written by – Rangers Report
Tom Worville of Analytics FC wrote in his introduction to football analytics that analytics in football can be used to, “find patterns in play or significant trends. These trends and patterns form conclusions and then can inform decisions.” In an article explaining how advanced stats are changing how we analyze football – Sam Gregory of Canada’s Sportsnet added, “We have our preconceived notions that we get from watching the game, and then we look at the numbers to see what more we can learn. Finally we can make an opinion based on the combination of these two things.”
In that spirt here’s a highlight reel of a young striker – who has a nose for goal, can separate from defenders, is strong on the ball & in the air, & who appears to be a very good finisher when given the opportunity. (My apologies in advance for the awful music – I suggest muting the video to truly appreciate it). Then let’s look at the statistics to get a fuller understanding of what he can bring to Rangers next season.
Recently we offered up an evaluation of Rangers goalscorers & it was concluded that you could easily make a case that the team’s best attacking player was actually honing his craft at the bottom of the Championship table with Cowdenbeath.
Calum Gallagher featured ten times while on loan for Cowdenbeath last season & despite playing with a poor team, he was able to outperform his peers at Ibrox against the likes of Hearts, Hibs, & Queen of the South. Gallagher scored five goals in that time & while that total did not match the eventual output of Nicky Clark (eight goals) or Kenny Miller (seven goals) he played nearly 1,200 less minutes then those Rangers forwards.
To illustrate Gallagher’s successful results we broke down the scoring to a per 90 minutes average rather then just look a the overall tallies. Gallagher scored 0.58 goals per 90 minutes, while Clark & Miller both scored 0.36 goals per 90 minutes. What does that mean? Well, if you projected those numbers over a full season of 36 games -Gallagher would have scored 20.88 goals, while Miller & Clark would project to score only 13 goals.
That inspired me to investigate further. Let’s begin with some basics – how much did each forward play? (Remember: Gallagher’s stats were with Cowdenbeath.)
Even though Gallagher’s sample size is small, & Ryan Hardie’s is even smaller, that should not mean they are simply brushed aside. If anything the results that each of those forwards had in their limited time suggests that both deserve to see a drastic increase in their time this season – especially given the hope that this is Rangers last season in Scotland’s second tier. If not now, when?
But how dare you compare Gallagher’s 770 minutes to Clark’s 2000 minutes or Kenny Miller’s 20,000 minutes over his entire career? It’s such a small sample size.
Micah Blake McCurdy, a freelance hockey analyst, recently discussed small sample sizes on his Twitter feed. He argued, “Samples don’t have sizes. They are measurements you take and then there are estimates of how good those measurements are. Whenever anybody says ‘oh but that sample is small’ what they mean is ‘my intuition suggests a substantial uncertainty in that measurement’. I’m not suggesting that any of us should rely less on their intuition. Just the opposite, you should hone it on the whetstone of stats.”
If you watched the highlights of Calum Gallagher featured above – your intuition likely told you, “There’s a player there. There’s somebody that has the pace, strength, & intuition to be a productive goalscorer if given the chance.”
The data, albeit small, supports those assertions. Of course you could compile a bunch of the goals scored by Nicky Clark & Kenny Miller, put some (hopefully better) music to it & walk away with the same conclusions. That’s why it’s important examine the evidence provided by the actual results to properly assess each of these players.
More & more fancy stats people are applying ice hockey concepts to their coverage of football. The idea of combining goals & assists to accumulate points is an additional way to assess a forwards value to his team’s offensive success. Remember, these are the raw, overall totals. Converting these numbers to a per-90 minutes metric creates a much more meaningful assessment.
Playing for a poor Cowdenbeath side that scored the fewest goals in the Championship last season, Gallagher was nearly scoring or setting up a goal per game (0.82). According to Fitba Fancy Stats, as a team Cowdenbeath only average 0.89 goals per match. That makes Gallagher’s 0.82 points per 90 minutes that much more impressive. Probably the most surprising element of these results is that Gallagher also averaged more assists per 90 minutes then Miller or Clark did. Yes, the sample size is small but what’s more impressive? Kenny Miller’s four assists in 1,875 minutes or Gallagher’s two assists in 770 minutes? Furthermore, it took Nicky Clark nearly 2000 minutes to produce the same result as Gallagher did in less than half the time.
