<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by FL Smith on Medium]]></title>
        <description><![CDATA[Stories by FL Smith on Medium]]></description>
        <link>https://medium.com/@sincerefc?source=rss-3e5a8f5ae41a------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*5zhYVDneANPRlBGv7guaJg.jpeg</url>
            <title>Stories by FL Smith on Medium</title>
            <link>https://medium.com/@sincerefc?source=rss-3e5a8f5ae41a------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 23:57:50 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@sincerefc/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Who was the most “clutch” player this season? Measuring the impact of every Premier League goal]]></title>
            <link>https://sincerefc.medium.com/who-was-the-most-clutch-this-season-measuring-the-impact-of-every-premier-league-goal-77b7dfef7644?source=rss-3e5a8f5ae41a------2</link>
            <guid isPermaLink="false">https://medium.com/p/77b7dfef7644</guid>
            <category><![CDATA[visualization]]></category>
            <category><![CDATA[sports-analytics]]></category>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[football]]></category>
            <category><![CDATA[soccer]]></category>
            <dc:creator><![CDATA[FL Smith]]></dc:creator>
            <pubDate>Fri, 13 Jun 2025 11:52:35 GMT</pubDate>
            <atom:updated>2025-06-14T21:57:14.382Z</atom:updated>
            <content:encoded><![CDATA[<p><em>All goals included in the data for this article are from open play.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*vJ802dNQkXgZ4u1t_BGhOg.jpeg" /></figure><h4>Introduction</h4><p>Suppose you are a recruitment analyst looking for attacking players in the Brazilian Série A. Knowing nothing about the clubs and players, you look at the league’s top goal scorers.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*4f2lnL_uViOL8x27XYUZlQ.png" /><figcaption>Brazilian Série A Top Goalscorers, 2024 season (source: Transfermarkt)</figcaption></figure><p>This list lacks context: you have no information about these players besides their goal tally, age, and position. To get that context, the typical approach would be to watch footage of each player on the list.</p><p>But because watching match footage takes time, using tools like a player radar (below) has become a common way get that context quickly, and understand HOW players play: whether they dribble a lot; whether they shoot more often from inside or outside the penalty area; and how frequently they link up with their teammates during buildup.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/512/1*yu_wBikeqrZi4-OGlc1fOw@2x.jpeg" /><figcaption>An example of a player radar from 2022–23 (source: Opta Analyst)</figcaption></figure><p>But what if, rather than their play style, we want to understand the ‘significance’ of their goals? <strong>How much did the goals they scored influence their team’s results?</strong> And, on the other hand,<strong> did they miss key chances in game states that mattered?</strong></p><p>Some players earn reputations for being poor in big matches, while others become known as “big match players”. But how do these reputations relate to their performance over the 38-game season?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8tTrPTFEDsJ-3XSqwTjUFw.png" /><figcaption>Didier Drogba (left) and Mohamed Salah (right) have each garnered reputations as playing their best (Drogba) or worst (Salah) in the biggest matches of the season.</figcaption></figure><p>Being near the top of the goalscoring list is never a bad thing. But as a scout, without watching all their matches, it’s hard to know how “good” 11 goals are. Were they 11 key goals against the league’s best sides? Or were 10 of these goals against teams fighting relegation, in moments where the outcome was already decided?</p><p>The goal of this is to show that information with a single data point per player, that captures <strong>1) how difficult the matches they score in are</strong>, and <strong>2) how critical each goal was to that match outcome.</strong></p><h3>Part 1: Creating a score for match difficulty</h3><p>If you’ve played Fantasy Premier League, you’re familiar with this idea.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/585/1*nCxuDCpRNB3tyqfu-TInyA@2x.jpeg" /></figure><p>In Fantasy Premier League, each player’s fixtures have a “difficulty rating” between 1 and 5 (it applies to the player’s team). FPL doesn’t release the formula showing how it is calculated, but we can derive a similar score using available data:</p><p><strong>Before a match, we have the following information about a football team:</strong></p><ul><li>Their <strong>win percentage</strong> in their recent games (I used the last seven)</li><li>Their <strong>cumulative goal difference </strong>across those games</li><li>Their performance in the competition throughout the season (<strong>how many points they’ve accrued thus far</strong>)</li><li><strong>Their “ELO” rating at the time of the match</strong> (a rating system on which all professional clubs are ranked — your score improves from beating clubs with a higher score than your own, and declines for the opposite reason)</li></ul><p><strong>Using those numbers, we calculate a “Difficulty score” for a game</strong>, by normalizing the four numbers above, and taking a weighted average (these weights have to change at the start of a season, since not enough matches have been played to calculate teams’ recent win percentage, goal difference, or cumulative points)</p><p><strong>Depending on the result, we assign each match a value between 1 and 7</strong> (1 being easiest, 7 the hardest). More scaling is necessary to ensure that games with difficulty 7 aren’t measured as being seven times as “hard” as games with difficulty 1; thus, the range of values used to calculate the index was actually between 14 and 23.</p><h3><strong>Part 2: How </strong>to capture the <strong>importance of a moment</strong></h3><p>How much does any moment in a football match “matter”?</p><p><a href="https://youtu.be/zUZNXSUxUVU?si=VkqQkAdZ-oVefkqa">A version of this idea is in a video by Route None</a>, which examines who scored the most “match-winning” goals (goals that gave their team a lead with less than ten minutes to go in the game). However, this newly created “leverage” metric captures the importance of a goal at any moment.</p><p>Followers of <strong>baseball</strong> will be familiar with this idea.</p><p>In baseball, the high-leverage situations are in the late innings, when the score is tight, and one or two runs for either team can change the result. Football is no different.<strong> When the margin is within one, the result is in the balance.</strong> As you approach the end of the game, the “impact” of a goal to level the score or take the lead is larger, as there is less time for the opponent to score.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*YJ7wQmMyKY2LhLQJuFjYUA.jpeg" /><figcaption>Chelsea fans celebrate after taking a late lead against Manchester United in the FA Cup Final in 2007.</figcaption></figure><p>If the score margin is wide, the opposite is true. If the goal margin is three and the game is in the 87th minute, the outcome is virtually settled, and thus, goals scored by either team are not that “influential” on the outcome.</p><p>The data I used to measure how likely a team is to win based on the score and minute comes from a 2019 <a href="https://www.arcticicehockey.com/leverage-and-timely-goals-in-soccer-1550310291/">article</a> by Gabriel Desjardins. The likelihoods of winning, drawing, and losing were transformed into “expected points”, given that wins are worth 3 and draws are worth 1.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/585/1*4Djm53uwxqSTXUWc75FVlw@2x.jpeg" /><figcaption>Expected points depending on the score and minute of the game (source: ArcticIceHockey, 2019)</figcaption></figure><p>The differences between these numbers at each minute are the <em>differences</em> in expected points when scoring a goal.<strong> This is how I define the “leverage” of a goal — it is the ‘expected points’ you add by scoring.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/455/1*zUT8OB6sMz5jOuGR5Xh1oQ.jpeg" /></figure><p>Because there are not enough matches with a margin of three after less than 15 minutes, I had to approximate the value of several situations (shown below) so that every possible goal had a value.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/846/1*JTswpCsToJnEdwLPo9npQg.png" /><figcaption>The minute of a football match (X) versus the expected points (Y), based on whether a team is ahead, level or behind.</figcaption></figure><p>The values, which range between 0 and around 1.5 (points), are then categorized into one of five buckets.</p><ul><li>Bucket 1 (very low): 0 to 0.1 (0.1 = the 10th quantile)</li><li>Bucket 2 (low): 0.1 to 0.33</li><li>Bucket 3 (“typical” ): 0.33 to 0.66</li><li>Bucket 4 (high): 0.66 to 0.9</li><li>Bucket 5 (very high): 0.9 to 1 (1 = the 100th quantile — the biggest value)</li></ul><p>Some of the highest leverage goals from this season so far: <strong>Rodrigo Muniz </strong>vs <strong>Chelsea</strong> (2–1); <strong>João Pedro </strong>vs<strong> Manchester United</strong> (2–1, 90&#39;); <strong>Amad </strong>vs<strong> Manchester City</strong> (1–2).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*J_lQmpEzCymKWwF48vjgKg.jpeg" /><figcaption>Amad Diallo’s 90th minute winner against Manchester City was tied for the highest-leverage goal of the 2024–25 season (source: Getty Images)</figcaption></figure><p>All of these goals (and others) are tied for 1st place, as they gave their team the lead in the 90th minute or later.