Identifying Triple Threat Forwards Using Data
Who is the Michael Jordan of Football?
What is a “Triple Threat” Forward?
A triple threat forward is a player that adds value to their teams possessions through carrying, shooting, or passing, meaning that they are less easy to defend against as their threat is varied and less predictable. This idea first came to me when reading Ian Graham’s 2024 book “How to Win the Premier League”, where he discusses this notion with regards to that dominant Liverpool front 3.
The idea of usage led us to question whether attackers were adding value by using up possessions or by keeping them going. Our favourite type of striker was the "triple threat" - a player who could add through shooting, passing and dribbling. These players are difficult to defend against - they can choose to pass or dribble instead of shoot. And they use up fewer possessions than players whose only skill is shooting. At Liverpool we had the luxury of three triple threats in Roberto Firmino, Sadio Mane, and Mo Salah. Even so, Salah was the highest volume shooter, and benefitted from the passing ability of Firmino. Separating players' contributions into different types of actions - passing, shooting and dribbling - allowed us to begin to understand teamwork. Too many shooters and the ball does not get passed into dangerous locations. Too many passers and the ball does not get shot into the opponent's goal. Likewise, Salah's presence increased Firmino's impact on the team.
Adding value to possession through shooting is simple to understand: it is taking many high quality shots, therefore directly impacting the team’s chance of scoring.
Creating value via passing can happen in multiple ways. The most direct method is generating passes that lead to shots. However, passes can also create value indirectly by leading to situations that could lead to shots. In essence, high-value passes are ones that move the team closer to the opposition goal and into dangerous positions. These passes don't necessarily have to lead directly to shots to create value.
High-value carrying also manifests in various ways: from dribbles that directly create shooting opportunities, to carrying the ball into more dangerous areas of the pitch. It includes drawing fouls from opponents and maintaining possession rather than consistently losing the ball.
I believe that Ian Graham may have drawn inspiration from basketball's "Triple Threat Position" - a fundamental physical stance that allows players to either shoot, pass, or dribble while keeping defenders guessing. Basketball legends like Michael Jordan, Kobe Bryant, and LeBron James mastered this position, using their versatility to dominate the game. Graham notes that Liverpool's front three possessed similar varied skillsets. While the specific mechanics differ between sports, both concepts share the same core principle: players who can simultaneously threaten multiple offensive actions are inherently more difficult to defend against.
Modern football has seen a clear shift from specialised players to more versatile and generalised footballers. I would then argue that triple threat forwards are likely much more common today than in the past. Past eras predominantly featured specialists - defenders focused on defending, midfielders on passing or tackling, wingers on dribbling and crossing, and strikers on finishing. While this is a simplification, Manchester United's 1999 treble-winning side exemplifies this specialist setup. While exceptional all-rounders existed - Johan Cruyff and Dennis Bergkamp in attack, Franz Beckenbauer in defence - such versatile players were rare. Today, being limited to a single skillset is often a liability at top clubs, with even goalkeepers now expected to contribute beyond their traditional duties.
Understanding how forwards create value in possession is crucial for recruitment and building balanced attacks. While having multiple triple threat forwards is ideal, as demonstrated by Liverpool's Salah-Firmino-Mané combination, teams can also create effective attacks by combining players with complementary strengths. A front three composed entirely of players who excel only at shooting may struggle, for example, as they lack the passing and carrying abilities to consistently create opportunities for each other. The key is achieving balance - whether through multiple triple threat forwards or through a combination of players who excel in different aspects of attacking play (one that springs to mind here is the Jovic, Haller, and Rebic front 3 for Frankfurt all the way back in 2019 - the magic triangle! They had varying skillsets that complimented each other perfectly).
How Can We Measure These Threats?
Shooting ability is the most straightforward of the three threats to evaluate. We can assess it primarily through two metrics: shot volume and non-penalty expected goals. At the time of writing, Erling Haaland leads in both categories which demonstrates how effectively these metrics capture shooting threat.
Passing ability can be evaluated through several key metrics, each capturing different aspects of creative threat. Expected assists (xA) measures how likely a pass is to become an assist, while expected assisted goals (xAG) quantifies the quality of shooting chances that follow directly from a player's passes. These metrics directly reflect how passing impacts team scoring opportunities. Supporting metrics like key passes and passes into the penalty area serve as valuable indicators of passing that creates dangerous situations. Additionally, progressive passes and passes into the final third represent high-value actions that advance the team toward goal - a crucial element in building attacking threats.
