing the noise and being overly influenced by recent results, Lawro plots a course and stays with it no matter what. Trying to figure when (if ever) Chelsea would come good or Leicester/West Ham finally drop off would have been a hard task.

As he does every season, Lawro keeps his scores simple with 99 of the 101 predicted draws being predicted as 1-1 (which actually is the most frequent occurring draw)
Lawro's Score Predictions - the three in red (2-0,2-1,1-1) being his three main tips of choice
Lawro's advantage comes from him not being afraid of 'being boring', where the bookies love the 'Have a bang on that' kind of punter who blows £10-£20 on an accumulator (which compounds the issue of a bookies margin), profit is usually derived from a large number of small gains (which could arguably just as easily apply to football itself).

Another example of 'Boring Lawro' is his response when asked about his apparent success:

“If I could successfully predict all the results, do you really think I'd be here? I'd be on a yacht in the Caribbean…“

“If I could really seriously predict football results, I would be sitting on my ocean-going yacht, which would be moored off the Bahamas.”

Monday, 11 April 2016

This model is wrong but it may be useful

Expected Goals (xG/Goal Likelihood) models are increasingly common and try to add a bit more understanding to what’s going on in a game to try and get beyond just the scoreline or the top level shot stats.

None of this is to knock a lot of good work that’s being done in this field (there’s a list of useful links at the end of this), but anyone can build a model.  Technically you could call what Lawro does in his predictions a model even if it’s probably a fairly simple ‘mental flowchart’ to pop out the score at the other end.

The aim of this piece is give a very basic introduction to looking at Expected Goals.  It’s not even that big a leap from when managers talk about restricting the other teams ‘big chances’.

All figures that follow come using Opta data but will have been butchered and filleted by my own fair hands so any errors/omissions etc., will be mine not theirs. 

Analysis excludes Penalties and Free Kicks which are special cases and will be dealt with separately some other time although there’s plenty on free kicks from across the Big 5 Leagues on my blog.

If we start with a very simple model where we assume all shots are created equal we get the following for Premier League data:

So if we apply the 2010-2014 conversion rate to 2014/15 activity (excluding set piece/own goals) we get the following:

This, as I’ve mentioned before is a model, just because you call something a model doesn’t necessarily make it any good.

As always, the key thing with looking at any numbers is the inferences, you could look at the numbers and say:
  • Chelsea were lucky
  • Chelsea were more skilful in converting chances
  • Chelsea created better than average chances
  • The model's crap
Many the time I’ve been in a meeting post a marketing campaign and the conversation has gone something like this:

Boss: We forecast sales of £100k but only sold £73k, why was that?
Marketing Wonk 1: The weather hasn’t been very good the last couple of weeks?
Marketing Wonk 2: People’s budgets are stretched post-Christmas?
Me (in my head, as I'm a coward): It’s because the £100k figure was a nice round number you pulled out of thin air and has no real basis in fact

It’s a natural trait to try and rationalise any figures you see after the event but generally you should always be wary of any explanation you’re given and should try to at least run a quick sense test on things.

A nice example of this is the Baresi/Maldini stat of 23 goals conceded in 196 games that was doing the rounds last year even though it’s completely wrong (there’s a good piece here on it).

Going back to our crude expected goals model, obviously not all shots are equal, if we do a basic inside/outside box split we get the following:

Shots from inside the box convert at 4-5 times those outside the box, so this split helps differentiate between those who are shot heavy outside the box (e.g., QPR with 47% of shots coming from outside the box) with those doing their work in better positions (Man C with only 27% of their shots from outside the box).

At Leicester’s ‘Tactical Insights’ performance analysis event a few weeks ago, when asked about stats, Roy Hodgson said if Shots suddenly were the key thing he’d get people to shoot from half way, he was being a bit flippant as even a 3-year-old knows a shot at an open goal from 3 yards has a better chance of being a goal than a punt from 40 yards.

