xG Leaders in the SPFL: Liam Boyce is the Hero we Deserve, but not the Hero we Want

The Expected Goals leaders table has been updated after this week’s matches, but I thought I would make specific mention of the weekend that Liam Boyce had. I have sung the praises of the Ross County striker often this season, but even I was left speechless after Boyce knocked in a hat trick this weekend to help lead Ross County to a 5-2 victory over Dundee. Kane Hemmings, who has also been having a good season for Dundee, knocked in two goals of his own but was no match for Boyce. After Saturday’s match, when we look at the xG over Goals chart we see Boyce reaching Adam Rooney levels of overachievement.

xG_G-2

However, when we look at Boyce’s xG compared to his shots, we see Boyce being right where he should be based on the amount of shots he takes. Boyce works hard and gets in plenty of shots in good positions and it has led to 12 goals this season, third in the league.

xG_Shots

Boyce is averaging 0.34 xG per 90 minutes and 0.70 goals per 90 minutes, showing the ability to be clinical all over the pitch this season. Along with being a clinical finisher this season, the former Cliftonville man has shown bags of skill, no more evident than in his second goal on Saturday where he used an overhead kick to put County up 4-2. Boyce had a total xG of 1.16 on Saturday, which was higher than his opposition Dundee had combined, as well as both Inverness CT and Aberdeen in the 2-2 Willie Collum shit-show that occurred Saturday. As a Celtic supporter, I see Boyce as someone who can have similar success in Ronny Deila’s system as Leigh Griffiths has. With Nadir Ciftci’s ability a controversial topic among Celtic support could Boyce be the depth at striker that Celtic need with Leigh Griffiths hurt at the moment? And more so, if a larger club makes Celtic an offer it can’t refuse for Griffiths, could Boyce be the replacement for Griffiths that could be plugged in right away without losing anything? Or could he even earn himself a bigger move down south, as surely if he keeps his form he will have to be strongly considered for the Northern Ireland squad for the Euros in France this summer.

xGp90_Gp90

xG Leaders in the SPFL: The Fine Lines Between Relegation and Mid-Table

As the SPFL season moves closer towards the halfway point in the season, the gap between mid-table clubs and clubs fighting relegation is tight. Only seven points separate fifth place Ross County and eleventh place Kilmarnock. With such a tight margin in the table between so many clubs, advanced stats can give us an insight in which clubs will be able to fight off relegation and which should start to be worried. Looking at the stats at the club level, we can guess that a club like Ross County probably will not have to worry about relegation if they are able to continue producing the numbers they have been so far. On the other end, clubs such as Kilmarnock, Inverness Caledonian Thistle, Motherwell, and Partick Thistle have produced some numbers that suggest they really could be in trouble. With not only the automatic relegation spot for finishing in last (a spot that Dundee United seem intent not to give up) but also the relegation playoff spot for eleventh place (and likely matching up with either Hibs or Rangers), one player on these clubs in the bottom half of the table could mean the difference between relegation and another year of SPFL Premiership football and the lucrative money that comes with it (SARCASM KLAXON!!!).

Despite being in the bottom half of the table, the likes of Kilmarnock, Motherwell, Dundee, and Partick Thistle all have a player in the top 10 xG numbers in the SPFL. Each of these players could be the key between going down and staying up.xG_G.png

Looking at the Expected Goals over Goals graph, we can see Louis Moult of Motherwell, Kris Doolan of Partick Thistle, and Kane Hemmings of Dundee, and Josh Magennis of Kilmarnock. Partick Thistle seemed like they were destined for that eleventh place playoff spot after the first third of the season, but they have won three times and drawn once in their last five matches to move from eleventh to eighth in the table and the play of Kris Doolan has a lot to do with it. So far this season, Doolan has six goals, all from open play. Comparing this to his xG totals, it seems that Doolan is performing right where he should be.