</p><h3>Looking at the 2024–25 Season’s top goalscorers</h3><p>After 29 match days (<em>through Nottingham Forest vs. Aston Villa on April 5), </em>the Premier League goalscoring chart looked like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/500/1*CFmb7WriCjFskXCNcWsQTQ.png" /><figcaption>Non-Penalty goals for players in the 2024–25 Premier League through April 6. (source: FBref)</figcaption></figure><p>As discussed, knowing the impact of each player’s goals is difficult without having watched every game.</p><h4>Here’s how it looks after we add the “leverage” metrics.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*yk4vyG3xjfwZTeBYPpfs2w.png" /><figcaption>Low-leverage: doesn’t influence the result very much; High-leverage: significant impact on the result; Normal leverage: everything else.</figcaption></figure><p>Despite being poor in tournament finals, <strong>Mohamed Salah</strong> was tied with <strong>Jean-Philippe Mateta</strong> for the division high with four high-leverage goals. These were:</p><ul><li>@ Arsenal, Oct. 27, 2024, to make it 2–2 (81&#39;)</li><li>vs. Brighton, Nov. 2, 2024 – to make it 2–1 (72&#39;)</li><li>@ Newcastle, Dec. 4, 2024 — to make it 2–3<strong>*</strong> (83&#39;)</li><li>@ Everton, Feb. 12, 2025 — to make it 1–2 (73&#39;)</li></ul><p>It could easily be five for <strong>Salah</strong>: he scored to make it 2–2 in the game against Newcastle, but it fell slightly outside the criteria for ‘high-leverage’.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_kG7nnyUCTmd4-Z0WBehpQ.jpeg" /><figcaption>Mohamed Salah celebrates at St. James’ Park after scoring for Liverpool (source: Sky UK)</figcaption></figure><p>In contrast, <strong>Cole Palmer</strong> scored nine low-leverage goals this season. While he did score several important penalties, the most significant <em>non-</em>penalty goal he scored came against Newcastle on Oct. 27, 2024, making it 2–1 to Chelsea after 47 minutes.</p><h4>What if we show the difficulty metrics?</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*J2_U80E_WlyxIidzyz6G-w.png" /><figcaption>“Easier match goals”: goals in games with a difficulty rating of 1–2; “Harder match goals”: goals in games with a difficulty rating of 5–7; “Regular Match Goals”: everything else</figcaption></figure><p><strong>Jean-Philippe Mateta</strong> scored the most goals in “difficult” matches with seven. However, only two of them came against the top seven clubs (Chelsea and Aston Villa): the rest came against surging mid-table teams, like against Everton, on February 15 (the Toffees had won five of their previous seven matches entering the game). In terms of scoring against the actual top seven,<strong> Alexander Isak</strong> was the leader with six.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*MD_8CUjdlWvsmFXtI76VpA.jpeg" /><figcaption>Alexander Isak scored against five of Newcastle’s top six rivals this season, all apart from Manchester City (source: Getty Images)</figcaption></figure><h3>The Premier League’s goal scorers, in a Scattergram</h3><h4>The 2024–25 Difficulty Index</h4><p>Rather than counting the number of goals scored, the “difficulty index” shows the average strength of the opponents each player scored against. Results are shown below, for players who scored five goals or more.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fXoOd2Ed0_MEXs6ARCHCEw.png" /></figure><p>Among players with ten or more goals, the lowest average score is <strong>Jørgen Strand Larsen</strong>, who only scored two (out of 11) goals against sides in the top ten. As mentioned previously, <strong>Alexander Isak </strong>and <strong>Jean-Philippe</strong> <strong>Mateta</strong> scored against the toughest opponents on average. <strong>Liam Delap</strong> also finished with an impressive score from his ten goals (through 4/6).</p><h4>The 2024–25 Leverage Index</h4><p>The “leverage index” captures how important the goals each player scores are on average (left to right). Again, results are shown for players who scored five goals or more.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Y4I-UEzq53QQx4f4pnWAPQ.png" /></figure><p>Among players with at least 10 goals, <strong>Ollie Watkins</strong> finished with the highest score (one low-leverage goal out of 11). <strong>Cole Palmer</strong> ended with the lowest score (zero high-leverage goals in 11). Of the top three scorers, <strong>Mohamed Salah </strong>and<strong> Alexander Isak </strong>ended up with around the same score, with <strong>Erling Haaland </strong>slightly behind.</p><h4>Showing difficulty and leverage at the same time</h4><p>This produces a scattergram showing both indices on the same plot, for all players with at least 5 goals.</p><p>Before you see it, here are some notes about how to interpret the diagram:</p><ul><li><strong>Landing in the top right corner of this diagram implies you scored all your goals to win the match against the league’s best teams. Landing in the bottom left suggests you scored all your goals in meaningless situations against the worst teams.</strong></li><li>Being far to the right of the vertical or horizontal line is good, <em>but staying there at high volume is difficult</em>. If you score a lot, being closer to the middle is to be expected: the best players score in all game states, against all opponents. <em>However, you are ‘not clutch’ if you’re far into the bottom left quadrant.</em></li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*5Njtm9n99UEGbrDJ3OfHAQ.png" /><figcaption>Premier League 2024–25 (data up to April 5, 2025). X = leverage index, Y = difficulty index. Point size = # goals scored.</figcaption></figure><p>What defines a “clutch” player? <strong>Being towards the top right.</strong></p><p>Players like <strong>Rodrigo Muniz</strong>(7) and <strong>Harry Wilson</strong>(5) didn’t score that many goals, but they scored when it mattered most. <strong>Rodrigo</strong> <strong>Muniz</strong> scored against nearly all of the top teams (Liverpool, Chelsea, Arsenal, Man City, and Newcastle) – but virtually nobody else.</p><p>Some of the players who helped “mid-table” sides contend for Europe can be found here (Fulham: <strong>Raúl Jimenez, Rodrigo Muniz, Harry Wilson</strong>; Brighton: <strong>João Pedro, Danny Welbeck; </strong>Nottingham Forest: <strong>Callum Hudson-Odoi, Anthony Elanga</strong>; Bournemouth<strong>: Justin Kluivert, Evanilson). Beto’s</strong> mid-season purple patch for Everton was a series of important goals.</p><p><strong>Arsenal’s</strong> lack of a clear goalscorer is evident: they are the only team in the top 7 without even a medium-sized dot – their most “clutch” goal scorer was their defensive unit, which tallied eight goals, all of them in close games.</p><p>The more worrying datapoints are those in the bottom left quadrant. Being there suggests that you did not perform your best when the lights were bright. Notable players in this group include <strong>Marcus Rashford</strong>, <strong>Phil Foden</strong>, <strong>Noni Madueke,</strong> and <strong>Cole Palmer</strong>. Some mid-table players in this quadrant also “flattered to deceive”: while <strong>Iliman Ndiaye(7)</strong> and <strong>Kevin Schade (7) </strong>each had career-best seasons in 2024–25, they found that form in periods of matches with less urgency, against the weaker teams in the league.</p><h4>Adding the rest of Europe’s top five leagues</h4><p>Combining the Premier League data with similar data from the other top leagues, we created a similar scattergram for all players <strong>with at least ten goals</strong> in their domestic competition.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*V52mMaa1MzWh3DTMJnqYcQ.png" /><figcaption>Top 5 leagues in Europe, 2024–25 (data up to April 5, 2025). X = leverage index, Y = difficulty index. Point size = # goals scored. Players from the Premier League appear in different places because the data involves all 5 leagues</figcaption></figure><p>Of all players in the top 5 leagues, <strong>Ayoze Pérez </strong>scored when it mattered most.</p><p>Among players with at least 15 goals, the player whose goals were (on average) the most important was Kylian Mbappé. Real Madrid supporters will dispute this<strong> </strong>— they<strong> </strong>won’t have forgotten his sub-par showing in Madrid’s 0–4 defeat to FC Barcelona in September 2024.</p><p>Some of the names in this summer’s transfer rumor mill can also be found on the right side. <strong>Benjamin Šeško</strong>, <strong>Emanuel Emegha </strong>and<strong> Liam Delap </strong>have all been linked with moves to bigger clubs, and Frankfurt’s <strong>Hugo Ekitike </strong>didn’t do too badly, either.</p><p>Some of Europe’s biggest “flat-track bullies” can be seen in the bottom left corner. While <strong>Serhou Guirassy</strong> impressed in this season’s Champions League for Dortmund, his Bundesliga campaign left a lot to be desired in terms of match-winning goals. <strong>Mateo Retegui</strong>(22) led the Serie A in goals this season, but he scored only <em>one</em> high-leverage goal in his appearances through April 6. And finally, nobody had “emptier” stats than<strong> Harry Kane</strong>, who ended up in an area entirely on his own — although he won’t care, as he won his first major trophy of his career.</p><h3>Part III: missed chances</h3><p>The previous scattergrams showed the difficulty and importance of players’ goals. The following indices show when and against whom players missed big chances.</p><p>What defines a “good chance”? While some sites use an xG of 0.38, and others use 0.3, I defined them as shots with an xG of 0.28 or greater (0.28 was the cutoff point above which the ratio of shots to goals went up significantly).