Quantitative Look at Full Backs in the Canadian Premier League

In the second part of this series, I will study full backs of the Canadian Premier League. At the end of this article I hope to conclude on the highest performing full backs of the CPL 2019 season.

The analysis will be split between overall influence on the team, defensive qualities and what the full back provides in terms of attacking play. The role of a full back has changed in recent years, with Dani Alves, Jordi Alba, Marcelo, Andrew Robertson and Trent Alexander-Arnold leading the way with many attacking inputs alongside their traditional defensive responsibilities.

CPL Full Back Insights

Before starting my review, I wanted to bring attention to the fact I have never watched any of the Canadian Premier League soccer. I understand the structure of the league and know the names of the seven teams, but I do not know any results, league standings or values/wages of players. This enables my review to be purely statistical without any bias from preconception views of the players, or support/preference for a particular team.

As before, for our analysis we will only investigate players that have played over 1000 minutes in the CPL 2019 season, this provides us with a group of 15 full backs.

Initial Insights


Under the assumption that the more touches a player has per game, the more influential they are on the team, the first stat would suggest that Doner, Abzi, Awuah are heavily influential on a match whereas De Caroli and Verhoven influence is somewhat dampened.

The percentage of touches in the defending third begins to tell us the story of what type of full back the player is. With just one in every three of Franco’s touches being inside the defending third of the pitch, he seems to be the most attacking full back. It seems this is the way Forge set up tactically, since their other full back, Awuah is the second most attacking full back. HFX Wanderers seem to play with defensive minded full backs, with both Langwa and Bona having the majority of their touches in the defending third.

Defending Insights


*A duel is an 50-50 contest between two players of opposing sides in the match.

First and foremost, a full back is a defender and must be able to defend. Sliding in and winning 50/50 balls, tackling and disposing the opposition winger and cutting out through balls are all synonymous with famous full backs over the years (Roberto Carlos, Paolo Maldini, Lilian Thuram etc.). Viewing the two metrics, duels win percentage and tackling success rate, as a pair, as representation for 1-1 defending ability, Esua, Bona, Smith and Langwa perform favourably. De Carolis appears to underperform.

Intercepting and ball recoveries can quickly turn a defending situation into an attacking phase of play. It can turn a potential goal conceded into a potential goal scored. It is the key fundamental to the World Club Champions, Liverpool’s style of play. Equivalently, being disposed is essentially the opposite, resulting in an attacking situation being turned into a defending situation. Hence a player will be aiming to have a high metric for interceptions and ball recoveries but being disposed only a few times. Smith seems to appear the most favourable from these metrics, with Abzi, Frano, Bona also reflecting well. It is interesting to note Awuah seems to complete the most inception and ball recoveries, while also being the full back who loses ball possession the most. Langwa, despite performing well in the duel and tackling metrics seems to underperform in this metric.

Bona and Smith were the two players who performed the best across the first four metrics. They are also the two full backs with the most clearances – suggesting they are the best two full backs in the league based on the defensive side of the game.

Attacking Insights


There was limited useful data available for the attacking aspects of full backs. For a lot of the metrics, such as average distance of pass made (which would be able to shed some light on whether the full back typically does short safe passes or Ashley Cole/Trent Alexander-Arnold-style cross pitch passes to quickly switch the phase of play), there was very little difference between each player’s value. However I have found insight in the following metrics:

Only six goals for the season came from full backs, with Verhoven being responsible for two. Hence, I combined the goal and assists figures together. Verhoven performed best in this metric, with Abzi, Awuah and Smith also performing favourably. As discussed in the first set of metrics, HFX Wanderers full back appear to be defensive minded, which would likely explain why neither of their three regularly appearing full backs recorded a goal or an assist.

The chances created metric seems a fairer metric for full backs, as assists are so reliant on the final player’s shooting accuracy. If a full back sets up a really good opportunity to score but the striker has a poor shot, the full back should still get credit. Again for this metric Verhoven performed the best, with Awuah, Abzi and Smith also performing favourably.