Going into their match away to league leading Celtic, Motherwell have seen an improvement in fortunes, winning two and drawing one in their last five matches (which in the bottom of the SPFL represents a huge uptick). It does not seem like a coincidence that this improvement is coinciding in striker Louis Moult appearing in the Top 10 xG leaders. Moult has seven goals, six from open play. When we look at his place on the graph, we see based on his xG numbers that this might be a bit of overachievement for Moult. Fans of the Steelmen will hope he can continue this purple patch, though we might see some regression from Moult if he does not raise his xG numbers and get into areas with higher probabilities of scoring.

xG p 90_G p 90

On the other end of the spectrum, looking at both xG/G and xG per 90 min/ Goals per 90 min can see some players who might need to do more to prevent their clubs from facing the drop. Both Josh Magennis of Kilmarnock and Kane Hemmings of Dundee are having good seasons. Both have scored five goals, all from open play and have xG totals over 4 which puts them in the top 10 in the SPFL. Yet, Kilmarnock and Dundee find themselves very much in the relegation battle because of poor play recently. Kilmarnock have earned less than a point per match the last ten matches and Dundee have earned exactly one. No one would place the blame at Magennis’ or Hemmings’ feet for this poor play, but compared to the other players that make up the SPFL Top 10 in xG, Magennis and Hemmings have been underperforming when it comes to scoring goals based on their overall xG numbers. The Dee and Killie’s position in the table could come down to Hemmings and Magennis and if they can put their clubs on their backs and carry them to safety.

Sheet 1

The graph above is the same as the first graph on this post, but I added Miles Storey, striker on loan at Inverness Caley Thistle from Swindon Town. Storey isn’t in the top 10 of xG, thus why he wasn’t in the first graph but I felt it necessary to add him to this graph because of his absurd goal total compared to his xG output. Now, we have talked about strikers in the SPFL over-performing on some of the clubs in the top half of the table with Adam Rooney, Juanama, Liam Boyce and even Leigh Griffiths a bit. In 11 appearances, Storey has a modest xG total of 1.91 yet has managed to bag six goals thus far, all from open play. Storey’s play has seen Caley move to 7th (albeit tied with Partick Thistle and Dundee and only ahead on goal difference), but the striker’s loan is over in January. If Caley cannot hold onto Storey through the new year, they might find themselves in a sticky relegation fight. On the other hand, if John Hughes can extend Storey’s loan or even bring him to the Highlands permanently then Caley could find themselves in the safety of the top 6 come split schedule time.

Advanced Stats after 16/17 Matches in the SPFL

Not only did the inclement weather stop half the matches this past weekend in the SPFL Premiership, it gives teams an uneven amount of games played which can really bother someone with OCD tendencies, such as your humble author. This is where measuring stats for teams in players in per 90 minute metrics comes in handy. I thought the bad weather could offer an opportunity to update some graphs and charts on different metrics in the SPFL.

xG_Points-2

Or view an interactive version of this graph here.

Celtic far and away lead total expected goals and average expected goals in the SPFL, which makes sense given that Leigh Griffiths is the leading goal scorer of the SPFL and has nearly 4 more expected goals than the next highest player. Along with Griffiths strong play, Celtic have asserted their offensive firepower with a Total Shot Ratio of 0.74 (meaning they have 74% of the shots in a match on average), a Total Shot on Target Ratio of 0.75 (75% of Shots on Target in a match on average), and a total xG of 31.60, which is over 11 expected goals higher than the next highest club in the SPFL. While Celtic’s attack has been good, the graph below might even suggest based on their xG total they should be scoring more goals. If Celtic can find that fourth gear on their attack, they could really put some room on the table between themselves and the rest of the league.

Link for graph

xG_G Team

TSR & TSoR

Link for graph

Looking at TSR and TSoR, it is no surprise why Kilmarnock’s keeper Jamie MacDonald has the second highest save percentage in the SPFL. Facing that many shots and shots on target, he is bound to get the gloves on some of them! Kilmarnock have some ugly shot numbers and combined with their not great expected goal totals makes me believe come end of season they will be worrying about that relegation playoff with the contender coming out of the SPFL Championship. Heck, if Dundee United get their stuff together, Killie could find themselves in the automatic relegation spot.