</p><h4>The Difficulty Index for missed chances</h4><p>Results are shown below for all players with at least 5 missed chances (players further to the <strong>right</strong> missed chances in more difficult games). Each point’s size isn’t that player’s total number of big chances missed, but chances missed in general (shots with at least 0.08 xG that didn’t end up in the net).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*4fsRSOjODxbybKN5FT3Gmg.png" /></figure><p>The Premier League league leader in big chances missed was <strong>Ollie Watkins (</strong>13). Several of the names from the top of the previous difficulty index are up here — missing are <strong>Bryan Mbeumo</strong>, <strong>Jørgen Strand Larsen</strong>, <strong>Chris Wood, </strong>and<strong> Matheus Cunha, </strong>who<strong> </strong>each missed four or fewer such opportunities<strong> (Cunha </strong>only missed one).</p><h4>Leverage Index</h4><p>Below, we see the average importance of the missed chances for all players with 5 or more.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*39kLJ1JGMq1XEgDV-H_UHA.png" /></figure><p><strong>Ollie Watkins, </strong>who<strong> </strong>missed many of his big chances in less important game moments, scored far above average in terms of the leverage of his <em>goals, </em>suggesting that while he wasn’t the most prolific, he came up big for Aston Villa when it mattered, albeit against some weaker teams.</p><p>The context these indexes add to <strong>Nicolas Jackson’s</strong> nine big misses makes it clear why he continues to garner criticism from Chelsea fans.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gx1zDncm2eDQW1x9l2KNBg.avif" /></figure><h4>Does the good outweigh the bad? A ‘net analysis’</h4><p>As shown, every player accumulates a “score” for both their goals and their misses. But what if we put them together to examine how they each net out?</p><p>In the next scattergram, we show the <strong>difference between the score accumulated by scoring and the score accumulated by missing. </strong>Here’s an example using <strong>Alexander Isak</strong>.</p><p>Through April 6,<strong> Isak</strong> scored 18 goals:</p><ul><li><strong><em>Total difficulty score</em></strong><em> (as described, a score between 0 and 1 for each match, where 0.01 represents 2007 Derby County and 0.99 represents 2009 FC Barcelona) = </em><strong><em>8.93</em></strong></li><li><strong><em>Total leverage score</em></strong><em> (expected points added) = </em><strong><em>9.94</em></strong></li></ul><p>He missed 9 big chances:</p><ul><li><strong><em>Total difficulty score</em></strong><em> = </em><strong><em>2.44</em></strong></li><li><strong><em>Total leverage score </em></strong><em>= </em><strong><em>4.14</em></strong></li></ul><p>Thus, Isak accumulated a difference of <strong>6.49</strong> in difficulty (Y) and<strong> 5.8 </strong>(X)<strong> </strong>in terms of leverage.</p><p>Every forward with at least 10 missed chances and/or goals is shown below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/844/1*0CSEQsPNOnpjcQR7ZIxLEA.png" /><figcaption>Difference in players’ cumulative leverage &amp; difficulty scores accumulated through either scoring or missing big chances. X = difference in leverage, Y = difference in difficulty score</figcaption></figure><p>What information is being captured here?</p><p>Naturally, players who score more than they miss end up towards the top right. However, it’s not that simple. <strong>Erling Haaland, Bryan Mbeumo, and Jørgen Strand Larsen</strong> all had seven more goals than missed big chances, but each player ends up in a different place on the graph.</p><p>In the bottom left, only two players ended with negative scores on both metrics: <strong>Beto </strong>and <strong>Nicolas Jackson </strong>were both empirically the least clutch forwards in the Premier League this season.</p><p><strong>Mohamed Salah, Alexander Isak</strong>, <strong>Chris Wood,</strong> and <strong>Matheus Cunha</strong> were all clutch<strong>, </strong>each<strong> </strong>having a slight edge over <strong>Erling Haaland</strong>.</p><p>To make the diagram easier to read, we then divided forwards into 3 tiers:</p><ul><li>those who scored at most 3 more times than they missed big chances (Tier 3)</li><li>those who scored at least 7 goals more than they missed (Tier 1)</li><li>and everyone in between (Tier 2)</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*L5DeUy7dnYs0G6D6tNN7iA.png" /></figure><p>Among players in tier 3, <strong>Rodrigo Muniz’s</strong> goals were by far the most impactful relative to his misses. For those in tier 2, the results are clustered together, but <strong>Antoine Semenyo</strong> missed several more high-leverage chances than he scored important goals.</p><p>While it’s difficult to keep up a “conversion rate” of shots to goals as you get more opportunities, the four forwards on “non-elite” teams in tier 1 all had statistically strong seasons, but particularly <strong>Matheus Cunha</strong>.</p><h3>Numbers don’t lie — or do they?</h3><p>If you looked at these graphics and said, “This metric isn’t fair to players on good teams”, <strong>you would be right</strong>. Because you add less win probability by scoring early in matches, or scoring while already ahead, players whose teams win a lot don’t spend enough time trailing to accumulate many points.</p><p>To account for this, we split the players into 3 groups:</p><ul><li>Players on teams who won &gt;50% of games (Liverpool → Nott’m Forest)</li><li>Players on teams who won &lt;20% of games (Southampton &amp; Leicester City)</li><li>Everyone else (the teams from 8th to 18th in the final table)</li></ul><h4>The “goals” scattergram, divided into 3 groups</h4><p>Here are the players on teams that won &gt;50% of their games.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mkgXsOogVQ5u363X7a8bmg.png" /></figure><p>Next, the players on teams that won less than 20% of their games (there are only two).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-aDUugyXxOOCQxcu2zq68w.png" /></figure><p>Finally, here’s everyone else.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*dVRThzYWtaMoVU6fj_BGpw.png" /></figure><h4>The “net results” scattergram, divided into 3 groups</h4><p>Here are the players on teams that won &gt;50% of their games.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/844/1*K0fy7RaJskFW9_N5JQWDRQ.png" /></figure><p>Most fans would agree that this graphic separates the forwards who had good seasons from those who had poorer seasons.</p><p>Now, the players on teams who won less than 20% of their games (there is only one).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/844/1*nV52mvIAdQ5OvIDqGU1p_Q.png" /></figure><p><strong>Delap’s</strong> place on the graph — especially relative to Chelsea’s other forwards in the previous diagram — suggests they have done some smart business.</p><p>And finally, here’s everyone else.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/844/1*xWduXrHiEHMY_xxgnD5paw.png" /></figure><p><strong>Matheus Cunha’s</strong> place on this chart speaks for itself.</p><h3>Conclusion: Why <strong>do these numbers matter?</strong></h3><p>First, they provide clarity to counting stats.</p><p>Each team will have a set of key goals that earn them points in matches, and these statistics help clarify who does the heavy lifting for these teams.</p><p>As stated in the introduction, this helps us understand a player’s ability to execute in the important moments. Who’s accumulating their ‘counting stats’ in important games, and important parts of the game?</p><p>This can be adapted to other statistics, like successful crosses, or successful dribbles. Wingers receive the ball at a time and place in which it’s sensible to take on the full back a few dozen times per game. How many of those times do they complete actions that advance their team’s chance of scoring? And do they do it when it matters most?</p><p><strong>Secondly, it can help to either confirm or dispel narratives around players’ reputations.</strong></p><p>As mentioned previously, while <strong>Mohamed Salah </strong>does not have a good goal-scoring record in finals, he came up with the goods in key moments as much as any other Premier League player this season.</p><p>Finally, a neat use of this data is putting matches from different seasons or competitions onto the same scale. Following Arsenal’s 3–0 victory against Real Madrid in the 2024–25 Champions League, pundits compared <strong>Declan Rice’s</strong> pair of free kicks to other times players scored two free kick goals in a Champions League match. Using this data, we can compare each instance of a player scoring 2 free kicks and get an empirical answer as to which two were the most “important”.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=77b7dfef7644" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Measuring the Best Hold-Up Players in Football]]></title>
            <link>https://sincerefc.medium.com/measuring-the-best-hold-up-players-in-football-cf41c70cd603?source=rss-3e5a8f5ae41a------2</link>
            <guid isPermaLink="false">https://medium.com/p/cf41c70cd603</guid>
            <category><![CDATA[analytics]]></category>
            <category><![CDATA[sports-analytics]]></category>
            <category><![CDATA[football]]></category>
            <category><![CDATA[soccer]]></category>
            <dc:creator><![CDATA[FL Smith]]></dc:creator>
            <pubDate>Wed, 08 Jan 2025 19:21:02 GMT</pubDate>
            <atom:updated>2025-01-11T15:42:15.503Z</atom:updated>