Carrying ability is the least straightforward of the three threats to evaluate. A key indicator is carries into the penalty area, where players directly advance the ball into the most dangerous zone. This relates closely to touches in the penalty area, which indicates dangerous possession. Successful take-ons are another crucial element, particularly those leading to shots, as beating defenders in 1v1 situations creates numerical advantages and helps break down organized defences. Progressive carries and carries into the final third demonstrate a player's ability to advance their team up the pitch through dribbling. The value of carrying extends beyond just advancement - drawing fouls gains territorial advantage, while consistently maintaining possession through carries reduces the risk of counterattacks.
To evaluate a player's proficiency in each threat, I created weighted averages, placing more weight on the statistics I felt provided more value (xAG has a higher weighting than Passes into Final Third, for example) using the percentile rankings of the relevant metrics. See below for a breakdown of the top 5 forwards in each of the ensemble scores I have created.
Clustering
After establishing the measures for each threat, the next task was to determine how to categorize these potential Triple Threat Forwards. One approach involved identifying candidates by selecting players who scored highest across all three ensemble metrics combined. Initial analysis revealed the top 5 players based on their overall scores.
However, this approach proved overly simplistic and failed to capture the diverse skill profiles among attackers. A more sophisticated method involves using the k-means clustering algorithm, which employs machine learning to identify distinct groups of players based on their three metric scores. The algorithm naturally groups similar players together: those with high scores across all three metrics form one distinct group, while those excelling in passing but showing lower scores in other areas form another. This data-driven approach enables player classification without the influence of pre-existing biases.
The k-means algorithm requires only one input from the user: the desired number of clusters to be formed (and a random state value, actually). Selecting this number is more of an art than a science - it requires careful consideration and experimentation. I personally tested various values until I found one that best represented the different attacking profiles in the data. I landed on 6 clusters, a choice supported by the elbow plot below.
Here is a link to a page that can explain an elbow plot better than I ever could. But, in essence, it shows that the more clusters added will improve the accuracy of each cluster, but there is diminishing returns with every one you add.
Reviewing the Clusters
Looking at this 3D scatter plot, I've managed to cluster forwards based on three metrics: shooting, passing, and carrying threat. Each dot represents a player, with their position in space determined by their percentile scores in these categories. The cluster centres (marked with red X's) help define six distinct groups of forwards, each with their own characteristics.
Let's break down each cluster and look at who's ended up where - I think you'll find some of these groupings both validate what we see on the pitch and throw up some surprising insights about players we thought we had figured out.
Cluster 0: The Shooting Specialists (Strong Shooting)
This cluster represents players who excel in shooting, and less in the other two categories. Here we see some notable names such as Robert Lewandowski, Erling Haaland, Dominic Solanke, and Moise Kean. It's probably the easiest cluster to intuitively verify just by looking at the names - these players are typically centre forwards, so naturally the majority of their contributions come in the form of shots as they are closest to the goal.
This the largest cluster with 88 players. There are a couple of surprising inclusions in Nicolas Jackson and Darwin Nunez. Both also have relatively high carrying scores (low 60s), which might suggest they're better suited to a later cluster. However, their extremely high shooting scores (94 and 84 respectively) gravitate them towards this group instead.
Cluster 1: The Creative Shooters (Strong Passing & Shooting)
This is the first double threat cluster - players who excel in shooting and passing, perhaps representing forwards who drop deep and link rather than spearheading the front lines. I would expect fewer wingers to be present here as this skill combination feels more suited to central forwards.
This is the smallest cluster with 42 players, and there are some notable names in Harry Kane, Paulo Dybala, Bruno Fernandes, Eberechi Eze, and Phil Foden. At first glance, Foden's inclusion may be surprising - he is renowned for his dribbling ability and close control. However, his carrying ensemble score is only 53 for the data so far this season, suggesting he is not creating that much value through dribbling. The other names ring true for this category.
An interesting inclusion in this cluster is Antoine Griezmann. You may instinctively think that this makes a lot of sense for his skillset, but his shooting score is a measly 38. This must be a case where his high passing score (87) and low carrying score (33) means that this cluster suits him better than any other, even if his suitability is not that great.
Cluster 2: The Triple Threat Forwards (Strong Passing, Carrying, & Shooting)
These are the all-rounders we've been searching for - our true Triple Threats. The cluster validates our earlier analysis, containing all five top-ranked players from our overall scoring system. We see elite talents like Bukayo Saka, Mohamed Salah, and Kylian Mbappé here, which speaks to the cluster's quality. Interestingly, the group is dominated by wingers rather than traditional strikers, highlighting how rare it is to find central forwards who consistently excel in dribbling.