Another measure to include in this assessment is Total Shots Ratio (TSR). Usually applied to a team’s performance, TSR measures the ratio of total shots in a match that either team had. If Rangers had 15 shots in a match & their opponent had five, that would mean that Rangers had a TSR of 0.75 (15 shots divided by 20 total shots). This can give you a sense of what a team is doing with their possession, especially if you compare it to the actual possession numbers from a match. If a team had 55% of the possession, but only 45% of the total shots, you can argue that a team is experiencing ’empty possession.’ – holding onto the ball but not doing much with it.
While applying TSR to individual players only has so much value (there are after all 11 players in a side) it can still be an interesting comparison tool. Does a team perform better with certain players in the lineup?
To provide some context to the TSR data. According to Fitba Fancy Stats, Rangers’ TSR for the entire season was 0.65 (the third best in the league). When Kenny Miller was on the pitch, the team’s Total Shot Ratio was actually noticeably lower & it was slightly better when Nicky Clark was on the pitch. Miller’s number is concerning & should trigger further analysis by studying the match footage from last year to see what was causing this irregularity. There are ten other players on the pitch, maybe Miller was not the problem…but then again maybe his play was affecting the the team’s tactical approach. The data can’t explain it, but it should influence how performance analysts study Miller’s impact on the team’s play.
The same pertains to Calum Gallagher, who played for the team with the league’s lowest TSR (0.33). Even though Gallagher’s TSR is low (0.40), it further illustrated the positive impact he had on a bad team. The team’s TSR was significantly better with Gallagher in the lineup. What was it about Gallagher’s style of play that had this impact on the overall team’s play?
Another measure of a forward’s effectiveness can be an examination of their shooting statistics. Basically, how many shots they are getting in a match & how many are on target. This can give you a sense of their accuracy & efficiency in front of goal.
The following lists each player’s shots per 90 minutes, shots on target per 90, their overall shooting accuracy (what percentage of shots are on target), & their overall shooting percentage (what percentage of shots on target end up being goals).
|Sh per 90||SOT per 90||Shooting Accuracy||Shooting %|
While Miller & Gallagher’s shots on target numbers are pretty similar, you can see what enhanced Gallagher’s results was his shooting percentage. More than half of the shots he got on target ended up beating the goalie. Also, Nicky Clark’s impressive 2.30 shots per 90, is diminished by a dismal shooting accuracy of 0.411. This is where it is crucial to actually look at where these players are taking their shots.
Based on the information available, I will be using four different ‘danger zones’ for shot location this season. The terminology & thought process is based on similar models for the National Hockey League & the one created for football by Michael Caley.
Using the above depiction of the penalty box, these are the four different danger zones that I will be referencing:
Very High dangers shots: this is any shot coming within the ‘goal area’, or the within vicinity of the six yard box
High danger shots: any shot that originates from a central location in the penalty box – from the goal area out to the edge of the penalty box. The rationale is that the shooter is either relatively close to the goal or if he is further out, has more of the net to shoot at & can beat the goalie to either site. Also if the shooter is further out it is more likely to have the ball deflected, making the save more difficult.
Medium danger shots: shots originating from inside the penalty box but coming from either side of that central area. The shot is coming at an angle which means the shooter has less of the goal to shoot at.
Low danger shots: any shot that is launched from outside of the penalty box.
Note: You can click on the following charts to see them in more detail.
Four of Gallagher’s goals came in High Danger areas & his fifth came in a Very High Danger area. Of his twelve shots, nine came in either a high danger or very high danger area. The more shots a player take in these zones, the more likely they will be to score goals.