            <content:encoded><![CDATA[<h3>Measuring The Best Hold-Up Players in Football</h3><p>We talk about ‘hold up play’ — why don’t we count it?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/968/1*0YMttoWNe2Vvn_YcVSoPRA.png" /><figcaption>Álvaro Morata and Mikel Oyarzabal exchange gestures during a substitution in Spain’s match against England in the Final of Euro 2024. (Source: Getty Images)</figcaption></figure><p>With 22 minutes to play in the final of Euro 2024, Luis de la Fuente replaced captain Álvaro Morata with forward Mikel Oyarzabal: a substitution that helped Spain win their fourth international tournament since 2008.</p><p>About 10 minutes after Oyarzabal came on the pitch, I told a friend sitting beside me, <em>“Oyarzabal for Morata was a bad switch. Oyarzabal has no idea how to play effective central.”</em></p><p>Less than 5 minutes later, Oyarzabal scored the winning goal.</p><p>After watching the goal go in, I felt like a fool– but reflecting on the moment clarified an idea I’d been trying to work out.</p><h3>Player Roles in Football’s Tactical Evolution</h3><p>Today, football teams are set up so players have a defined role. Although there are many chaotic and random moments in a football match, the game has become less about free-flowing action and more about executing tasks tied to one’s position.</p><p>The tactics used by elite teams are so homogenous that there is a finite set of ways teams create attacking chances that can be categorized. In a recent data analysis of European matches, <a href="https://medium.com/@guillemmiralles1/ai-driven-soccer-formations-analysis-optimizing-team-formations-through-deep-learning-part-ii-1e4f60a33375">more than 90% of teams’ lineups were listed as either <em>4–4–2</em>, <em>4–5–1,</em> or <em>4–3–3</em></a><em>.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/937/1*uQogND_PpepS7naskM0P1w.jpeg" /><figcaption>Formation distribution for home and away teams in European Soccer Database matches (source: <a href="https://medium.com/@guillemmiralles1/ai-driven-soccer-formations-analysis-optimizing-team-formations-through-deep-learning-part-ii-1e4f60a33375">Guillem Miralles</a>)</figcaption></figure><p>Given a finite set of ways to attack an opposition goal, we can look at types of attacking play akin to fighting styles in mixed martial arts. <strong>Teams adapt their approach throughout games based on tactics, personnel, opponent, and game state</strong>. Graphics like those below help illustrate the contrast between teams.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/720/1*TL_N5i_J2B7DpLW59KwRsA.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/720/1*farhkNfywAhfx0KLaAQ4Hw.jpeg" /><figcaption>Premier League teams compared in 2020, using several of the most common ‘advanced’ attacking metrics (Source: Sky Sports, via Opta)</figcaption></figure><p>In 2020 (and even more so in 2025) , a significant number of attacks came from set pieces (not shown above), and “<a href="https://www.premierleague.com/news/4028988">high turnovers</a><strong>”</strong>,<strong> </strong>but many of the others are from buildup play, which comes in various degrees of speed and intricacy.</p><p>Football is a results business, and as manager Sam Allardyce famously reminded us, you don’t get extra points for having more 10+ pass sequences than the opposition (playing “<a href="https://youtu.be/NN2rngoO1p8?si=3nQxOS3daOJFenye">tippy-tappy football</a>”). The objective is to win– how you win is beside the point.</p><p>Similar to how teams can win multiple ways, the objective of a striker is to help score goals - but there are skills involved in playing striker besides scoring that players perform to varying degrees: they pass, press from the front, run in behind, <strong>hold up play</strong>, and act as a decoy to create space.</p><h3>What defines ‘hold up play’?</h3><p>This example from OneFootball is classical “hold-up play”, happening high up the pitch: Chelsea’s midfielders try to pass to Olivier Giroud, who, with his back to goal, attempts to control the ball and find a teammate running towards the opposition goal.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*r5WDhNeOu1giY2yt5ci9mw.png" /><figcaption>An example of ‘hold-up play’ from Chelsea vs Leicester in 2021 (source: OneFootball)</figcaption></figure><p>Giroud, regarded by many as the best target man in the modern era, describes hold-up play <a href="https://www.getfootballnewsfrance.com/2018/a-self-analysis-by-olivier-giroud-i-am-a-bit-of-an-old-school-player/">in a 2018 interview</a> as the ability to “[control the ball and] lay it off…or to keep possession high up the pitch”.</p><h4>“Lay it off”</h4><p>Hold up play can happen closer to or further away from goal. In the following clip from the Euro 2024 final, Morata holds it up to bring Spain out of their defensive half:</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fplayer.vimeo.com%2Fvideo%2F1042048679%3Fapp_id%3D122963&amp;dntp=1&amp;display_name=Vimeo&amp;url=https%3A%2F%2Fvimeo.com%2F1042048679&amp;image=https%3A%2F%2Fi.vimeocdn.com%2Fvideo%2F1964942747-b1230a10361f577e70cfce55bb9b85a7e856358b67983991d075f54ba1d761ca-d_1280&amp;type=text%2Fhtml&amp;schema=vimeo" width="1802" height="1080" frameborder="0" scrolling="no"><a href="https://medium.com/media/453dbd913692d0dd0970ad8e50d3e37a/href">https://medium.com/media/453dbd913692d0dd0970ad8e50d3e37a/href</a></iframe><p>In the 2000s Premier League, many teams played with “target men”, and thus may not have occured to anyone to quantify hold-up play, since it was so much more common. Many fans associate “hold-up play” with big, physical players like Olivier Giroud — but there are many modern examples of less physically imposing players holding up the ball.</p><p>In the next clip, Lautaro Martínez receives the ball in a tight area while facing away from goal. He controls it before passing to Hakan Calhanoglu, who sets Inter on a break. This looks similar to the Giroud example, despite the action being performed by a 5&#39;9&quot;, 72 kg player.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*WHjnTjmNdov24jrubgTAbw.gif" /><figcaption>Lautaro Martinez vs. Lazio (source: CBS Sports)</figcaption></figure><h4>“Keep possession high up the pitch”</h4><p>As Giroud explains, “hold-up play” also means simply holding onto possession, where the player waits for teammates to run beyond them, as shown below:</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fplayer.vimeo.com%2Fvideo%2F1042049874%3Fapp_id%3D122963&amp;dntp=1&amp;display_name=Vimeo&amp;url=https%3A%2F%2Fvimeo.com%2F1042049874&amp;image=https%3A%2F%2Fi.vimeocdn.com%2Fvideo%2F1964944105-ef5b3c7d2616dca3bc876d2f9f5a4e2aa43e638a07f40b4582fa0982903b2c77-d_1280&amp;type=text%2Fhtml&amp;schema=vimeo" width="1802" height="1080" frameborder="0" scrolling="no"><a href="https://medium.com/media/2894bba71ae144b24984aba20bb35a80/href">https://medium.com/media/2894bba71ae144b24984aba20bb35a80/href</a></iframe><p>This type of hold-up play less high up the field is common against a press: teams try to either bypass the opposition by going direct, or ‘play through’ the press and build up quickly.</p><p>Playing through a press is difficult and presents significant risk. High turnovers — ball recoveries from pressing — <a href="https://www.premierleague.com/news/4028988">are more common than ever</a>, driven by the rise of high pressing and other teams’ attempts to play through it. As shown in the clips, playing direct can be an effective way to evade the press–which is facilitated by players who can hold up the ball.</p><p><strong>Holding play up successfully against the press takes the game from a state of danger for your team to one of danger for the other team.</strong> Because these actions are both impactful and less common than they used to be, it has become more important to understand who does it well, and at high volume, especially from the perspective of a team who expects to face this situation regularly.</p><h3>Back to the Euro 2024 Final</h3><p>Spain created chances through Nico Williams and Lamine Yamal throughout the game, with Morata connecting the play by dropping in to hold the ball up. This aspect of Morata’s game is part of why he has spent his career playing at elite European clubs despite having unexceptional goalscoring record.</p><p>With Oyarzabal in his place, Spain struggled to build through the center of the pitch, and began using the flanks to break England’s press, having lost the ability to play a direct ball up the field.</p><p>These freeze-frames from after the substitution show how when right-sided defender Robin Le Normand has the ball (as in the video before), Oyarzabal’s movement is more similar to that of the other front players, on the last line of England defenders.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*AH3SRae-GSSiMbvhHKd-Kw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*q78NPHz3AKnJcB0clQAuxA.png" /></figure><p>With the ball in midfield, all three of Spain’s forwards attempt to stretch the England back line.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*XHb4Rwp6hM5_0AhoAO38vw.png" /></figure><h4>Comparing Morata to Oyarzabal</h4><p>Fans familiar with Mikel Oyarzabal know he is less of physical player than Morata, being more of a technical attacker. In this game, his role was to stretch the defense, rather than to drop deep and hold up the ball. <strong>It’s not that Oyarzabal is “bad” at hold-up play — he just doesn’t do it. </strong>Other players in his team typically do that job (<em>Morata for Spain, and players like Willian José, Alexander Sørloth or André Silva with Real Sociedad</em>).</p><p>Spain’s winning goal comes from a quick sequence of ground passes, shown below.