What you’re left with then is a balance between more and more detail (angle of shot, headers, defensive pressure) better explaining a team/player’s activity against over-complicating things and creating a ‘black box’ approach which spits out a final number that may be harder to explain to players/management.

Using data is better than not using data, using detailed data (e.g., Shot Location) is better still and even better again is combining it with video.

In the chance below, Rooney has a simple tap-in so at the point the shot is taken the likelihood of a goal is close to 1, if you treated that shot the same as all others from that location then Rooney (and Man Utd) would then be outperforming xG, but unless you had the video you wouldn’t know if this was luck, good finishing, good positioning by the forward or good chance creation.  

Beyond the shot itself, subsequent versions would have a more fluent xG at any given time (there’s a number of people building non-shots models) which basically say ‘I have the ball in this position, what’s the likelihood I score before giving away possession’.   In the case of the Rooney goal you could take it several steps back with the likelihood increasing as each part of the move is successfully completed.

The example below from Basketball gives a good example of this in practice:
As does this video from Prozone’s Paul Power (Expected Goals example is about 8 minutes in but the whole video is well worth a watch).
Performance.LAB Innovation Seminar: Game Intelligence - Paul Power from Prozone Sports on Vimeo.

The concept that some areas are better to shoot from that others isn’t a hugely difficult one to grasp although if the example from American Football is to go by from earlier this year, maybe teams aren’t fully optimising their activity.
Single Match Expected Goals – Will Gurpinar-Morgan
Expected Goals – Martin Eastwood

Wednesday, 18 November 2015

Garry Monk: From cover star to bottom of the bill (and back?)

When you're hot, you're hot. When you're not, it suddenly looks like everyone's out to get you...

With the Swans finishing 8th in 2014/15 there was even talk about Monk being a possible candidate for next England manager with Hodgson presumably departing after England underwhelm at Euro 2016.

As the highest finishing English manager last season, this wasn't as fanciful as you'd think, with the other front runner Alan Pardew liable to head butt someone or call them an 'Old C*%t'
Garry the Gaffer - LMAs Cover Star at the start of the season
Early on in the season with 8 points from the first 4 games and sitting in the Champions League places, everything was looking pretty rosy and mid-Sep I got an email from Leaders in Sport talking about their conference in November with Monk as one of the main attractions:
Leaders email - other speakers included World Cup winner Fabio Cannavaro and a range of top managers from sport across the world (the unique part above misses out Tate/Britton but still an incredible acheivement)
As much as I'd have love to have gone to hear from Monk, Leaders is a bit out of my price range to go just as a sports fan rather than as someone actively working in the industry.  I did however keep an eye out on Social Media to see what came out of the event.

I was a bit surprised though to see Monk ending up at the end of the second day in possibly the graveyard slot alongside Per Mertesacker:
I was sure Monk had a higher billing in the original schedule and looking back at the original event leaflet that seemed to be the case:

As I wasn't there, I've no idea why the change is made and there may be a perfectly valid reason, but have the horrible feeling that the organisers felt it might be a bit awkward to talk about maintaining success with a speaker who (at time of writing - 18th Nov) is 1/4 to be next Premier League manager to leave.

Sack Race odds however should be taken with a pinch of salt, where on a normal market (e.g., Win/Draw/Lose for a match) a bookie might price it so the total odds are around 103% giving them a theoretical small margin, for sack race the figure is 164% (Sky Bet) as a large part of it is rumour and also to protect against getting scalped by someone 'in the know'

This brings me to the question, 'Should Garry Monk's job be at risk?' For me the answer is absolutely no, the last couple of months haven't been pretty and if I had to describe the Norwich performance in a word it would be 'Tetchy'.

If you look at 4 of the 5 league defeats this season the common theme has been a lack of cutting edge up front and maybe a lapse of concentration or sloppiness at the back resulting in a defeat.