I wanted to look into the pass performance of the full backs in the opponents half. Obviously a higher pass success rate is best but I also included the number of passes to help give representation to the success rate. Despite having the best end product in terms of assists and goals, surprisingly Verhoven seems to only play a few successful passes in the opposition’s half per game. Awuah and Frano seem to play the most passes and have the most involvement in the opposition’s half and both have high pass success rates.

It appears Awuah and Verhoven are the two best attacking full backs in the league. Despite not being at the top of any metrics, Smith performs quite well in the attacking metrics. This, alongside his supportive defensive works makes me consider him as the league’s best full back. How does your club’s full backs perform?

Studying for a Career in the Sports Industry

Several times over the past few years, I have considered studying a masters or MBA within sports analytics or finance. I never found a central list of many courses, let alone anything that compared different courses at different universities.

Below is a list of the courses I found and felt were interesting enough to be bookmarked. I have to admit that this list is not extensive and one thing I have seen is that the number of potential courses is rapidly increasing as the sports analytics and finance industries both grow. I am hoping that this list helps someone who is starting to research potential courses in this field.

The list is split into analytical degrees, finance degrees, other sports related degrees and then a final section on shorter courses, some of which are completely free. The courses is heavily tailored to soccer, due to that being my personal preference when originally seeking out potential courses. One thing I quickly found was that hundreds of universities offer Sports Management based MBAs, so I have not provided a list for them.

Analytical Degrees

Finance Degrees

Other Degrees

Free/MOOC Courses/Short Courses

Canadian Premier League 2019 Review

In a series of posts, I will complete statistical analysis to determine the league’s best player in each position. Use the links below to take you to each position’s analysis page.

Before starting my review, I wanted to bring attention to the fact I have never watched any of the Canadian Premier League soccer. I understand the structure of the league and know the names of the seven teams, but I do not know any results, league standings or values/wages of players. This enables my review to be purely statistical without any bias from preconception views of the players, or support/preference for a particular team.

CPL Goal Keeper Review
CPL Full Back Review
CPL Center Back Review
CPL Winger Review
CPL Central Midfield Review
CPL Center Forward Review

Canadian Premier League Background

The Canadian Premier League (The CPL) was founded on 6th May 2017 and played its first full season between from April 2019 to October 2019. The 2019 league had seven teams from five different Canadian provinces. An eighth team will join for the 2020 season, the Atlético Madrid owned: Atlético Ottawa.

As with many North American sports, the league is used to determine playoff qualification and seeding, with the playoffs determining the competition winner. CPL team play against the three Canadian MLS based teams (Toronto FC, Vancouver Whitecaps and Montreal Impact) in a second competition, the Canadian equivalent of the FA Cup (the Canadian Championship). The winner of the Canadian Championship is awarded Canada’s berth in the CONCACAF Champions League.

The CPL’s focus is to improve the level of talent and interest in soccer in Canada. To achieve this, the league has several unique features. These include a minimum quota of Canadian players on team rosters and starting line-ups, requirements for Canadian under-21 players, and a Canadian university draft. In addition, the CPL releases all of the data collected, for free, to aid the league’s coaches and to encourage fan and analyst engagement. In order to obtain a copy of the data, visit the Centre Circle section of the CPL website.

Canadian Premier League Analysis : Goalkeepers

The below is is a portion of an article I wrote for Sports Tech Daily – which can be read in full here.

CPL Goalkeeper Insights

Before evaluating each keeper’s stats – let’s determine which keepers played a substantial amount of minutes in the season. We want to compare as many keepers as possible, but we want to ensure each keeper has played a sufficient amount of minutes to ensure that their stats are reliable and fair to compare against.

Player Club Total Minutes Played
N. Ingham York9 2340
M. Carducci Cavalry 2250
C. James Edmonton 1979
T. Henry Forge 1979
C. Oxner HFX Wanderers 1489
T. Farago Valour 1440
M. Village Pacific 1440
N. Wirth Pacific 1080
M. Janssens Valour 1080
J. Williams HFX Wanderers 1028
Q. Roberts Forge 720
D. Powley Edmonton 540
N. Giantsopoulos Cavalry 450
C. Vance York9 90
M. Silva York9 90

I decided to only inspect keepers that have played over 1000 minutes in the CPL, giving us a group of 10.