 

PDO and Luck in the SPFL

With half of the SPFL Premiership fixtures for Saturday cancelled due to inclement weather and no way for your average Scottish Football loving American to watch the matches that were not cancelled (seriously, SPFL I will give you around $20 United States Dollars a month to watch any match from the weekend online. Heck, I would even watch it tape delayed!), I turned to twitter to see what was going on in the three matches that were being played. Upon checking twitter, I saw this tweet describing the first goal in the Saint Johnstone-Ross County match. IMG_1187

You can commend David Wotherspoon for pressuring the keeper into forcing this mistake, but I think most would agree there was a bit of luck for Saint Johnstone to get a goal in this situation (to be fair to Wotherspoon, he scored later in the first half from an absolute rocket of a goal).  I had briefly discussed Saint Johnstone’s success despite low shot totals, TSR and TSoR, but the Saints match with Ross County, including their fortunate first goal, prompted me to look at PDO in the SPFL this season.

For lack of a better word, PDO measures luck (though some have a problem with that word). We discussed Dundee United’s lack of luck in relation to PDO, but as a refresher PDO is used to explain how a team performed historically, since if a team is scoring a lot of their chances and not allowing goals in the chances they allow, they are probably doing well. It measures a team’s efficiency. The formula for PDO is ( (Goals For / Shots For) + ( 1 – (Goals Against / Shots Against) ) ) * 1000. Looking at PDO and graphing it with xG we can see “lucky” and “unlucky” teams in the SPFL this season.

PDO-xG -Luck-

In the top right quadrant, we can see “Good and Lucky” teams, or teams with high PDO and high xG so far this season. Among these teams, it is no surprise we see the top four teams of the SPFL Premiership in Celtic, Hearts, Aberdeen, and Saint Johnstone. Celtic, unsurprisingly, have a high Expected Goal total and a good PDO. Aberdeen and Hearts have solid, though not spectacular numbers in both categories to back up their position on the table. Then we come to Saint Johnstone. The Saints are on the “Good and Lucky” side, but are awfully close to the top left quadrant of “Bad and Lucky”. Saint Johnstone’s PDO is a sky high 1102 on the back of their ability to seemingly score goals despite not having a high amount of shots. Their high PDO combined with their lower xG total points to a West Ham in the EPL this season type rise in the standings that might not be sustainable all season. Then again, this has been said about the Perth club for awhile now and they continue to churn out top 6 finishes in the SPFL.

In the bottom right corner we see Ross County as the lone occupants of the “Good and Unlucky” quadrant. County’s xG output suggests the Highland club should be scoring more goals, but their sun-1000 PDO hints at some bad luck for the Staggies. With the club in a top six position when few predicted such a strong start, could they dream of a possible European spot finish if the bounces start going their way, like their Highland rivals Inverness Caledonian Thistle enjoyed last season?

Speaking of Caley, Inverness Caley Thistle are the lone occupants of the “Bad and Lucky” quadrant, though they are not far off from joining the “Bad and Unlucky” quadrant. In that quadrant we see the previously mentioned Dundee United, who are on their own island in the sea of awful numbers.  Closer together in the “Bad and Unlucky” quadrant are fellow relegation worriers Kilmarnock, Motherwell, and Partick Thistle. In addition to the clubs who might have to start worry about that relegation playoff, we see Hamilton and Dundee who sit in the table 5th and 7th respectively. These clubs being near the top half of table might suggest that the talent level gap between teams fighting relegation and finishing in the top six on the table is not that great.

PDO-xPDO

Here we see the SPFL Premiership Clubs PDO compared to their xPDO. Again, in the Dundee United focused post we discussed the idea of xPDO brought up by the great Dutch football blog 11tegen11, but to refresh, normal PDO is calculated using shooting and saving percentages while expected PDO is calculated with xG per shot created and xG per shot conceded (formula: xPDO=( (xG For / Shots For) + ( 1 – (xG Against / Shots Against) ) ) * 1000)). Saint Johnstone stick out again with this graph with a high PDO, but their expected goals totals is not as high with the quality of their chances not backing up their high goal scoring. This is another sign that we might see some regression by Saint Johnstone in the table. Then again, they could continue to flick analytics in the eye.