  • The goal conceded away to Watford was route one, flick on and a finish.
  • Stoke at home was an Omnishambles of multiple mistakes resulting in their 4th minute penalty.
  • Arsenal's first was where Ash totally misjudged the corner after Swansea were arguably the better team.
  • Norwich away was a lack of concentration at a corner again.  The defeat away to Southampton probably the only result which wasn't debatable.

Any side can 'If only' pretty much any match but it's small margins between success and failure.

There's no point saying 'Why aren't we like West Ham or Leicester' as any season there's likely to be 1 or 2 teams who start the season strong but come the end of the season they'll probably both finish no better than 8th (although that's not to be sniffed at).

Likening the current situation to that when Laudrup was fired is to me a false comparison, despite winning the league cup and the Europa League win in Valencia, things seemed on a downward spiral since the club went on a mid season break shortly before the league cup final, returning to get battered 5-0 by Liverpool the week before Wembley.

As happened with Laudrup, we're seeing with Monk now, stories of dressing room disquiet emerge, rumours of Pep off to Brentford, things get exaggerated and 'misquoted' as 'club in crisis' gets more clicks than 'club in sticky patch that they'll probably turn around'

The dressing room have let it be known that they are irritated their manager takes too much credit when Swansea win. When they lose, the complaint is that he is too quick to move the blame on to the players. - Neil Ashton Daily Mail

The big difference for me between the two situations though is the length of time the bad run has gone on for, the chart below shows Swansea's rolling 38 game total:

38 Game Average - Laudrup's peak happening just prior to that Liverpool game (and also still partly buoyed by Swansea's end of season form under Rodgers).
In the long run then, things look decent but ultimately 'in the long run we are all dead' and the short term is what gets the attention.  Much has been made of Swansea being 'Not just another club', but with the new TV deal on the horizon, it's possible the owners will get twitchy although I hope that isn't the case.

There's been talk (how accurate I don't know) about the hierarchy looking for Monk to bring in a senior hand (possibly a director of football), whatever you think about that kind of structure at a club I find it hard to see how it's anything other than destabilising for Monk in the current climate.

As mentioned in a piece by the Swans Trust today, there's been a huge leap forward in facilities at the club in the last couple of years but that maybe there isn't much budget to buy our way out of any slump, 'too good to go down' is a dangerous game to play but unless there has been some almighty disruption between manager and team I so no reason to feel our Premier League place is truly at risk.

In terms of financials, I'd disagree with the need to expand from an income point of view, 6,000 extra seats would generate £2-4m a year, depending on number that were sold as season tickets or to Kids/Senior Citizens etc., (and how many games sold out).  Although that's not chicken feed, a mid-table club in 2016/17 will probably be taking in £125m in TV money so while 'every little helps', from a financial perspective any change in gate income will not produce a massive change, especially short term as it would require capital investment to do the expansion.  Doing it to increase the fanbase and longer term drop prices would be great but that's a different issue.

Although things have gone quiet on the takeover front, selling once Premier League status for 2016/17 was confirmed could be the optimal time.  It was interesting to see this tweet by sports lawyer Jake Cohen on the possible sale of Crystal Palace:
One of the things that makes Swansea stand out as a place is also something which stands in its way in terms of growth of the club and that's its location.  Having pretty much nobody to the North, South or West and your historic rivals to the East means that you can't bring in people from a wider catchment area like Palace could do, although as mentioned above with Swansea and the gate money, I don't particularly see the burning need to drive up attendances or create a bigger ground when so much of the £ is in the TV deal.

If you're a Champions League level club and can pull in corporate hospitality and high end spenders then match day is important, otherwise you're effectively a media company that has a few people come along to watch, however dispiriting that may sound.

Friday, 30 October 2015

Free Kicks in the Big 5 leagues analysed

Last February I presented at the Opta Forum on efficacy of crossing/shooting from Free Kicks, I've been planning to do an update at some point but Gylfi Sigurdsson's free kick vs. Villa was a good reminder to pull my finger out and get on with it.
When they go in like that one they make the defensive side look helpless and a goal 'unstoppable' - Guzan's positioning makes him irrelevant if the ball goes successfully over the wall. As great a free kick as it is, it's nowhere near the top corner when it goes in as it doesn't need to be that accurate to be a goal.