Six of Miller’s seven goals came in High Danger areas & his seventh tally came from a Medium danger area. He had three shots in Very High Danger areas but failed to score. While finishing was a bit of an issue for Miller last season, he simply wasn’t getting enough shots off to enhance his effectiveness. Both Gallagher & Miller had low shot totals – but Gallagher was playing for the worst team in a subpar league. Yes, Miller’s service was poor but when you are evaluating a 35 year old forward – you must wonder how much of this may come down to a player playing out the final stages of his career? No one can question his work ethic or his instincts – but it is time to question the results.
I know that Miller does a lot of work off of the ball, something that is difficult to measure using statistics. But it would be shortsighted to say that a younger player couldn’t do the same kind of work with the proper coaching.
The first question I would have for Nicky Clark is – why the hell is he taking so many shots from outside the penalty area? He took 15 of his 51 shots from long distance & none of them were free kicks. He’s not going to score from out there.
While he did score one goal from a Low Danger area it should be noted that it was a header so it would be the textbook definition of a fluky goal. Clark does not have the skill-set to be scoring from this range & honestly these shots are wasted opportunities.
To be fair to Clark, none of the other forwards that we are looking at scored from a Low Danger area. A forward does his damage in the penalty area. Clark scored four of his goals from a High Danger area & three from Very High Danger areas. This is where he can be effective – picking up the scraps in front of goal.
One last metric I’d like to use looks to measure how effective a forward is when his team actually has the ball. It is based on some ideas that North Yard Analytics’ Dan Altman discussed in his short series of videos on YouTube. While evaluating a player based on a per 90 minutes ratio is so much more valuable then simply looking at final totals – what if a player plays on a bad team? What if a forward, like Calum Gallagher, played on a team that only had 46% of possession in his games? How many goals shows real value if your team doesn’t have the ball as much as the opponent?
So, rather then assessing a forward based on 90 minutes, Goal Scoring Efficiency Rating (GSER) looks at the goalscoring rate for how many minutes his team actually possessed the ball. (Keep in mind that the possession statistics for each player are not exact. They are for an entire match & often the player does not play the entire 90 minutes)
In Gallagher’s 770 minutes Cowdenbeath only possessed the ball for 354 of those minutes. He scored five goals in those 354 minutes giving him a Goal Scoring Efficiency Rating of 0.014 – a number that doesn’t mean much without some comparables.
Rangers possessed the ball for 1,050 of the 1,875 minutes that Kenny Miller played. Miller scored seven goals in those 1,050 minutes leading to a GSER of 0.007. That’s half of Gallagher’s rating. If you applied Gallagher’s Goal Scoring Efficiency Rating to those same minutes he would have hypothetically scored 14 goals.
Nicky Clark’s Goal Scoring Efficiency Rating was identical to Miller’s 0.007, while Ryan Hardie’s was 0.017.
The consistent theme in all of these stats is that there was very little separating Nicky Clark from Kenny Miller & neither player was particularly impressive.
Their results weren’t horrible, but they also weren’t very inspiring. I think most rationale supporters felt the same way from watching the two on a regular basis last season. Clark & Miller were average performers last season. Average is OK if your aspirations are to be a mid-table team. The expectations for a club like Rangers is to win the league. That is the bar that was set by Graeme Souness & is the standard that is part of the culture at Ibrox. Finishing third in Scotland’s second tier was an embarrassment. Average performances won’t cut it this season.
There’s no guarantee that Calum Gallagher would be any better. However, Gallagher’s results in limited time warrant more opportunities to justify his worth. Maybe his statistics would regress with more minutes & the results would be on par with Clark & Miller. Then you would know that he isn’t the answer. But you can only know that by giving Gallagher the chance to prove himself with Rangers. If he actually produced at a rate similar to his results with Cowdenbeath then Rangers would be much better off.
Keep in mind that most of the names linked to a move to Rangers have not been forwards. Conventional wisdom in modern football is that you build your team by solidifying the backline first. You develop a team from a strong defensive foundation first. There is a very good chance that Mark Warburton begins the season relying on the forwards that are already at the club. Knowing that – it would be foolish to not give Calum Gallagher a chance to prove his value this season.
If he could succeed at Cowdenbeath, why can’t he succeed for Rangers?
You can follow Rangers Report on Twitter @TheGersReport