As you may know, I have had some time on my hands recently. In addition to taking a bit of a holiday, during my free time I have also been writing a book explaining the various competitive edges I have uncovered in the last couple of years researching football.

Now the majority of this book may never see print, but that’s not really the point – the point of this is to explain to stats laypeople how their football teams can improve. A lot of the material inside it isn’t completely unique, but it adheres to some extremely important requests decision makers inside of various sports have with regard to analysis.

  1. It’s practical.
  2. It’s hopefully easy to understand.
  3. It focuses on what is important: winning.

In the past, there has always been a fear among analyst types that by writing about edges and explaining too much of their research, they would be giving away hugely valuable information for free.

What if that’s the only edge I ever uncover? Do I diminish my own chances of success by making everyone else smarter?

For a while, I also held this fear until I realized something I explained in a mailbag a couple of weeks ago – the well on improvement goes incredibly deep. I could cut this or any number of other chapters out and still end up with a book chock full of innovation and improvement.

At the end of the day, it seemed a lot better choice not to live in fear and just work hard at being awesome.

Amusingly, even if a team knew all of this information and kept it to themselves, they would not be guaranteed success. As I have been explaining publicly for the last two months, simply possessing great information isn’t enough – it’s how you use it that truly matters.

Unnamed Book Title Chapter X – Shot Locations
We’ll start here because I think this is one of the easiest concepts to grasp from an analytical perspective and one of the hardest to teach to football people. It flows from four basic principles

  • The closer a shot is to goal, the more likely it is to be converted.
  • Central locations are better than wide. (This mostly has to do with angles of the goal covered by the goalkeeper from wide shots.)
  • At the same distance, shots with feet are far more likely to become goals than shots with the head.
  • Crosses are hard.

If you want to visualize these principles, they look like this:

Heads_vs_Feet_final

feet_crosses_vs_not

Principle 1 is very easy to explain and understand. Principle 2 is also quite straightforward, with or without the trigonometry involved to prove the concept. 3 and 4 might seem open to argument, but are conclusively proven by data analysis.

From these principles flows one core concept: We want to push players to take the highest quality shots possible. And related to that, at a team level we need to find ways of attacking that create high quality shots.

I’m not suggesting these are ground shaking ideas – far from it. These are basic truths about the game of football. Why then are they hard to teach?

Consider the following common football cliches:

You can’t score if you don’t shoot.

Go ahead son, have a pop.

We want our players to have the freedom to try things on the pitch – we don’t want to limit their creativity.

He unleashed a thunderbastard from distance!

Players are often taught to “try their luck from range,” and to “test the keeper.” In my experience, they are rarely taught where good shooting locations are and not to shoot unless they are inside them. This creates a problem because the whole concept of shot locations competes directly with a far more powerful force: habit.

We try to automate as much behavior as possible on the pitch so that players can use their active brains to concentrate on reading the game. When it comes to shots, you are dealing with hundreds of thousands of past attempts at every level of football played, where that player has taken a shot. If they have never been taught where to take them, a player then needs to think before every time they might shoot whether that shot is good or not, and then choose whether to take it.

The combination of fan culture where long-range goals are the ones that make the highlight shows and get the plaudits from the pundits and the sheer weight of habit make this a hard concept to convey, especially to mature professional footballers. This is even more true if no one has bothered to explain the concept of good and bad shots to players previously.

At some point not limiting creativity of your players begins to directly butt heads with the probability of winning games.

Okay, whatever – it seems hard for some players to learn this… Why bother?

Because taking higher value shots makes your attack and scoring output more consistent. As a result, it should make you more likely to win.  

These are two huge edges to grind over the course of a season.

It helps to add some very basic math into this to fully illustrate the scale of what I’m talking about.

Take the picture below.

Picture1

A shot from this location would have about a 3% chance of being a goal (.03 expected goals or xG). If a player is completely by himself with no teammates around, it might make sense to take the shot, but on average you will get 1 goal for every 33 shots from this location. For most teams that is about one goal for every two to three games worth of shots. Not great, Bob!