 

The Curious Case of Dundee United

Without the mega money television deal their neighbors down south have secured, much of Scottish club football, with it’s grand history and passionate fanbases, remain virtually unknown to the world outside the United Kingdom. Even the largest of Scottish clubs located in the city limits of Glasgow cannot escape the fate of a smaller existence in the football world.

Those days seem ages away.

While the rest of the world may not be aware of it, Dundee United have a rich, deep history that fans of the club can be proud of. During the 1980’s, the Terrors success saw them become part of “The New Firm” with Aberdeen. In 1982-83, United won the Premier League, joining Aberdeen as the only non-Glasgow club to win the league since 1964. United even beat Barcelona in the semi-final of the 1987 UEFA Cup. While those heights have not been reached by the club for some time, the past few years have seen recent success with top six and even European spot finishes, a Scottish Cup, and a well known youth development system that has sent the likes of Ryan Gauld, Barry Douglas, Andrew Robertson, Gary Mackay-Stevens, and Stuart Armstrong to bigger clubs for decent profits. Yet, this year Dundee United find themselves in last place in the SPFL, manager Jackie McNamara sacked, and Mixu Paatelainen trying to pick up the pieces and keep United in the SPFL Premiership and away from the bare knuckle brawl of a division that is the Scottish Championship. So how did United find themselves from possible European Spot contender to relegation fodder? We can look at advanced stats for some answers.

Here we have the stats for this year’s Dundee United team, as well as stats for United after 16 matches (the same spot they are this season), after 24 matches, and the complete 2014-2015 season. Looking at the big picture, Dundee United had obtained 23 more points last season at the same point of the season they are currently. They then picked up 14 more points in the next 8 matches, before finishing the season by picking up only 11 points in the next 14 matches. Clearly something happened between the first part of the season and the last that saw United start their downward trend to where they are today. Now, you might be wondering why I chose to highlight where United were after 24 matches. I can assure you that this was not a coincidence picking this part of the season. After this point in the season, something changed for United that I believe led them to where they are now. That is the first game Dundee United played without these guys.

Celtic-europa-556194

Of course, the table can lie about the strengths and weaknesses of a team, so there is a need to look at all the stats to get a better idea of just what is happening. One stat we can look at is PDO. PDO was originally developed for ice hockey, you calculate PDO by the following formula: PDO = ( (Goals For / Shots For) + ( 1 – (Goals Against / Shots Against) ) ) * 1000. PDO is used to explain how a team performed historically, since if a team is scoring a lot of their chances and not allowing goals in the chances they allow, they are probably doing well. It measures a team’s efficency. When we look at Dundee United’s PDO this year compared to last year, it is a lot lower than last year. Scoring 16 goals less and allowing 13 goals more than the previous year at the same point will do that. Along those lines, the great Dutch football analytics blog 11tegen11 has come up with the idea of Expected PDO, or xPDO. They describe the difference between the two as “Normal PDO measures how efficiently a team has converted shots into goals, and prevented the opponent from doing so. xPDO measures how efficiently a team converts shots into goals, and prevents the opponent from doing do, given the shot quality for and against, and assuming league average conversion of xG into goals.” The formula for xPDO is xPDO = ( (xG For / Shots For) + ( 1 – (xG Against / Shots Against) ) ) * 1000. A team that has a PDO that’s much bigger than their xG might be “lucky” (a word that causes great hand wringing in the football analytics community) or their good results might be unsustainable. Similarly, a team with a much lower PDO than their xPDO might be considered “unlucky” or on the receiving end of some bad bounces. With a PDO of 902 and a xPDO of 989, Dundee United could certainly claim to be the bad side of Lady Luck.

https://public.tableau.com/views/DUxGAvg/xGAvg?:embed=y&:display_count=yes&:showTabs=y

xG Avg-2

Looking at Dundee United’ Expected Goals per match in 2014 16 matches in, 24 matches in, and 38 matches in compared to this season, we see a big increase. This season, United are averaging 0.85 xG per match, compared to 1.18 xG per match at the same point of the season in 2014-2015, 1.25 xG per match before United sold Gary Mackay-Stevens and Stuart Armstrong, and 1.08 at the end of the 2014-2015 season. We see a dip of 0.17 from that match 24 to the end of the season, a dip that continued this season for reasons we will continue to explore.