Guzan arguably still had in the back of his mind Sigurdsson's free kick against him last season when he went the other way:
A few weeks ago the boot was on the other foot as Christian Eriksen scored twice from free kicks v Swansea, both of which he put to Fabianski's side:
The question is, as a keeper, do you 'trust' your wall and leave it to whether the free kick taker get the ball over the wall, or do you try and second guess the taker and risk leaving a massive gap for them.

When facing elite players, I argued in my Opta Forum presentation that maybe a different strategy was needed and highlighted this example from a Nice B game:

Sigurdsson and Eriksen are arguably two of the best free kick takers in the league but sample sizes are generally too small to get a true conversion rate for an individual, but we do have enough data to start to make some assumptions is general.

I don't have the same quality of data as provided by Opta for my forum presentation (so any errors will be mine rather than Opta data issues) but I have put together an analysis of the outcomes of over 9,000 direct shots from free kicks taken in the big 5 Euro leagues (last 5 full seasons for Premier League, last 3 full seasons for La Liga/Serie A/Bundesliga/Ligue 1).

At the top level, overall 5.8% of direct free kick shots end up in a goal but of course that varies by location:

As you'd expect, the closer and more central a free kick, the higher the conversion rate, when looking at distance from goal (in a straight line and ignoring any angle to goal), conversion drops sharply from 25 yards out:

The other point to make is the relatively high conversion (6.9%) for long distance wide of goal attempts, my guess is that a lot of these are from free kicks that are 'put in the mixer' and end up going straight in.  It's open play rather than a free kick, but this goal from Andros Townsend is a good example of an effort that probably wouldn't have been classed as a shot until it ended up going in.

Going back to the Nice B example, if you know on average someone will score from a certain position 10% of the time, then if their unusual defensive set up resulted in a lower conversion then obviously that'd be a better strategy.  The risk however is that even if you tested it in training over a large sample and were confident conversion dropped to 5%, if you did it in a Premier League match and it didn't work, you'd risk looking a fool (where you never look stupid if you do what everybody else does, regardless of if it's the best option or not).

One of the other areas I covered in the Opta Forum presentation was how different teams approached free kicks in attacking areas, some shot pretty much whenever possible and some only shot in the prime positions.

Looking at the highest/lowest free kick shot volumes, tho two Madrid teams come out at opposite ends of the spectrum, I would expect Real to have more possession than Atletico in general but there's no doubt that Atletico are passing up the lower chance shots in favour of crossing rather than shooting.
Top Teams by Season, Real Madrid with 3 of the top 20
Atletico with just 6 direct shots in La Liga last season, down from 11 the year before and 15 in 2012/13. Lens' shot volume will in part be due to them finishing bottom in Ligue 1 last season.
Atletico's strategy was noticeable in the 2013/14 analysis I used for the Opta presentation, but last season they took things even further with just 6 direct shots, all from fairly good areas:

Looking at the teams with the highest shot volumes, there were a lot of Serie A teams in the top 20, which is also seen when comparing the leagues as a whole:
You'd need overall free kick volumes to know if Serie A's extra shots were coming because they have more free kicks in general rather than just chose more to shoot rather than cross or retain possession.  Even with several years data, you're starting to get into the realms of small samples when considering conversion rate, there was no obvious reason i.e., location why the Bundesliga conversion was higher than the Premier League's
Bringing it all back to Gylfi again, Swansea host Arsenal again on Saturday and in the same fixture last season he managed this, which from 29 yards you'd expect an average player to convert around 4% of the time:

How much above average the likes of Sigurdsson and Eriksen are when it comes to free kicks is hard to quantify but feels like they take the art of free kick taking to a different level and that in some cases having a standard wall helps rather than hinders them.