Now add an additional player to the mix that is making a vertical run toward the goal.

Picture2

shot-loc-animation_loop

A completed pass here creates a likely 1v1 with the goalkeeper from fairly close range. The resulting shot is approximately a minimum of .40 xG, so it would be converted into a goal at least 40% of the time.

How often does that pass need to be successful for passing to be a better option than shooting?

The mathematic answer is easy, but I can tell you from experience that most players gut instinct gives them an answer far higher than the actual probability.

If we do the math(s), we find the shot in example 2 (after the pass) is 13.3 times more likely to become a goal than the shot from distance (.40/.03). Even if your players could only pull off this successful pass one in every ten times, it still adds positive expectation to the result at the end of the game.

However, it doesn’t mean that making this pass every time is the correct way to go about it. In every strategy game in the world, you need to vary your strategies to have the highest chance at success. Football is no different.

There’s another factor here that gets far more math-y, but it involves variance. The ride on the variance train is a bumpy one, and we would far prefer to minimize variance whenever possible in an effort to create more certainty in the outcome.

Consistently creating fewer, higher quality chances lowers the variance in scoring output. The example used in a great piece of writing by Danny Page is Team Coin vs Team Die. Team Coin takes 4 shots a game, making their expected goal return 2. Team Die takes 12 shots a game, which on a six-sided die would also make their expected goal return 2. Simples. However, this is where variance comes into play.

If we simulate these two teams playing each other 10000 times, what we find is that Team Coin wins an average of 40% of the time, Team Die 36% of the time, and a draw occurs 24% of the time. The point return over the course of the season would be 1.42 for Team Coin and 1.36 for Team Die.

.06 points per game difference seems microscopic, right? Take that across a whole season, however, and you end up with 2-3 points more for the team with lower variance in their chances. 2-3 points is often the difference between relegation and survival, and it can also be the difference between Champions/Champions League spots or automatic promotion vs the playoffs.

This is counterintuitive across the entire sport, so let me state it again:

All things being equal, you would rather have fewer chances with a very high quality than many small quality chances.

Creating the big chances is hard, but it’s also damned important in giving your team the best chance of ending the season at the top of the table.

The variance train also makes a huge difference in player value when it comes to scoring goals.

sameplayer_combined

This is the same player, two seasons apart. As you can see, nearly every detail in this plot is identical except one thing: the goal return. In one season the player scored three goals and was considered good, but not great. In the next season, he scored 10 across almost exactly the same number of shots and was considered one of the best players in the league.

What changed?

Nothing.

His expected goals per shot were the same. His shot locations were almost shockingly similar, despite playing in a couple of different roles. And yet one season the goal return was below xG and the other it was basically twice expectation.

That’s the role that variance plays in the game and in our perception of player value.

Should it? Almost certainly not.

What about shooters who we are sure are good at shooting from long distance?
This is a question I get asked all the time whenever I present this information to coaches. “We know David and Frank are good shooters from long range – shouldn’t they be allowed to shoot from distance when they want to?”

The answer to this question and so many other ones in football is: Lionel Messi

Messi is assuredly one of the most skilled shooters in football. His accuracy from almost any range is precise and in many cases astounding.

This is Messi’s shot map from the 2014-15 season.

Lionel-MESSI_14_15_logo

Nearly every one of his 181 non-penalty shots that season were clustered in good areas. The ones that are outside the good areas are almost exclusively from direct free kicks. This is despite the fact that we know he is skilled at shooting.

The rationale behind this is simple: Barcelona know that shooting from prime locations makes them far more likely to score, which in turn makes them more likely to win. They pass up tons of possible shots over the course of a possession in search of a better one, a choice that few teams and players in the world make.

However, the teams that do make this choice tend to be the teams that have the most success.

To put it another way: if shooting only from prime locations is good enough for Leo Messi, shouldn’t it be good enough for you?