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fplayer.vimeo.com%2Fvideo%2F1042085948%3Fapp_id%3D122963&amp;dntp=1&amp;display_name=Vimeo&amp;url=https%3A%2F%2Fvimeo.com%2F1042085948&amp;image=https%3A%2F%2Fi.vimeocdn.com%2Fvideo%2F1964988761-5c212437206e78f7c4e501219488efe3cc1e59da7b6dcdc903f0843a461f60d2-d_1280&amp;type=text%2Fhtml&amp;schema=vimeo" width="1802" height="1080" frameborder="0" scrolling="no"><a href="https://medium.com/media/ed9fcb9d6a1b96d8ef1ac54b39c7f522/href">https://medium.com/media/ed9fcb9d6a1b96d8ef1ac54b39c7f522/href</a></iframe><p>Oyarzabal’s quality is on display: as England’s defense retreats, he has the skill to find Marc Cucurella and score first-time.</p><p><strong>Notice where the action begins</strong>: Spain have already bypassed the England press, and are onto England’s back line. In this scenario, the part of the game Morata is more effective than him at is not required.</p><p>As a forward, Oyarzabal is both more effective in terms of technical skills, and at the same time<strong> </strong>less effective as a link between defense and attack. Comparing him and Morata on paper using commonly available stats misses the context of their respective roles. Measuring hold-up play helps provide context when comparing players, the kind of which is possible with video analysis.</p><h3>Turning Hold-Up Play into a Tangible Metric</h3><p><a href="https://www.getfootballnewsfrance.com/2018/a-self-analysis-by-olivier-giroud-i-am-a-bit-of-an-old-school-player/">As Giroud </a>says, “hold-up play” means controlling an upfield ball “to lay it off for a teammate or to keep possession high up the pitch”<strong>.</strong></p><p>This action can be broken into three parts: <strong>the pass </strong>to the player who holds it up<strong>; </strong>the<strong> control </strong>of the pass<strong>; </strong>and the<strong> subsequent action (or lack thereof) </strong>the player makes<strong> </strong>after controlling it<strong>.</strong></p><h4>Part 1: the pass to the player who holds it up</h4><p>What defines an “upfield ball”?</p><p>The ball is either hit along the ground, typically from distance (the clip of Morata is about a 35-yard pass), or the ball is not along the ground and requires controlling or winning via a duel<strong> (</strong>shown in the Bowen and Martínez examples). In any case, <strong>when the player controls the ball, they are not running towards the opposition goal.</strong></p><p>Teams being pressed in their own half are periodically forced into playing this type of pass (sometimes it’s just a hopeful kick), and yet many of the more technical forward players don’t possess the ability to hold up the ball at all.</p><p>These passes are different from the “progressive” passes captured in the graphic below<strong>, </strong>which are typically<strong> </strong>to a player who receives it and moves toward the opposition goal.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Ezg5J_LfLGE9pPfye9lvaA.png" /><figcaption>Progressive reception leaders by zone, Premier League 2023–24 (source: New York Times). “Hold-up” play typically occurs in the middle three-fifths of the pitch, in the area outlined in black.</figcaption></figure><p>When players receive these passes, their teams are often in a ‘counter attack’ scenario, or advancing forward against an unset defense. The following clips shows a ‘progressive reception’ by Morata in the same game. Both passes to Oyarzabal in the sequence of his goal are examples as well.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fplayer.vimeo.com%2Fvideo%2F1042057664%3Fapp_id%3D122963&amp;dntp=1&amp;display_name=Vimeo&amp;url=https%3A%2F%2Fvimeo.com%2F1042057664&amp;image=https%3A%2F%2Fi.vimeocdn.com%2Fvideo%2F1964953614-86f71846e455cb48a0233189ab9c65593ddfda8b1aa8858fcf8c9b2e07ff0e7a-d_1280&amp;type=text%2Fhtml&amp;schema=vimeo" width="1802" height="1080" frameborder="0" scrolling="no"><a href="https://medium.com/media/a49c16e860f2dc9e78184e4a594ea047/href">https://medium.com/media/a49c16e860f2dc9e78184e4a594ea047/href</a></iframe><p>In these examples, the player receiving the ball is moving toward the opposition goal, either as they receive the pass, or directly afterwards.</p><h4>Part 2: controlling the ball</h4><p>The start of a successful action is either an upfield pass that is “<a href="https://www.fifatrainingcentre.com/en/resources-tools/football-language/in-contest/in-contest.php">in contest</a>”, <strong>such that a player must receive it with their back to goal as a defender competes with them.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/634/1*ElS0Cisr16g2TL_oTCBrLA.jpeg" /><figcaption>An example of ‘hold-up play’ from Olivier Giroudin 2015 (source: Getty Images)</figcaption></figure><p>Because of their body position, the player typically can’t immediately drive the ball towards the opposition goal. Winning a duel, “holding off” a defender, or receiving the ball with back to goal on the run (as shown by Jarrod Bowen) is the first part of successful hold-up play.</p><p>When a winger decides not to take on his man and waits for teammates to advance beyond them, that is hold up play ONLY if the winger had to actually compete to receive the ball.</p><p>If the player receives a pass while at an angle at which they are immediately able to dribble towards the opposition goal, <strong>that is a</strong> <strong>progressive reception and not an example of hold-up play, </strong>as the player doesn’t have to work to maintain possession.</p><p>The following clip shows an example of such a reception by a winger.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fplayer.vimeo.com%2Fvideo%2F1043967725%3Fapp_id%3D122963&amp;dntp=1&amp;display_name=Vimeo&amp;url=https%3A%2F%2Fvimeo.com%2F1043967725&amp;image=https%3A%2F%2Fi.vimeocdn.com%2Fvideo%2F1967274674-c056d5b9b72f8a3fabe1a2bd8a72c3a71ced47756a888e069219eba0ee6c7c75-d_1280&amp;type=text%2Fhtml&amp;schema=vimeo" width="1802" height="1080" frameborder="0" scrolling="no"><a href="https://medium.com/media/cfd2a60006878157ffb8e8071199eb13/href">https://medium.com/media/cfd2a60006878157ffb8e8071199eb13/href</a></iframe><h4>Part 3: the subsequent action</h4><p>Once a player has controlled the upfield pass, they either “lay it off” immediately, bring the ball down and dribble as Bowen does, win a free kick, or pass to a teammate (Lautaro Martinez), <strong>in each case retaining possession. </strong>Possession must remain with their team after the action without interruption from the opposition, or contested possession (in other words, the player who holds it up can’t be the one to lose the ball).</p><p>Hold-up play doesn’t necessarily require more than 1 touch. Here, Morata plays a through ball while holding it up in one move, one of the most advanced skills for a forward at the top level.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fplayer.vimeo.com%2Fvideo%2F1043974636%3Fapp_id%3D122963&amp;dntp=1&amp;display_name=Vimeo&amp;url=https%3A%2F%2Fvimeo.com%2F1043974636&amp;image=https%3A%2F%2Fi.vimeocdn.com%2Fvideo%2F1967283233-b7f5c6a23634f3e3e4ebe8f3b8cedb4d361ed721c75cb61a63ba0db44895746e-d_1280&amp;type=text%2Fhtml&amp;schema=vimeo" width="1802" height="1080" frameborder="0" scrolling="no"><a href="https://medium.com/media/585df4a1a23c98d7aba246c61d984f74/href">https://medium.com/media/585df4a1a23c98d7aba246c61d984f74/href</a></iframe><p>Hold-up play is not the same as winning an aerial duel, often called a “knock-down, or “knock-on”, although knock-downs are related to this action.</p><p>Possession must remain with the team for at least 3 more seconds. The pass must reach a teammate facing toward the opposition goal. If a player holds it up without passing, at least 2 of the team’s most advanced players must be Y meters closer to goal from the start to the end of the action (<em>see Bowen vs Arsenal</em>).</p><h3>Conclusion: the “hold-up” Statistic</h3><p><strong>I propose to count the following actions as “hold-up play”:</strong></p><ul><li>hold up with<strong> </strong>a single touch<em> (Morata vs Juventus)</em></li><li>hold it up and lay it off with 2+ touches <em>(Morata vs England, Martínez vs Lazio)</em></li><li>hold it up &amp; retain possession, pass later <em>(Bowen vs Arsenal)</em></li></ul><p><strong>Counted separately: </strong>Knock-downs and knock-ons, without having maintained possession with the feet.</p><p><strong>Don’t count:</strong> Progressive carries or passes where the player runs forward with the ball after receiving it. (Mbangula vs Milan)</p><p><strong>The “counting statistic” (happens a number of times each game)</strong></p><ul><li>Hold-up actions into holding possession, winning a free-kick, or passing it (these exclude progressive carries, where players receive a pass and then run towards the opposition goal, like Samuel Mbangula above)</li><li>1-touch lay offs (these passes can go backward or forward) (like the first Morata example, and the latest Morata example from A.C. Milan)</li><li>the knock on headers (separate, but also knock on headers is a relevant aspect of a striker’s hold up play)</li></ul><p><strong>The rate statistic: successes/attempts</strong></p><p><strong>successes</strong>: described above</p><p><strong>attempts</strong>: the number of balls that they received of that nature, including all the failure outcomes (headed it on and lost possession; lost duel; couldn’t control the pass into them; lost possession while trying to control ball; controlled ball, tried to pass it, gave it away)</p><h4>What we can do with the hold-up statistic</h4><p>Adding more context to what’s happening on the pitch at a time of an attempted hold-up may be illuminating.</p><p>Metrics like the <em>hold-up player’s distance to their nearest teammate</em>, and <em>average distance of passes after successful hold-up actions</em> can help to identify the optimal shape for teams hitting upfield balls and attempting to counter. With lots of data, visualizing passing patterns where line thickness shows frequency shows how the ball moves into the next phase of play when it’s held up successfully.