Other Posts:Gomis: Lyon+Swansea Shot Locations
Value of Corners - Outcome of 25,000 Analysed

There's plenty of free kick compilations out there but as a Swans fan enjoyed this goal compilation of the U18s Kyle Copp who has some great ones already on his showreel:

Twitter: @we_r_pl
Stats: Created using Statszone , Whoscored  and Squawka

Monday, 17 August 2015

Swans 2 - Newcastle 0 Stats and Chalkboards

The dream scenario before any match is one where everything goes in and you have a crushing 5-0 victory, the more pragmatic version is more like the game on Saturday: Early goal, control the game, hopefully getting a second and strolling to a comfortable win (the extra man making it even easier).
Shot by Minute - For 2nd game this season, Swans dominated the shot count
2-0 down and a man down you can understand McClaren looking to keep things tight and hope for a miracle late on, but Newcastle didn't manage a single shot in the second half.

I was expecting this to be the lowest number of shots that Swansea have conceded in the Premier League, but there were two games (Dec 2012/Jan 2013) against Sunderland and Reading where they only had 3 shots (to make that stat even worse both those teams were at home to Swansea in those matches.

The pass volumes show how in control of possession Swansea were in the 2nd half bar a small period right at the end of the match:
Pass Volumes - some periods where Newcastle averaging less than a pass a minute
Ash also had the distinction of making over 100 passes, this is a rarity in any match but when it happens it's usually from someone like Alonso at Bayern (sometimes Leon too) where it's a defensive midfielder as a pivot between defence and midfield.  For it to be a centre-back with this number of passes shows the level of dominance.
Williams' pass map by half, his only misplaced one in the 2nd half coming from a trademark diagonal
Other Posts:Gomis: Lyon+Swansea Shot Locations
Value of Corners - Outcome of 25,000 Analysed

Twitter: @we_r_pl
Match Stats: Created using Statszone , Whoscored  and Squawka

Monday, 10 August 2015

Chelsea 2 - Swans 2 Stats and Chalkboards

"It's a marathon not a sprint" etc., but the point (and even more importantly the performance) is a big step towards continuing the momentum of last season and it's not often a team comes away from Stamford Bridge having had the better of the chances.

Although the Penalty incident resulted in a red card and the equaliser and maybe curbed Chelsea's threat from there in, the majority of the chances had come from Swansea before that anyway.
Shots by Minute - Swansea's 18 later got downgraded to 17 by Opta as they removed an attempt that came directly from a corner (which personally I'd still class as a shot)
The shot that was chalked off was the first of these two in the video below but arguably for me they we both actually shots rather than crosses
In 38 league games last season, Swansea only had 18+ shots in 4 games and 2 of these were against QPR so this (even if just 1 game) was quite a change from last season - even if 3 of the shots came in a matter of seconds as part of Ayew's goal.

It wasn't exactly cutting edge insight but on the morning of the game I'd tweeted about how many of Ayew's goals seemed to be tap ins or from rebounds or saves off the keeper.

There comes a point where it's a case of anticipation, judgement and general fitness to be able to be in the right place at the right time rather than being a 'lucky goalhanger' as probably 6 or 7 of his 10 league goals last season came from close in finishes without a keeper to beat.

One of the most noticeable aspects of the game was the number of times big gaps seemed to open up in the middle of the pitch (one such incident led to a drive upfield from Ki ending in his hamstring strain), possibly due to Matic being over-exposed in this area.  It was a surprise for Cork to be left out but this did result in Swansea's approach being 'attack is the best form of defence' which worked superbly.
Average Pass Volumes per minute - Averaged over 5 min periods. Chelsea with 2-3 periods of control in the first half but then only a brief spell towards end of 2nd half (obviously impacted in part by being down to 10 men).
The obvious other highlight was Montero's performance, in particular a 5-10 minute spell early on in the second half (although his first take-on of the match was actually after just 6 seconds which was a sign of things to come).