How can we teach this to players?
There are many ways you can go about introducing and reinforcing the concept to players and coaches alike, but I have two favorites. One of these is an experiment and the other involves a bit of paint.

The Experiment

  • Put two one-yard wide targets on either side of a normal goal, spanning the entire height of the goal.
  • Mark out shooting locations at 9, 15, and 23 yards away from goal.
  • Ask players to estimate how many of their own shots out of ten they expect will hit either target from each distance.
  • Roll the ball to the players and have them shoot for the targets. Do this 10 times from each distance and record the results. (You can do this over a few days, and you can also do it multiple times over pre-season and early in the year.)
  • Tabulate the results by individual and by population and present them.

The reasoning behind this easy: merely telling players about the concept is one thing. Having them learn it for themselves is a far more potent learning tool. Additionally, it teaches you a bit about their accuracy from distance in a laboratory setting. Game settings are obviously much harder, and you expect performance to be worse there.

Even if players are two or even three times as likely to score from 23 yards as an average player, that still only takes their expected goals per shot from that range from .03 to .09. This continues to compare poorly to the .40 you would get from completing throughball passes to runners into the box. That said, you also have learned who you are more likely to want taking long shots versus a packed defense to help mix up your strategy.

The Paint
Explaining to players constantly about what is and is not a good shot annoys everyone and will wear thin pretty quickly. As an alternative measure, you can paint rings like you see on the shot maps on each of your training pitches. The solid blue line below corresponds to an 11% chance or better for a normal shot taken with feet, and it’s convenient because it basically rubs up against the 18-yard box). The black line is a 6% chance or better. Anything beyond these lines is considered a poor shot unless something special is going on (like a 1v1 with a keeper).

6-11-crop

Explain to the players what the lines mean, and then let them evaluate themselves every time they take a shot. In behavioral economics terms, this is a nudge, and it’s one that will likely have a large impact across your entire club as time progresses. If the senior players in your team buy in to the concept and start to police it, you’ve already succeeded.

Other Ways to Reinforce the Concept
You have seen the shot maps above. A further method to highlight the issue to players is to review their own past shot locations and how successful they have been from each spot. Another way is to have them highlight past players they thought were good at shooting from specific locations and review how those players performed as well.

Almost inevitably, everyone remembers the long range screamer that went into the top 90. No one pictures the other 30 shots from that same distance that never turned into goals.

This is a classic example from a Premier League legend.

Frank-LAMPARD_2008-2009_logo

Massive volume of shots from long range. Zero open play goals outside the prime rings, and only 1 from a direct free kick.

Yet Lampard was a great goalscoring midfielder – he just didn’t get many goals from long distance, even though many people would suggest that was an important part of his game. (And to be fair, he had to deal with packed boxes constantly, which is why there are so many blocks in the chart.)

Even great players can improve their games, as long as we find the right ways to teach and emphasize the concepts.

Final Thoughts

The concepts in this chapter aren’t new – they have been floating around the football analytics stats sphere for years. Hell, plenty of proper football people understand this stuff intuitively as well, but I’m not sure I’ve ever seen it explained from base concepts there in public.

A couple of other notes before I go:

I think both the shot quality rings idea and the shooting location experiments listed above developed out of a discussion between DOCTOR Marek Kwiatkowski and myself about how football teams might train these things, but details are pretty fuzzy now.

As far as I know, no teams have currently painted their training pitches with shot quality rings. I would love to get feedback from people if/when this ever happens. Feel free to send me an email and a photo to mixedknuts@gmail.com or hit me up on twitter at @mixedknuts.

I also have never actually seen the results from the experiment listed above. It’s incredibly basic, but would be fascinating to track the data across age groups, league levels, table placement and any number of other wrinkles I’ve probably never even considered. It’s also potentially a really simple way to track shooting skill as players continue to develop.

Ted Knutson

  • Toshack

    From theory to practice. Great stuff! Lots of great stats stuff being published out there, but not often does it try to make the leap from knowing something to discussing what to do to change things around in real (pitch) life.