</p><p>Another thing to observe is how many teammates progress past the hold-up player after a successful action. This sheds light on how effectively a player enables their team to advance up the pitch. from the time of a successful hold-up, to the end of the team’s possession, measuring how many players were able to advance past the location of the hold-up.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*w3B9zluzUXUJaZ_kbVaApw.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*dsVEH-LR3WmQi98hRge-yw.jpeg" /><figcaption>Two examples of Hold-Up Play, when teams either retain possession higher up the pitch (left) or attempt to counter quickly (right).</figcaption></figure><p>When possessions end with several teammates upfield from the position where the<strong> hold-up action</strong> occurred, the hold-up player has allowed their team to retain possession higher up the field(<em>Bowen vs Arsenal).</em> On the other hand, if only 2–3 players have advanced beyond the location of the hold-up action at the end of the possession, the player who held-up the ball did so to kickstart a counter, or ‘on the break’ (<em>Morata vs Juventus)</em>.</p><p><a href="https://www.linkedin.com/in/finn-laister-smith/">-Finn Laister Smith</a></p><p><a href="https://x.com/SincereFC">sincere FC</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cf41c70cd603" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[A mirage of counting statistics,  and “usage rate” in football]]></title>
            <link>https://sincerefc.medium.com/a-mirage-of-counting-statistics-and-usage-rate-in-football-ee82c5a915f9?source=rss-3e5a8f5ae41a------2</link>
            <guid isPermaLink="false">https://medium.com/p/ee82c5a915f9</guid>
            <category><![CDATA[statistics]]></category>
            <category><![CDATA[football]]></category>
            <category><![CDATA[soccer]]></category>
            <category><![CDATA[premier-league]]></category>
            <dc:creator><![CDATA[FL Smith]]></dc:creator>
            <pubDate>Mon, 06 Jan 2025 12:06:01 GMT</pubDate>
            <atom:updated>2025-01-06T12:06:01.921Z</atom:updated>
            <content:encoded><![CDATA[<h3>A mirage of counting statistics, and “usage rate” in football</h3><p><em>What we can learn from how teams divide their possessions</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*uyZKklcy5JcqUdbPRAk0dw.png" /><figcaption>Bruno Fernandes, Morgan Gibbs-White, Martin Ødegaard, and Michael Olise were some of the most influential players in the Premier League in 2023–24. (source: Google Images)</figcaption></figure><p>In 2025, there is more mention of goals and assists in football than ever. Sure, your favorite player might be hard to press in midfield — but how many “G/A” have they got this season?</p><p>When elite football clubs look for attacking talent in leagues around the world, the goals and assists figures are a natural place to begin. But when looking at the goals and assists figures of players we don’t watch, we lack an understanding of the opportunities each player had to produce those goals or assists and the share of the team’s opportunities that they used.</p><p>Whether a player will adapt to their new team and role is a key aspect of recruitment, especially in games that require players to share the playmaking “load” (Football, Hockey, Basketball, etc.). Max Odenheimer expanded on this idea in a <a href="https://statsbomb.com/articles/soccer/memphis-depay-why-it-hasnt-worked-out-at-man-united/">2017 piece on Statsbomb</a>, describing the issues Memphis Depay dealt with at Manchester United after moving from PSV Eindhoven in 2015.</p><figure><img alt="Memphis Depay at Manchester United" src="https://cdn-images-1.medium.com/max/619/1*nXVfhNKFVr-2_k0RziGdZw.jpeg" /><figcaption>“At PSV, Memphis was the star man and preferred option. At Man United, he was splitting possessions with Anthony Martial, Juan Mata and Wayne Rooney -Max Odenheimer, StatsBomb 2017</figcaption></figure><p>Odenheimer writes:</p><blockquote><em>“At PSV, Memphis put up elite numbers at a very young age and looked like an all-world prospect. What those numbers didn’t account for was the massive number of possessions he was using…[Depay] certainly put up exceptional counting stats in 2014/15–0.85 NPG+A/90 — but they lose some shine if you consider that when he was on the pitch he was using almost 20% of PSV’s possessions. “</em></blockquote><p>Had they been widely available, a look at Depay’s conversion and usage statistics would have led fans to understand why he failed to repeat those levels of attacking output at Manchester United. Not only was he against higher-quality opposition, he was given far fewer opportunities per match to create and score.</p><p>These idea of ‘usage’ has been discussed in the NBA since the late 1970&#39;s. In basketball, usage measures the percentage of a team’s plays a player ‘uses’ while on the floor.</p><p>Football is different from basketball, in that a basketball game has far more “scoring plays” than a football match. Football fans might argue that a player missing 30 shots is fine, as long as they score the 31st, provided that the 31st shot wins a game. The same can be said about passing: matches can be won with one key pass, and a player’s previous errors are quickly forgotten.</p><blockquote>“[In football,] Shots are less ‘used’ than they are created, and holding high-risk high-reward ‘home run’ type passes to the same measuring stick as regular possession play feels like it’s missing a trick.” <strong>(</strong><a href="https://www.americansocceranalysis.com/home/2021/1/4/mvp-positions-and-the-problem-with-usage-in-soccer"><strong>Kieran Doyle, American Soccer Analysis</strong></a><strong>)</strong></blockquote><p>Thus, ‘usage’ in football doesn’t really mean the same thing as it does in basketball (yet). But nevertheless, understanding these numbers is a crucial part of the scouting process for attacking players.</p><p>When we look at young players’ game films, we can imagine how they might adapt to new systems or roles in training. However, <strong>the actions they put on film are the practical match-speed skills and habits that they are developing</strong>. They might play X amount of direct passes per touch or have X amount of possession per match to do what they do. Players like Memphis Depay get accustomed to those patterns.</p><p>While assisting and goal-scoring <strong>is</strong> about one key moment, the statistics I propose are to inform the viewer about the <em>nature</em> of players’ G/A totals, and <strong>how often</strong> <strong>they help their team get closer to achieving its objective </strong>(scoring)<strong>, given the number of times they have the chance to do so</strong>.</p><p>These statistics describe things the “eye test” already reveals with enough exposure to that player.</p><h3>Current comparative football graphics like this one are not always informative.</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/768/1*yDSuPovBS27PiKn_syU7TQ.jpeg" /><figcaption>Lucas Moura and Sadio Mané in a comparative visual shown on Super Sunday in 2019 (source: Sky Sports UK)</figcaption></figure><p>A typical statistical comparison like the one shown above can be hard to interpret. The numbers show that in 2018–19, Sadio Mané played a larger role in Liverpool’s attack than Lucas Moura did in Tottenham’s: Mané has more counting stats per 90 minutes. However, at the time Liverpool largely played through Sadio Mané and Mohamed Salah in attack, whereas Tottenham played more through players Lucas Moura (Harry Kane, Son Heung-Min).</p><p>Mané and Lucas are shown in a graphic together because they play in similar positions on the pitch. They are both wing players — but they don’t actually fulfill the same ‘role’ within their respective teams. Moura’s efficiency might have been exceptional for his level of ‘usage’ in Tottenham’s attack, whereas Mané’s efficiency might look poor once we consider the number of times he received the ball in the left attacking position.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*pQ6nhJdlt7qce08UPo9jKQ.png" /><figcaption>2018–19 possession statistics for Lucas Moura (Tottenham) and Sadio Mané (Liverpool) (Source: FBRef)</figcaption></figure><p>These are things Tottenham and Liverpool fans know from watching their teams play<strong>:</strong> however, they are not interpretable from the original graphic. From such numbers, we cannot tell how good a player is compared to anybody else, or how <strong><em>efficient</em> </strong>they are, due to a lack of percentages, rankings, and/or <strong>“</strong>rate statistics<strong>”</strong> (besides their shot conversion rate).</p><p>Returning to the example of Memphis Depay at PSV, by looking at statistics like the ones shown in that Sky Sports graphic, Depay may have appeared to be outperforming other young wingers at the time.</p><p>Appearances can be deceiving.</p><h3>Counting Statistics and Rate Statistics</h3><p>Goals and assists are “counting statistics” — you simply accumulate them.</p><p>Currently, the most commonly used rate statistic in football is<strong> shot conversion</strong>. <a href="https://www.statmuse.com/fc/ask/highest-shot-conversion-rate-in-premier-league">StatMuse</a> is one of many sites showing conversion in terms of accuracy {shots on target/shots}, but there is also goal/shot conversion, as shown in graphics like the one below:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MJ4YVsQPHhwq4ouorHOYJg.jpeg" /><figcaption>Liverpool forwards’ <strong>conversion rate</strong>,<strong> </strong>measured<strong> </strong>mid-way through the 2023–24 season (source: Sky Sports)</figcaption></figure><p>Rate statistics help to give context to counting statistics (in this case, goals).