Initial Insights – goals conceded, total clean sheets, goals conceded per 90 mins


Oxner was the keeper who conceded the least amount of goals and Ingham the most. However once we reflect the amount of minutes played by each keeper, Carducci, closely followed by Henry conceded the least amount of goals per 90 minutes played and Wirth the most. I prefer to focus on goals conceded on average per 90 minutes rather than the number of clean sheets – but it is worth noting that both metrics provide the best performers (Carducci and Henry) and the worse performer (Wirth).

Many factors beyond just the goal keeper contribute to goals conceded; quality of the rest of the team, quality of the defence, shots faced etc, that lets move onto metrics more goal keeper focussed.

Further Insights – errors leading to goals, goals conceded outside of the box, expected goals conceded


**Expected goals to concede is a CPL determined metric. Each opportunity for the opposing team is assessed with the likelihood of it resulting in a goal being scored, to give an expected goal conceded value. (i.e. big chance with a 60% likelihood of being scored = 0.6 expected goals to concede). Hence for Ingham the CPL determined due to opportunities faced, he should have conceded 29 goals. He actually conceded 34 goals, meaning he under performed by 5 goals based purely on shots faced.

Across the group, there was very few errors leading to goals. The main take from this metric would be the 4 keepers (Carducci, Henry, Farago and Wirth) who maintained concentration across their minutes played and did not produce a single error.

Apart from the occasional wonder goal, you would consider a high quality keeper to always be in position to save all shots from outside of the box. This metric reflects favourably for Carducci, who did not concede a single goal from outside of the box. Ingham seems to have difficulty saving shots from outside of the box – a trait opposition teams should aim to take advantage of when playing York9.

The expected goals to concede is a sophisticated metric and takes into account the quality of each opportunity faced. For example a shot taken by the preferred foot of a striker is assigned a higher probability of resulting in a goal vs a shot by their weaker foot/head. For a detailed insight into how this metric is calculated , click here.


Ingham again reflects poorly in this metric, conceding 5 more goals than would be expected based on shots faced. However this metric reflects worse on Williams, who conceded 8 more goals than to be expected. Henry performs favourable in this category, having conceded 4 goals less than predicted. Carducci was his closest competitor with a differential of 2. I converted this metric into a percentage to remove any bias in the size of expected goals. Williams – 74% performance stands out sharply from the rest of the pack.

Good Traits – % of crosses caught, % saves collected, % saves diverted to safety rather than danger,  , passing ability, % saves made stood up vs dived for


I wanted to do some additional insights on ‘goal keeping best practice items’.

Firstly, catching crosses. If a keeper catches a cross, it results in possession for his team. In contrast to punching, which could result in either team gaining possession and may lead to additional shots on goal. Farago and Village performs best in this category, with Janssens’s data demonstrating hesitance in catching crosses at all.

Similarly, if a keeper collects the ball after saving, it prevents the opposition team having additional opportunities to score and gives the keeper’s team possession. Again, Village and Fargo perform best in these category with Janssen performing poorly.

A keeper will never be able to collect all shots, but they are able to direct the ball into a safe area of the pitch. Williams, Ingham and Village again perform well in this category. Janssen shows a tendency to divert the ball into a dangerous area rather than a safe area – a weakness multiplied by his poor stats at collecting shots.

There seems to be merit in investigating the ratio of shots made stood up vs. dived for. If a keeper has good positioning, then they should be able to make simple saves, rather than requiring ‘Hollywood’ diving saves. However, this metric is somewhat heavily influenced by the type of shots the keeper is facing (i.e. weak shots close to him, difficult shots towards the corners etc.). Without putting too much influence on this metric (due to the variability that gives rise to it), Farago performed best.

In modern times, the keeper’s passing ability has become a much greater focus. Henry and Ingham have the highest pass success rate. However, Henry has the lowest average passing distance of all the keepers, suggesting Henry typically plays short, safe passes to his defenders.