    You have (tons of) data. That’s not enough. You turn it into information. That’s not enough. You transform it into knowledge (of a coach and then a player). That’s not enough. You turn it into skills so that you can use the knowledge on the pitch. That’s still not enough. You have to mix it with experience and 10 000 hours of training. And painted circles and shooting training from different distance is one way to make the transformation of data into successful performance on the pitch.

  • Ron IsNotMyRealName

    Could it be that Messi has such an optimal shot location because…he’s Messi? And he plays with other great players at Barca. Perhaps a lesser player on a lesser team cannot achieve that.

    This is giving me flashbacks to the days in baseball analytics when people suggested that the pitcher has no control over what happens to a ball once it is in play, aside from whether or not it is a home run. There’s not a one size fits all solution here, I don’t think.

    The Steph Curry viz that has been posted on this site in the past is brilliant not only because of its simplicity, but because it shows at a glance not only where Curry is better, but objectively how good he is (and if you have any savvy about the game, how good he is relative to his peers). What football lacks is any sense of that objective “how good” measure, and that’s the issue with football people — they understand, as baseball people did, that the players matter.

    • Paul Tiensuu

      The point about Messi’s shot locations is not that he shoots mostly from optimal locations but that he does so despite also having by far the best long time out of box conversion rate. Of course, the fact that he gets to optimal locations is due to the quality of himself and his team, that is not contested. What Ted tries to say is that other, particularly the lesser, teams and players should strive to get to those locations too, to score more goals. Messi can convert the shot from distance with 3-6 times more likely than an average-good player in a top league. When the players get to the optimal distances, that difference drops significantly. So the worse your team and your players are, the more they should be inclined to only shoot from the prime scoring zone and not waste their attacks by taking desperate shots.

      • Ron IsNotMyRealName

        I don’t think that’s true about Messi at all.

        I think he’s saying Messi is a great player because he gets great shots.

        Where I sort of lose faith in all of this xG stuff at least as a generalization is when I watch someone like Coutinho shooting from outside left into the opposite corner. You can’t tell me that’s a 5% shot for him.

        • Paul Tiensuu

          I don’t quite understand what’s the point of your last paragraph. To me it seems to repeat the point about Lampard: you remember him for his great long shots, yet when you look at it in whole, he converts those long shots at extremely small rate. Even if for Coutinho that rate is a little bit better than I think it was for Lampard, he still misses more than 19 out of 20 tries from distance. That it would predict that one exact time when he is successful would not be generalisation but ability to predict a one separate event and stats do not do that.

          Anyway, whatever it is in my post you think is not true about Messi, is statistically proven, so there’s really no dispute, and your second paragraph really misses the point of this article. Mr. Knutson uses Messi as example, because statheads have noticed that Messi’s big volume shot conversion is on its own level from any location, and particularly from longer distances where the differences are bigger. Anybody else who gets to those really high volumes of shots sees his conversion rate drop to more humane level. Or more accurately to how Mr. Knutson does this, is that the xG gives you the average for all players in the database, and Messi’s goals constantly surpasses his xG, from any shooting location. That’s why it is for Messi less advantageous than to others to get to those better shot locations, because the difference is smaller there, yet he appears to agree that it is advantageous even for him. So, Mr. Knutson concludes that it is advantageous, a fortiori, for everybody else. He may be wrong too, but it’s a good argument and not countered by what you said.

          • Ron IsNotMyRealName

            I think you’ve misunderstood the article. Simple as that.

            I guess I’ll keep my point to myself since it’s apparently not as straightforward as it seems to me. Knowledge is power, and all that.

  • Ron IsNotMyRealName

    It’s fascinating that even a shot with feet from basically the dot has like a 50% or less chance of being a goal. I don’t think many people would expect that, and I wonder how that changes based on the specifics of the shot and how the ball got there (dribble, through ball, square ball, rebound, weirdness, etc.)