</p><p>In modern football tactics, a subset of players in each team typically score most of the goals, just as there are a subset of players making the ‘key passes’ into dangerous areas of the pitch. Because of these defined roles, it’s possible to create metrics like the ones shown above, but related to <strong>passing</strong>.</p><p>A <a href="https://www.nytimes.com/athletic/5470579/2024/05/04/progressive-actions-premier-league-explainer/">May 2024 piece by John Muller</a> in the New York Times describes “progressive actions” (progressive passes, carries, and receptions), and why they are among the most important actions in football matches. The piece contains a useful visual for illustrating the importance of rate statistics for chance creation and key passes.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QiCaVMHScSyVZl9F-J9n_Q.png" /><figcaption>Progressive pass leaders by zone, Premier League, 2023–24. (Source: The Athletic)</figcaption></figure><p>Most fans would agree with the broad takeaways from this visual: in the final third, <strong>Bukayo Saka</strong>, <strong>Martin Odegaard</strong>, and <strong>Bruno Fernandes</strong> made the most ‘key passes’ over the 2023–24 season.</p><p>But these are <em>counting stats, </em>and<em> </em>they lack the context described before.</p><p>These numbers fail to capture:</p><ul><li>The number of times a player has the ball in that situation (<a href="https://zone14.ai/en/blog/the-zone-14-in-football/">zones 13, 14, and 15</a>, with teammates ahead of them)</li><li>The number of times they attempt such an action (a key pass or cross from that position, or a “progressive dribble”),</li><li>The share of a team’s “possession-ending actions” that the player uses (by either losing the ball or attempting an action), specifically in these attacking areas.</li></ul><p>A question to ask Manchester United fans would be: <strong><em>is Bruno Fernandes leading in these statistics evidence that Manchester United were attacking well in ‘23–24</em></strong>? For all of the key passes Fernandes makes in those areas, <strong>how often does he attempt them, and what share of Manchester United’s final third actions do those attempts take up? </strong>Most fans would probably say that the relationship isn’t that simple.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*IFaoUsilmHEy7ahdjzySaw.png" /><figcaption>Premier League leaders in Progressive Actions (pass/carry/reception) , 2023–2024 (Source: The Athletic)</figcaption></figure><p>Fernandes led the entire Premier League in progressive actions<strong> </strong>in 2023–24. On the one hand, he was the only creative outlet in a Manchester United team lacking inspiration. On the other hand, in certain matches, the Red Devils’ attack may have benefitted from Fernandes <em>not</em> attempting ‘progressive actions’, and shifting his play style towards that of somebody like <strong>Martin Ødegaard</strong>, who attempts fewer expansive or risky passes in order to keep the ball (notice that Ødegaard doesn’t appear anywhere on that graphic).</p><h3>Usage Rate and ‘Possession-Ending’ Actions</h3><p>In basketball, a player’s usage rate captures the percentage of their team’s possessions that end with them<strong>. </strong>Basketball possessions either end with shots, fouls, or turnovers. <a href="https://medium.com/r?url=https%3A%2F%2Fwww.americansocceranalysis.com%2Fhome%2F2021%2F1%2F4%2Fmvp-positions-and-the-problem-with-usage-in-soccer">Kieran Doyle’s 2021 article</a>, referencing Odenheimers’ earlier work, describes a footballer’s “usage rate” as the share of a team’s <strong>possession-ending actions</strong> they use. “Possession-ending actions” include incomplete passes, shots, failed take-ons, dispossessions, and failed receptions.</p><p>Since there are many more turnovers of possession than shot attempts in football, this statistic measures something a bit different about a football player than it does a basketball player. While it’s not as directly related to shooting, it’s still about <strong>dominating the ball when their team has possession</strong>.</p><p>The following data from <a href="https://www.americansocceranalysis.com/home/2021/1/4/mvp-positions-and-the-problem-with-usage-in-soccer">Doyle’s article</a> shows MLS players’ possession-ending actions per 96 possessions and their “usage rate,” which is a player’s share of their team’s total possession-ending actions per 96 minutes.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/604/1*FI1_7T60DZlatG6uZrEJBw.jpeg" /><figcaption><strong><em>“Action usage %”</em></strong><em> = the number of “possession-ending actions” a player takes, divided by the team’s total possession-ending actions. (Source: Kieran Doyle, American Soccer Analysis, 2021)</em></figcaption></figure><p>The data is hard to interpret without exposure to the MLS. Toronto FC’s Alejandro Pozuelo has<strong> </strong>a usage rate of 15.3%, and registers 20.5 possession-ending actions per 96 minutes: what should that mean to a non-MLS watcher?</p><p>The following table shows the same data for the <strong>2023–24 Premier League season.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jWHki36sfX1iBCnOphxoJA.jpeg" /><figcaption>Premier League leaders in Possession-Ending actions in 2023–24 (source: FBRef)</figcaption></figure><p>Premier League fans can connect this data to their own experience and knowledge who the division&#39;s best, and most ‘ball-dominant’ players are. Unsurprisingly, Bruno Fernandes is among the top 5 in possession-ending actions alongside other ‘playmakers’ like <strong>Trent Alexander-Arnold</strong> and <strong>Kevin De Bruyne</strong>.</p><p>Here’s what we can understand from viewing these data points (team action%, SCA%, and actions/96) together:</p><ul><li>Who is ‘aggressive’, or ‘direct’, <em>(more possession-ending actions per 96 minutes)</em></li><li>Who is ball-dominant in their team <em>(high in actions/96 and action %)</em></li><li>The difference between a playmaker closer to the opposition goal and a playmaker in buildup play <em>(they either have high usage and low SCA% if far from the goal, or high usage and high SCA% if nearer to the goal)</em>.</li></ul><p>This brings us back to the comparison of Bruno Fernandes and Martin Ødegaard.</p><p>These two are considered the premier “number 10” players in the Premier League. But they are very different players: <strong>while Fernandes and Odegaard each ended with 18 G/A in 2023–24, Fernandes (24.23) finished in the top 5 of possession-ending actions per 96, while Odegaard had nearly 8 fewer actions per 96 minutes (16.89), putting him outside the Premier League’s top 50.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*drOQ21zy4Dryw4okuqhucw.jpeg" /><figcaption>Possession-Ending Actions for Bruno Fernandes (Manchester United), Martin Ødegaard (Arsenal) and Morgan Gibbs-White (Nottingham Forest). (Source: FBRef)</figcaption></figure><p>This data above is not the final version of what we aim to measure, but it provides helpful context to view alongside a player’s “end product” (goals, assists, shot-creating actions), since we can compare all attacking players on these terms.</p><p>This comparison illustrates how safe (or risky) each player’s approach is, which is dictated by their manager’s tactics as much as it is by their own ability. Ødegaard is less “direct” than Bruno Fernandes: he ends fewer possessions per 96 and he tries fewer such actions. By that metric he has lower ‘usage’ — but this season, Arsenal fans have frequently cited the team’s <a href="https://x.com/KNGDM_Prince/status/1868161693119099159">“risk-averse”</a> approach as something that holds them back from winning games. This ‘risk-averse’ style begins with Ødegaard, who needs to be more courageous and unpredictable to unlock defenses.</p><p>The third player in the graphic is <strong>Morgan Gibbs-White</strong> from Nottingham Forest. Gibbs-White is generally regarded as a talented player, but far below Bruno Fernandes and Martin Ødegaard in terms of reputation and impact.</p><p>While Bruno Fernandes registered many more shot-creating actions than Gibbs-White(219 to 149), Gibbs-White was also near the top in actions/96, and finished 2nd in <strong>share</strong> of his team’s shot-creating actions (20.35) behind only Fernandes (21.35), suggesting that he was more aggressive and more influential than nearly any other player.</p><p>You might hear a fan say they “rate” Morgan Gibbs-White, but because Nottingham Forest rarely received media attention while sitting near the relegation zone in 2023–24, the extent of Gibbs-White’s impact compared to other players in his position across the league often goes unnoticed without proper context.</p><p>To understand the true nature of Gibbs-White’s “counting stats”, we need to know more about his usage, as well as how many opportunities Nottingham Forest get in attacking possession, relative to teams like Manchester United or Arsenal.</p><p>We need to know how many opportunities players like Gibbs-White, Ødegaard and Fernandes have in attacking positions that can lead to ‘progressive actions’ or ‘shot-creating actions’, and what each player does with them.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_Womef2ZFSUM_3YWKs2agg.jpeg" /><figcaption>A post by Statmuse comparing Morgan Gibbs-White and Martin Ødegaard from December 14, 2024 (<a href="https://x.com/SleeperFooty/status/1868019106869571982">Source: X</a>)</figcaption></figure><p>The above graphic lacks exactly that context.