Montero's explosive pace and acceleration mean that injury risks are always there, one comparison (showing my age) could be with the 110m hurdler turned Rugby Union hurdler Nigel Walker who was on a different level speed wise to those around him but often gained niggling injuries which kept him out for periods at a time.

To save me time, if you're after some more chalkboards from the game, the FourFourTwo Statszone review of the match is pretty good this week looking at Montero/Shelvey among other areas.

Other Posts: Impact of True to 22Gomis: Lyon+Swansea Shot Locations
Value of Corners - Outcome of 25,000 Analysed

Twitter: @we_r_pl
Match Stats: Created using Statszone , Whoscored  and Squawka

Thursday, 9 July 2015

Swans: True to 22 - Some thoughts and figures

Back on Sunday I wrote about how '..The way clubs treat the Away Fans initiative is symbolic of what they think of their fans' so I was blown away yesterday when Swansea announced their True to 22 scheme.

I'd argued previously that although Swansea were one of the clubs putting the most effort into the Away Fans Initiative, I felt it was better spent on lower ticket prices rather than programmes, free shirts etc., I suggested that the £200k would best spent on those who may find it hardest to afford to go and without means testing everyone that'd probably be the Under 25s.

What Swansea have done has gone hugely beyond that, in their statement launching the scheme they estimate the saving to be £300k but personally I think it'll be closer to £500k.

The figures are a bit back of a fag packet but Swansea's figures estimate someone going to every away game will save £250, not everybody will end up buying tickets via the Jack Army subsidy but I think a huge amount will, especially for the bigger (and usually more expensive) games.

Last year's Chelsea allocation was 2,200, I've seen the odd post mentioning it may be 3k this year.  If we take the lower figure that could be £60k subsidy in one game with similar if not more for Man Utd Away.

Clubs are prohibited from charging more for away fans than the equivalent seats for home fans (Newcastle have got fined for this in the past) so teams can't just whack up the price knowing Swansea will pay the difference, but it may be that after finishing 8th and being in their 5th season in the Premier League there may be fewer Category C type fixtures such as the £25 v Arsenal.

Ultimately whether it's £300k or £500k it's still relatively small beer to the sums clubs are receiving from TV etc., but is a hugely symbolic step, time will tell if the Swans are swimming against the tide or are able to force the hand of a few other clubs as well.

For anyone who feels that home fans are being missed out, 19 games at £22 each=£418 and early bird prices for season tickets were from £419 so from that point it seems fair and that's before you've factored in travel costs etc.,

While you'd have too be mad to criticise the scheme there will be some consequences of it.  The system is being changed slightly so Jack Army members who are season ticket holders can still buy in the second week of sale (which seems fair enough), what they haven't said is if the 30% of any allocation that was kept back previously for the 2nd week will still be happening.  I'm assuming so or otherwise the full allocation could well be gone in the first week.

Even if the 30% are held back then that means for Chelsea for example there could be 660 tickets available for non-ST holders which could end up going in minutes (maybe I'm over-hyping things but £22 for first game of the season at the title holders could go quickly, even if it's live on TV and a 5.30pm kick off).

This seems like the perfect time to make use of the Jack Army points system which for the last couple of years has been largely redundant.  There's plenty of ways you can quantify who 'deserves' first crack at tickets, but is hopefully something that'll be looked at and that's before we get on to anything Wonky Sheep related....

There's plenty of conspiracy theories flying around and some inter-forum bickering but they are legit as confirmed by Vice-Chairman Leigh Dineen:
It's a bit of an extreme scenario but the main question is what happens if 3,000 Jack Army season ticket holders want to go to Watford but there's only 2,000 tickets? When the ticket office (virtual and real) open at 10am one Monday morning, who is at the front of the queue, do Wonky Sheep get priority or already have a number of tickets allocated to those who meet the ticketing criteria or do Wonky Sheep customers risk the ultimate nightmare of a weekend in Watford without even getting to watch football?