  • Paul Tiensuu

    Hey, I took a look at some long time conversion rates recently and was wondering how they relate to the xG’s, to which I have no access. Simply, is it because they took shots from good/bad locations, or because they are good/bad shooters that they have had good/bad conversion rates in the long run. Your example of Lampard who is famous for his long shooting, but who would probably have been well advised not to shoot from distance, reminds me of another player that I suspect was ill-advised to take a lot of long distance shots because “he scores from 40 yards”. So, that would be Gerrard. Could you, please, with your immense data, reveal Gerrard’s G/xG numbers from some seasons at least? Even better if you can give the maps like that, which also show the difference between free kick and open play goals.

  • allanderek

    A great article. I have a minor point. I’m not sure I follow your logic regarding Lionel Messi. He may be one of the best shooters from all locations but he is also playing in one of the best teams, and in particular one of the best possession/passing teams. So, you say “if shooting only from prime locations is good enough for Leo Messi, shouldn’t it be good enough for you?” But surely your data is really only for Lionel Messi playing in the Barcelona team. So that should be “If shooting only from prime locations is good enough for Lionel Messi in the Barcelona team shouldn’t it be good enough for you playing in the Barcelona team?”

    Earlier in the article, you give the graphical example of the blue team attacking the red team, with a shot having xG of 0.03 whilst a shot taken after a completed through pass having an xG of 0.4. You don’t quite state it exactly, but roughly speaking you’re stating that passing is the better option if your probability of successfully completing the pass multiplied by the probability of scoring from the resulting shot is higher than the probability of scoring from where you are. (Of course there are other caveats such as which is more likely to result in a secondary chance etc but roughly that’s the basic idea). But if you’re in the Barcelona team, then the chance of completing the pass is surely higher than if you’re in some other lesser team.

    Another point; passing and shooting are of course not the only two options. Messi may be one of the best shooters, but he’s also one of the best dribblers around. In many situations dribbling around one or more defenders to create a better shooting opportunity may well be the best option for Lionel Messi, whilst shooting from where you are is the best option for someone with less than Lionel Messi-like dribbling abilities.

    To be clear, I’m not disagreeing with your point at all. I’m merely suggesting that Lionel Messi is perhaps not that convincing an argument here.

    • Paul Tiensuu

      The point is this: your average top level player scores at rate of 2-3% from a distance from where Messi scores, from 2009 on (according to whoscored.com, which is for all teams in all competitions bar the domestic cups against any but Real Madrid and such), with about 9% rate (ok, so only 3-4.5 times more often, but he held, for several seasons, a 12% rate from out of box). When you get closer, to an area from where the average player scores with, say, 40% accuracy, Messi converts with a rate not much better than the average player, about 45%-50%. For Messi, it means he has to get 5 long shots for 1 clear-cut chance to make it an equal choice to shoot as to play a one-two. He might even accomplish that easily. The average guy, however, needs to take 15-20 long shots to make up for one clear-cut chance. The advantage of trying to get a better chance is thus much greater for the non-Messi player, and although he probably will not get there as often as Messi and his Barcelona, still, even if Barcelona’s completion of those killer passes was twice as good as that average top level team’s, that average team would still have much more advantage from trying it.

      For example, see the success story of Leicester, who have taken 4th lowest percentage of shots from outside the box in PL this season, and are the worst passing team in the league (70% completion). They’ve also scored the least goals from out of box (2) and the most from in the box (54). Have they been a bit lucky? Well yes, it shows that even with their passing, it pays to try to get closer to goal where you don’t have to be that lucky to score.

      • allanderek

        Thanks. I was perhaps a bit misleading in stating I didn’t follow the logic. I completely understand the point, and even stated that I agree with the point. My only quibble was that appeal-to-Messi was not a good way to make this point. Your example of Leicester is more convincing.

        You can stop reading now, the rest is just a clarification of the above paragraph.