In the search for the Canadian Premier League’s best keeper: there does not seem to be a clear winner. This is an indicator that the CPL’s goal keepers are strong in different categories and there is no keeper that emerges at the top of the pack. Based on the metric studied, the strongest performing keepers seem to be Carducci and Henry, with Farago also displaying good behaviors. Who would you pick?

Click here to read the remainder of the CPL series.

Sourcing Sports Data Sets

Below is a copy of an article I wrote for Sports Tech Daily


Before you’re able to test any of your sports theories or hypotheses, you will need data.

The collection of sports data is a rapidly growing industry. There is a huge commercial demand for sports statistics. For years, betting firms companies were the only ones collecting sports related data. Now, media outlets, broadcasters, sport teams themselves, players’ agents and scouting divisions are all trying to obtain and create data driven insights.

For soccer, there are currently several companies that provide very detailed data sets for the most popular soccer leagues. Opta Sports is one of the best known companies and charges various fees for different levels of data. One of the main features of Opta Sports soccer data is they are able to provide full x,y pitch location co-ordinates of every on-the ball event. Squawka and Deltatre are two other well-known companies that provide detailed sports data. Other companies operating within this space offer slightly different services, such as Stats Perform who perform sophisticated data analysis on these data sets and provide insights to over 1,800 global organisations.

There are also several websites providing free data sets for many of the leagues. Many of these websites started out by providing information to betting pundits, but are now providing a platform for many enthusiastic soccer fanatics to create personal blogs. The most popular websites are:

https://www.soccerstats.com/ : team and player data for over 200+ leagues. In addition SoccerStats provides various statistical analysis including percentage of games in league won by home/away team, won by 1/2/3 goals.
https://footystats.org/ : team and player stats for 600+ leagues
http://football-data.mx-api.enetscores.com/ : scores, lineup, team formation and events
http://www.football-data.co.uk/ : historic results and betting odds.
http://www.footballsquads.co.uk/ : current and historical squad details for clubs and national teams from all across the world
https://www.statbunker.com/ : team and player stats for multiple leagues
https://fbref.com/ : team and player stats for multiple leagues
http://sofifa.com/ : EA Sports FIFA game based players and teams attributes

Additional sources

Kaggle : Programming website which users can access various data sets
GitHub : Another programming website that has various data sets available
Reddit : Many independent data collectors provide links to their data

Canadian Premier League

Canada reformatted their soccer league system in 2019. One of the new features of the Canadian Premier League (CPL), with help from their data partner Stats Perform, is to provide all league data to the public for free. This initiative was done to help develop interest and knowledge of Canadian soccer. Oliver Gage, Head of On-Field Performance and Recruitment for the Canadian Premier League explains “The ability for Canadian fans, aspiring analysts, scouts and coaches to access this data on our league is an essential part of a wider strategy to promote this side of the game in Canada. Empowering a community with the ability to hone their analysis skills, will no doubt help our clubs, players and Canadian soccer as a whole in the long run. I look forward to seeing, questioning and promoting the articles and ideas which the release of this data will no-doubt encourage.”

This data set is a great place to start your analytics path. The data provides over 100 player focused metrics. For outfield players, the stats go beyond the basics of minutes played, number of substitute appearances made, but includes figures such as: passes attempted in the attacking third, aerial dual win percentage, fouls committed in the defending third. For goalkeepers, the data includes number of crosses caught, number of diving saves made, number of saves made with body.

In order to obtain a copy of the data, visit the Centre Circle section of the CPL website (https://canpl.ca/centre-circle-data/). The CPL is eager for people to publish their data insights and findings onto the official Centre Circle Data Twitter handles @canpldata #CCData and #CanPL, in order to increase awareness.

This article was written by David Martin. Check out his site ‘Actuaries FC’ where he’ll be writing more sports analytics content as well as writing with us.

The Journey Begins

Thank you for joining me.

I am currently an Actuary by trade, with a huge interest in football. I will be using this website to document and present football insights I have gathered via applying actuarial and analytic techniques.

For years I have been interested in football analytics, reading multiple books and searching the website for any sort of material. I will also be sharing interesting information I find online.

Good company in a journey makes the way seem shorter. — Izaak Walton