An insight comes from some unexpected inclusions: players like Yankuba Minteh and Nico Williams made the cut despite relatively modest shooting scores (45.12 and 46.74 respectively). Their high scores in passing and dribbling pulled them into this elite group, as they aligned more closely with these all-rounders than any other cluster profile. While additional clusters might help distinguish these edge cases more precisely, the current classification demonstrates remarkable accuracy. Out of 43 players in this cluster, only a handful represent unexpected inclusions - suggesting that our six-cluster approach effectively captures the Triple Threat forwards. The presence of these few outliers actually validates our methodology, showing how the clustering algorithm prioritizes overall skill profiles over individual metric thresholds.
Cluster 3: The Creative Dribblers (Strong Passing & Carrying)
This cluster represents the forwards who can create threat through a combination of dribbling and passing, I would imagine usually in the same possession. In my eyes, the archetypal forward for this cluster would be Bukayo Saka but, as mentioned previously, he found himself in The Triple Threat cluster due to his shooting proficiency. Forwards who do find themselves in this cluster leave a lot to be desired in the shooting department.
Some names that validate this cluster are: Jadon Sancho, Dejan Kulusevski, Jeremy Doku, Takefusa Kubo, Morgan Rogers, Bernardo Silva, and Dwight McNeil. The players clustered here tend to represent one of two profiles: creative midfielders used on the wing, or high-usage dribblers with poor shot selection.
I don't believe there is a single central forward in this cluster. Perhaps the most surprising inclusion is Bryan Mbeumo, as I personally associate him with being a dangerous shooter. However, with a score of 37 in shooting, it's now clear that the majority of that danger comes from good finishing.
Cluster 4: The Bad Forwards (Strong in Nothing)
Perhaps my favourite cluster, the forwards who do not excel in any one of the 3 skillsets. Some notable names include Rasmus Højlund, Marcus Rashford, Jordan Ayew, John McGinn, and Jack Harrison.
The clear issue with this cluster is players that are categorised as forwards but have not actually played as a forward much or at all. Some names to confirm that are Joelinton, Patrick Dorgu, Keane Lewis-Potter, and Ben Johnson. Unfortunately, I am limited to what fbref categorises for player positions - of course I could manually go through and deselect players who do not fit the bill for playing as a forward this season, and this could be something I do in the future to improve the clustering.
Despite that, the clustering is intuitive, and there are no real surprising inclusions. The forwards are all having bad seasons, or play for a bad attack. The inclusion of 2 Manchester United forwards instantly made me confident in the clustering.
Cluster 5: The Dribbly Shooters (Strong Carrying and Shooting)
Here we have the forwards that are dangerous on the carry, and create value in their shooting. There are some names in that cluster that made me nod my head in agreement: Alexander Isak, Dango Outtara, Mohammed Kudus, Brennan Johnson, Gabriel Martinelli, and Alejandro Garnacho. If we relate this to Football Manager, I imagine these players as good inside forwards rather than inverted wingers.
Alexander Isak might be a surprise, but his score of 58 in passing most likely prevented him from joining the Triple Threat gang. I don't really find any of the inclusions that surprising - the players are typically wingers, but with some central forwards who drop deeper and carry or act as wingers at the centre of attack and carry often. This is the group that I had imagined Nicolas Jackson joining but, as stated earlier, due to him having such a high proficiency in shooting, the algorithm felt he belonged in shooting specialists instead.
It's interesting to note that there is no carrying or passing specialist. The lack of a passing specialist makes sense - passing, whilst important to attackers, is more commonly associated with playmakers who are usually deployed deeper on the pitch. This means that good passing attackers typically excel in the other threats too to warrant their deployment in forward positions.
The reason for the lack of a carrying specialist is less immediately apparent, but it's reasonable to conclude that carrying is more of a complementary skill. Passing and shooting can be valuable threats on their own, however carrying without good passing or shooting provides limited value - beating a man only to lose the ball adds little to the team's attack.
Nevertheless, I would anticipate that with more clusters, these additional specialist profiles would emerge, but at the cost of becoming overly complicated.
Evaluating Team’s Forward Lines
Putting this analysis into practice, we can take a look at some teams in Europe’s forward lines and see which clusters their players fall into. I have created 3-pronged radar charts that use the average scores in each threat for team’s forward lines, it should capture the balance and level of a team’s attack in terms of individual personnel.
Real Madrid
Kylian Mbappe - Triple Threat
Rodrygo - Triple Threat
Vinicius Jr - Triple Threat
Real Madrid's attack captures exactly what Ian Graham originally spoke of, a trio of forwards who excel in each of the 3 threats. The chart displays their proficiency across all three attributes, with carrying being their standout skill at 86.9. This bias towards carrying ability - notably higher than their shooting (77.1) and passing (78.0) - is likely due to the forwards nominally all being wingers.