        Sometimes, it helps to use a bit of mathematical notation. In a given situation, with a low xG rating, the choice that faces the player is something like: Pr(S) Pr(P) * Pr(R)

        where: Pr(S) is the probability of scoring from where you are, Pr(P) is the probability of successfully developing the situation into a better shooting opportunity by passing or dribbling, and Pr(R) is the probability of scoring from the resulting better opportunity. If Pr(S) < Pr(P) * Pr(R) then you should pass up the opportunity to shoot and try to develop a better opportunity. Of course this model is over-simplified, since for example shooting from where you are may result in a better opportunity from a re-bound, or attempting to develop into a better shooting opportunity might result in an *even* better chance, such as a penalty. But for the sake of demonstration this model will do.

        Now, the original argument essentially said "Messi is very good at shooting from situations with low xG, but a very high proportion of his shots come from situations with high xG, he is therefore often passing up the opportunity to shoot from low xG situations in favour of developing the chance into something better. Since Messi is better than you are at shooting from low xG places you should pass up those shooting opportunities more than he does".

        This rests on the assumption that Messi does indeed *get* an average proportion of low xG shooting opportunities. Let's assume that is true (my intuition is that it is true). It also rests upon the assumption that Messi is *correct* to pass up such shooting opportunities, let's assume that is also true.

        For Messi, his decision is based upon whether or not Pr(Sm) < Pr(Pm) * Pr(Rm), where Pr(Sm) is the probability that *Messi* scores from a particular low xG position and similarly for the others. Similarly for the average player (or a player you care about) we have to decide if Pr(Sa) < Pr(Pa) * Pr(Ra).

        So I *think* the original article was stating that because often Pr(Sm) < Pr(Pm) * Pr(Rm) and Pr(Sm) is higher than Pr(Sa) then that implies that Pr(Sa) < Pr(Pa) * Pr(Ra).

        My point is that, it is quite possible that the implication is false, because whilst Pr(Sm) may be higher than Pr(Sa), Pr(Pm) is also higher than Pr(Pa) (not to mention Pr(Rm) is likely greater than Pr(Ra)).

        In plain English rather than maths, Messi is really good at shooting from low xG situations, but he's also *really* good at developing those situations in to opportunities with higher relative xG *and* playing in a team that supports that endeavour. Hence, because it is good for Messi, does not make it *necessarily* good for more ordinary players playing in more ordinary teams. Again, I think it probably is good for those ordinary players to pass up such low xG shooting opportunities, just that appeal-to-Messi is not a convincing demonstration of this.

        • Paul Tiensuu

          Thanks for this. I get the point very well and agree with it to an extent. There was an article here about it earlier, check it out:

          http://statsbomb.com/2014/02/mythbusting-is-long-range-shooting-a-bad-option/

          It says that historically teams, particularly those not at the top of the table, have more often made the good decision when they’ve shot when they had the chance, and no team usually makes the good choice by passing the ball to the flank. It’s not very conclusive, because that 3.6% of goal advantage results from shots “struck from a central position outside the penalty area”, so not all long distance shots, and it is also likely that players take that decision (as any decision to shoot) more often when there already is a decent chance (and it still, as Mr. Knutson says, would often be better to make that one more pass when they already have that nice 5% chance to score because that’s also often a situation where that pass would be easier to complete and it would get directly a better scoring chance) whereas they pass the ball more often when the opponent is well organised in defence and goal is less likely anyway.

          But I think Mr. Knutson’s point is that the teams should train to develop the situation better rather than just take desperate shots because once they get to those prime locations the conversion rates go so much up. On the other hand, some stat heads (I think Knutson knows such people from his last job) recommend focusing on getting that shot conversion higher to get an advantage. These are not mutually exclusive options, of course.

          Messi of course epitomises both virtues, but why I think his point about Messi is good is that while for example Leicester converts those few long shots they take at a miserable rate, Messi does that exceptionally well, so he gets less advantage from getting close to goal than Leicester (although also gets there easier). For us the problem may be to define when the team is better of passing the ball forward and when to shoot from low xG chance, as you suggested. But for the coach it should be to make his team both to convert any shot at better rate, and to get to better shooting locations.