https://public.tableau.com/views/DUShotsPerGame/ShotsPerGame?:embed=y&:display_count=yes&:showTabs=y

Shots Per Game

One could assume with a lower xG average this season, Dundee United would also have a lower shot average and this is the case. This season(8.75 shots per game), United are averaging more than a shot less at the end of last season (9.92), 2 shots less to 16 games in last season (10.75) and 24 games in (10.875). Before Billy McKay’s arrival a few matches into this season, United’s attack was even more anemic, though it is clearly still not at the level it was last season.

https://public.tableau.com/views/DUSOTPerGame/SOTPerGame?:embed=y&:display_count=yes&:showTabs=y

SOT Per Game.png

Shots on target per game tells a similar story, though this season’s Dundee United have a slightly higher average than last season’s team at the end of the season, but not the team last season that featured Armstrong and Mackay-Stevens. Those two are not exactly volume shooters and are more creators than scorers, so while their absence was felt on the offensive end when they left, there was another departure from United. A striker who wasn’t afraid to shoot it anywhere and was rewarded with 14 goals last season, by and far the highest on United.

image-3

Stats for every player that played for Dundee United last season can be found here, though looking at Nadir Ciftci’s stats, you can see he clearly did not mind shooting. Ciftci had more than three times as many shots as the next United player. He knocked in 14 goals last season, though only 9 were not from penalties and with an xG total of 9.95 for the season that seems about right. Cifcti, Armstrong, and Mackay-Stevens were a huge part of United’s attack last season, each averaging over 0.5 Goals and Assists per 90 minutes. Take those three out without proper replacements and you will see one of the main reasons why Dundee United have struggled this season.

https://public.tableau.com/views/DUxGConAvg/xGConAvg?:embed=y&:display_count=yes&:showTabs=y

xG Con Avg-2

On the other end of the field, Dundee United have allowed 13 more goals this season than at the same point as last season. Yet, you probably cannot blame the back four for United’s bad start this season (or you can at least hypothesize that their performance is good enough to earn more than 8 points), as their xG conceded per match is nearly identical to the end of season total from last year and 16 matches into last year. In fact, the xG conceded total from this season is lower than it was 24 matches into United’s previous season, though no one has accused Gary Mackay-Stevens and Stuart Armstrong of being the defensive anchors of any teams.

https://public.tableau.com/views/DUShotsConAvg/ShotsConAvg?:embed=y&:display_count=yes&:showTabs=y

Shots Con Avg

https://public.tableau.com/views/DUSOTConAvg/SOTConPerGame?:embed=y&:display_count=yes&:showTabs=y

SOT Con Per Game.png

Save %

Similar to xG conceded, the total number of shots Dundee United are conceding to their opponents this season is actually lower this season that the same point last season and 24 games in. They are conceding slightly more shots on target than last year. The higher shots on target has meant more work for Dundee United keeper Michal Szromnik. Szromnik went from backup to Radoslaw Cierzniak to starter this year, though a 6% decrease in save rate from last year to this year might be an indicator of a downgrade.

The relegation of Hibs and Hearts (albeit Hearts’ financial problems gave them nearly no chance to stay up the year they were relegated) have shown that no club is “too big” in Scotland to stay up. Hibs and Rangers struggles and failure to win promotion from the Championship last season shows what a struggle that league is, no matter how big of a club you are. By experiencing some bad luck, not finding adequate replacements for their three best offensive players last season and a possible downgrade at keeper, Dundee United find themselves in serious relegation trouble. There are lots of points still available, though with the 11th place SFPL Premiership club having to play a playoff with a Championship club, pulling themselves off the bottom of the table does not assure Premiership football for the Terrors next season. There are a few other SPFL clubs that have flaws that make them susceptible to relegation and Dundee United should theoretically more resources than these clubs. Chairman Steven Thompson and Manager Mixu Paatelainen need to substantially improve the squad in January. If they do not, dark times could get even darker on the tangerine half of Dundee.