</p><p>The basketball equivalent of looking at goals/assists is the total number of points or assists players accumulate over a season. When a basketball team scouts a player, they contextualize the player’s points and assists per game using the player’s usage and efficiency statistics. This helps them understand how the player’s<em> counting stats</em> relate to <em>the offensive load they carry for their team</em>.</p><p>By looking at output, usage, and efficiency together, we are trying to evaluate all players using the same criteria. In this era of football tactics, although players fulfill different roles from moment to moment, in the final third a specific few players take on the majority of shooting or creative responsibilities.</p><p>Therefore, it is apt to compare players in terms of the role they have in their team, where that “role” is related to their level of offensive load. The same way role players in basketball are evaluated based on their number of touches, minutes and field goals per game, football players’ statistics can be viewed with the same context (as a “first option”, “second option”, “role player”, etc).</p><p>The following chart shows Premier League players’ possession-ending actions and share of their team’s shot-creating actions.</p><h4>Premier League ‘Usage Rate’ vs SCA per 90, 2023–24</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-n8zM9WnyXKEuWgGo2BhtA.png" /><figcaption><strong>Possession-Ending actions per 96 minutes</strong> vs. <strong>% of Team Shot-Creating Actions</strong>, 2023–24 Premier League(min. 1000 minutes). Points with names have either ≥<strong>15 G+A</strong>, or ≥<strong>21 possession-ending actions/96</strong>.</figcaption></figure><p>We can interpret being farther out on the horizontal axis as having been more ‘ball-dominant’, and being farther up on the vertical axis as having been more influential within their team’s attack.</p><p>Three key pieces of information missing from this visual.</p><p>First, although you can infer who has a large role in their team from their possession-ending actions and share of shot-creating actions (SCA),<strong> </strong>the share of shot-creating actions isn’t yet measured based <strong>specifically on the moments when a player is on the field</strong>, the way statistics like <a href="https://www.ijcai.org/proceedings/2020/0648.pdf">VAEP</a> (value-added expected possession) are calculated.</p><p>Second, the measures of possession-ending actions are also not specific to the final third, as discussed previously.</p><p>And third, it misses the context of how many opportunities each player receives to incur ‘actions’, whether it’s creating from the wing like Michael Olise and Bukayo Saka, creating from zones 13–15 (near the goal) like Bruno Fernandes and Morgan Gibbs-White, or creating as a striker, like Ollie Watkins or Erling Haaland (who does not ‘create’ at all). An experienced Premier League fan understands that although all players accumulate these statistics, the players play different positions, are relied on by their teams to different extents, and play for teams that possess the ball in differing amounts; and therefore accumulate their counting statistics in vastly different ways.</p><p>To untangle the sea of information we’re faced with when measuring all footballers at once, we need to understand the role someone plays within their team’s hierarchy, and how efficiently they play it: where they are on the pitch, what they do with the ball (cross, pass, shoot, dribble), and how much they/their team do it.</p><ul><li>How many times it results in an <strong>attempted</strong> progressive pass (cross, through ball), or a <strong>shot</strong></li><li>How many times it results in a loss of possession (dispossessed, miscontrol, failed take-on)</li></ul><h3><strong>Where we are going: Usage Rate in the Final Third</strong></h3><p>For attacking players, the ‘usage’ statistics described be more specific to actions relevant to goal creation: rather than all “possession-ending actions”, we are interested in <strong>a) the</strong> <strong>possession-ending actions that happen within shooting range of the opposition goal </strong>(rather than a loose pass or failed take-on/reception in the defensive third of the pitch), <strong>b) the shot-creating actions</strong>, and <strong>c) the opportunities each player took to produce a) and b)</strong>.</p><p>These statistics will be most useful during recruitment, <strong>when looking at the G/A figures of players in a league we do not watch.</strong></p><p>Suppose that in the Russian Premier League the player currently on top of the goalscoring chart, striker Y, has scored 25 times this season. How many shots is that from? And what percentage of his team’s shooting opportunities in the final third did striker Y take?</p><p>Alternatively, suppose RPL midfielder X has made 65 key passes in 30 appearances.</p><ul><li>How much does midfielder X’s team have the ball in the relevant areas, and what share of that possession does midfielder X use?</li><li>Out of all the receptions he made in a position to make a key pass, what were the outcomes? (Dispossessed, failed take-on, miscontrol, shot, key pass (failed or successful), or some type of ‘recycle pass’ (one that has ≤0 impact to <a href="https://statsbomb.com/soccer-metrics/possession-value-models-explained/">xT</a>, expected Threat, essentially retaining possession)</li></ul><p>This added context can reveal how certain players have accumulated their counting stats via an <strong>unsustainably</strong> high usage rate (Memphis Depay), while others have achieved comparable statistics with much less time on the ball, due to their team not having the ball, or them not being a primary option on their team.</p><h3>Counting New Events on FBRef</h3><p>Here, we are using Trent Alexander Arnold as an example.</p><ol><li><strong>Attempts at shot-creating actions and key passes</strong></li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*XtnVNDYFjIsIC6PuikkMVw.png" /><figcaption>“Passing” statistics for Trent Alexander Arnold (source: FBRef)</figcaption></figure><p>The area where they currently show the passes people make into the area doesn’t include attempts in those areas (we will need to view touches in the final third in buildup or on the break as well as these to understand how often players even have the OPPORTUNITY to play a key pass).</p><p><strong>2. Final-third specific possession-ending actions</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QniN_19YS-O46aYpiIRAbw.png" /><figcaption>“Possession” statistics for Trent Alexander Arnold (source: FBRef)</figcaption></figure><p>Events such as mis-controls, dispossessions, and other stats mentioned above aren’t shown<strong> </strong>in<strong> </strong>the<strong> final third </strong>specifically. Currently, we only see carries into the final third (and those are the successful ones).</p><h3>Defining the Statistics</h3><h4>Usage Rate Statistics</h4><p>Suppose that in a match, one team has 600 total touches in the final third.</p><p>There were 122 possession-ending actions attempted in the final third.</p><p>There were 53 possession ending actions OUTSIDE the final third (“turnovers”) — this includes missed passes, unsuccessful dribbles and forced/unforced losses of possession.</p><p>Each player’s usage rate is their share of the “attempted possession ending actions” in the final third, plus the possession ending actions outside the final third. Ideally, this is calculated using the actions that happened when a player was on the pitch (perhaps they were substituted after 60 minutes).</p><p>A player’s number of, and share of their team’s touches in the final third are also relevant. Whereas the above rate statistic measures “end product”, the share of touches illustrates whether the player is someone who spends more time on the ball, or off it. The number of touches tells you how much the team has the ball in general. Are they a possessive team or a counter-attacking team?</p><p><strong>What are “attempted possession-ending actions”?</strong></p><ul><li>passes: passes that don’t qualify as “progressive” but still increase xT, and end in the final third; attempted progressive passes into the final third; attempted passes (progressive or not) ending in the penalty area</li><li>dribbles attempted</li><li>crosses attempted</li><li>Shots attempted</li></ul><h4>Usage Counting Statistics</h4><p>You can contextualize every player’s attacking output (goals, assists, and SCA/90) using their usage rate, to understand how ball-dominant (and therefore how efficient/wasteful) they are, as well as how ball-dominant their team is.</p><p>Suppose a player has got X Goals and Y assists this season. You can see their usage rate — but you can also understand the volume of opportunity the player receives as a function of their team’s playing style.</p><ul><li>Player has X Goals, Y Assists</li><li>Player attempted K possession-ending actions in the final third in total, J possession-ending actions outside the final third</li><li>Player had N touches in the final third overall (over the season)</li><li>Player’s TEAM had XX touches in the final third total</li><li>Playery’s TEAM averages XXX attacks per game</li></ul><p>Players’ data can be viewed through the lens of the number of attacks their team has each match, similar to how per-game stats benefit from the context of “pace” in the NBA. <strong>This</strong> is the full context that the earlier graphic comparing Morgan Gibbs-White and Martin Ødegaard’s assists needs.</p><p>These stats tell us more than “who creates the most”. They can show who has the ball, how much; they can show who creates the most, given how much they have the ball; and they can show who’s most efficient at those actions.</p><p>Finn Laister Smith<br>@sincereFC</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ee82c5a915f9" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>