          Finally, the assumption about Messi getting average proportion of low xG chances caught my eye. It would be interesting to get stats on this, but it appears to me to be very difficult to create that data, because it would mean to quantify the shooting chances not taken. Barcelona does get high xG shots both absolutely and proportionally more than your average team, but on the way to those high xG locations Barcelona always has passed at least one low xG, and because Barcelona keeps the ball a lot at a distance from where other teams tend to shoot they probably pass much more of those low xG chances by each high xG shot (and now Bayern even more so). But is that often enough to even the shot chance proportions compared to teams that get less high quality shots, and can Barcelona (or Pep’s team) even have a greater proportion of those? And same for Messi as a player.

          • Paul Tiensuu

            I mean that the difficulty in both these issues, would the teams turn more attacks into goals if the players passed in those situations where they shoot with <0.05 chance, and how many low xG chances a team passes per game, are tough to study because that study would imply quantifying the action that was not taken.

          • allanderek

            Yeah okay, generally speaking I can see that if you’re explaining something like this concept to a non-stats person, the simple statement “Messi is really good at long shots and yet does not take many of them”, is pretty easy to articulate, pretty easy to understand, and can be said without reference to mathematical notation. The fact that just to clarify my point I reverted to a kind of formal notation should tell us that the nuances of this point are not for everyone.

            The original article was about the explaining (and training) shot expectation. So I guess I withdraw my complaint. It’s an interesting subject and one that clearly we quite enjoy discussing, but for non-stats people the simple appeal-to-Messi is likely at least a good starting point. If they object, with some of the objections I’ve raised, then great, they’re interested and will likely understand the more rigorous analysis.

            All that said, I think including your Leicester example is probably a good combination to the appeal-to-Messi. Some people have an intuitive feel for “what’s right for the very best is not necessarily right for the rest of us”.

  • Max

    How can you measure the dangers of passing badly, losing possession and running the risk of a counter attack?
    I reckon that avoiding this might be part of the rationale when teams /players go for the long range shot.

  • Aj

    Hi Ted

    I have a doubt regarding the comparison you make between probabilities of shooting from inside the box versus shooting from outside.

    While considering the probability of shooting inside the box, shouldn’t the probability of bringing the ball inside the box also be accounted for? As I would understand it, the probability of the ball being brought inside the box would be less then the probability of the ball being brought outside the box.

  • Juhan

    One thing that seems to be left out is what else can happen after a scoring chance. One outcome is a goal, one is a corner, one is a catch by the goalie(and a possible counter), one is a rebound that can fall to an attacker or a defender(who might be under pressure to clear or can just play the ball out). And if you try to get to a better shooting location, then there are different outcomes as well. For example the one-two might be intercepted and you might concede a counter.

    Intuitively it feels like shooting on goal from 20m will more likely have an outcome that allows the team to continue attacking or at least wouldn’t get into a position where they’re vulnerable to a counter. With trying to pass through a defense it might be more likely that you’ll lose possession.

    Obviously the skills of the players will come into play here. Barcelona trying to get Messi in a one on one situation is going to be much more likely than Aston Villa trying the same thing.

    Overall I think the difference between long range shooting and trying to make a pass isn’t so big that you’d only have to make one throughball out of 13. Without taking the full range of alternative outcomes into account I think we won’t get a great picture. I think the eventual answer of ‘what’s the optimal shooting range’ is going to be ‘It depends’. And not only on whether the player in a shooting position has someone making a run near him but also how good both players are at shooting from range, passing and finishing from closer positions.

  • Gregan78

    i think the ‘scenario’ you describe with the through ball to the runner beyond the defence is fair. If a player has that picture then your stats back up that probability says he should play the through ball.

    As a side issue and coaching point, having runners that make goalscoring runs behind the defence is key for that scenario. Are clubs calculating this? The Ronaldo run, hitting that offside line at top speed as the ball is played. Any stats? The Packing/Impect receiving stats will give an indicator although it doesn’t show number of runs, only completions.

    Good work.