Barcelona
Lamine Yamal - Triple Threat
Raphinha - Triple Threat
Robert Lewandowski - Shooting Specialist
Barcelona have the best attack in La Liga right now (11.2 more expected goals than Madrid in the same number of games!). Their front line has a different composition than Madrid's. Rather than 3 triple threat players, they have 2 triple threat wingers and a centre forward who excels primarily in shooting with Lewandowski. The chart above demonstrates this tactical difference clearly - their shooting score of 92.8 significantly outweighs their passing (72.4) and carrying (70.2), suggesting an alternative strategy of having 2 players who can cause danger in a variety of ways to free up their dangerous finisher.
Fulham
Raúl Jiménez - Shooting Specialist
Adama Traoré - Triple Threat
Harry Wilson - Creative Shooter
Alex Iwobi - Creative Dribbler
Fulham represent a different kind of balance. Firstly, their attack is not as good as either Barcelona's or Madrid's, and this is a result of player quality - as shown by their lower overall metrics (shooting 69.9, passing 65.3, carrying 60.5). But, the chart shows their attack maintains relative balance across all three threats. They have 4 players making up the distribution of minutes across their front line, each falling under a different cluster. The main centre forward, Jiménez, is a shooting specialist, supported by Wilson and Iwobi, who are both dangerous passers and excel in shooting and carrying respectively. Then you have Traoré who is a triple threat (perhaps surprisingly). This balanced rotation means that whoever pairs with Jiménez, the team maintains a spread of threat across all 3 areas.
Southampton
Tyler Dibling - Creative Dribbler
Cameron Archer - Shooting Specialist
Adam Armstrong - Bad Forward
Poor Southampton have one of the worst attacks across all of Europe. One could argue a lot of that is owed to Russell Martin's sterile tactics in the first half of the season, but their balance reveals additional problems. The chart shows not only significantly lower overall metrics (shooting 38.3, passing 22.6, carrying 44.7) compared to our previous examples, but also a severe imbalance between the three threats. Cameron Archer and Adam Armstrong share the centre forward role and occasionally play out wide, with Archer excelling in shooting whilst Armstrong struggles across all metrics. Meanwhile, the young, talented Tyler Dibling shows promise in passing and particularly in carrying. The glaring weakness in passing (22.6) directly relates to Graham's original point - without good passers "the ball does not get passed into dangerous locations."
It's worth noting that teams with superior overall attacking statistics may appear to have a more balanced forward line, but this could be an artifact of their general attacking quality - players in these attacks are naturally more likely to have good metrics in each threat.
Limitations
While the clustering approach provides valuable insights into player profiles, it's important to understand both its strengths and limitations. The reliability of these clusters depends on several factors. First, k-means clustering is sensitive to the initial random placement of cluster centres, meaning running the algorithm multiple times might produce slightly different groupings. To address this, I ran the algorithm several times to ensure stability in the results.
To validate the clusters' meaningfulness, I cross-referenced the groupings with my domain knowledge of player roles and playing styles. The resulting clusters align well with recognizable forward profiles we see in modern football, suggesting they're capturing real patterns rather than arbitrary divisions. That said, players on the borders between clusters might sometimes be classified differently if their performance metrics shift even slightly.
These clusters should be viewed as helpful guidelines rather than rigid categories - they're tools for understanding player profiles rather than definitive classifications. The fact that they align with observable playing styles while revealing some unexpected groupings suggests they're striking a good balance between confirming expert knowledge and uncovering new insights.
Conclusion
The emergence of triple threat forwards and the data-driven clustering of attacking profiles reveals how modern football has evolved beyond traditional specialist roles. While teams like Real Madrid showcase the power of having multiple triple threat forwards, Barcelona and Fulham demonstrate that balanced attacks can be achieved through different combinations of player profiles. The Southampton highlights how both tactical approaches and player profiles must align to create an effective attack.
What's particularly interesting is how this analysis validates Ian Graham's original observation about Liverpool's front three while showing that there isn't a single "correct" way to structure an attack. The key appears to be achieving balance - whether through multiple triple threats like Madrid, complementary specialists like Barcelona, or a diverse rotation like Fulham. The absence of pure carrying or passing specialists among forwards further supports this, suggesting that in modern football, these skills are most valuable when combined with other threats.
If you would like to review my methods, find my code here!
Really enjoyed this. You took your time on this analysis and it really shows in the quality of the results.