Jumping and heading as attributes are sometimes mixed up.
Heading dictates how accurate a player's headers are. Players with high heading score from corners and their clearing headers are more likely to fall to safety or their teammates.
Jumping dictates (amongst other things like positiong, strength, aggression and determination) whether a player will win an aerial challenge. A striker with a higher jumping will win more goalkick and long balls thrown forward to him. A striker with a higher heading ability will score more goals from crosses.
A central defender with a higher jumping ability will win more defensive headers or at least do enough to put the striker under pressure. A central defender with high heading ability will score a great deal of goals from corners.
So jumping doesn't equal leaping ability - it's leaping plus physical size. And here is the interesting part, height directly influences jumping - it's built in. To prove this point I created an FM08 game with only regens. After 5 years the game had generated over 32,000 regens. The analysis has to be calculated from regens because regens are pure - not biased by the researchers at SI. Biases blur the underlying patterns in the data.
Anyway of the 32,000 regens only 125 have a jumping attribute of 16+. 124 out of 125 of those players were 1.92m to 1.98m tall - or 6 feet 3 inches to 6 feet 6 inches. The 125th, well he was 1.89m. There is nobody below 6 feet 2 inches with a jumping attribute of 16 to 20.
This doesn't mean all tall players have high jumping, but it does mean that all high jumping attribute players are tall in regens. This is important when you judge regens in FML. If you are looking at a target man or a central defender, check out their height - because their jumping ability may never get over 15 if they are too short. And jumping is a vital attribute for both those positions.
Some people say height is a cosmetic attribute. This analysis proves otherwise. The match engine may not use height to determine who wins headers, but the jumping attribute is directly influenced by it - so in a way, it does.
Showing posts with label FML regens. Show all posts
Showing posts with label FML regens. Show all posts
Tuesday, 16 September 2008
Sunday, 24 August 2008
Judging regens
How do you judge a regen without the judging potential skill? Or how do you sort out the good 4 star potentials from the average 4 star potential players? Similar to the last post, I've analysed a set of data from FM08 to determine which attributes are correlated with potential ability.
Sometimes when you see a regen you head gets turned by 18 to 20 scores in certain attributes. Depending on which attributes those scores represent it could mean absolutely nothing, or something very special. Just like the previous post on strikers, I conducted similar analysis but opened it up to all positions. This analysis therefore doesn't factor in position specific attribute importance (e.g. heading, marking and tackling for DCs). So just treat it as a rule of thumb.
The five attributes that are most correlated with potential ability - that means the higher these attributes are the higher potential ability will be (and vice versa) - are:
Technique 0.49
Anticipation 0.47
First Touch 0.45
Composure 0.45
Passing 0.43
Good players invariably have good technique. This should be the first attribute you look at (ignoring position specific attributes for now). Anticipation, first touch, composure and passing round out the top five. Sometimes when you watch the match engine (ME) it's easy to see the effect of pace and acceleration on a match because you can see that some players are physically faster than others - and it's easy to assign that fact to the reason why you scored or conceeded. How do you observe technique? How do you measure composure?
Michael Lewis in his great book Moneyball, a story of revolution in baseball scouting, reported that young pitchers are measured on the speed of their fast ball rather than control or disguise because all you need is a speed gun. Control and disguise are much harder to measure, but ultimately far more important if you want a successful career in major league baseball.
So maybe when you think the ME is biased towards fast players what you are really doing is judging what is easy to perceive rather than what is really important. Just like going to a football game in real life you spend most of the time watching the ball - which means you miss the off the ball movement or anticipation of a striker. It's easy to witness a composed finisher in action - but how does composure manifest itself in your midfield general?
Out of interest the five attributes you should completely ignore when evaluating a regen because they have no bearing on potential ability are:
Influence 0.18
Natural Fitness 0.16
Long Throws 0.15
Flair 0.13
Aggression 0.09
You will often find regens with 17+ for long throws and natural fitness - and I think this analysis proves, if nothing else, that those attributes are meaningless in judging the quality of the player.
Sometimes when you see a regen you head gets turned by 18 to 20 scores in certain attributes. Depending on which attributes those scores represent it could mean absolutely nothing, or something very special. Just like the previous post on strikers, I conducted similar analysis but opened it up to all positions. This analysis therefore doesn't factor in position specific attribute importance (e.g. heading, marking and tackling for DCs). So just treat it as a rule of thumb.
The five attributes that are most correlated with potential ability - that means the higher these attributes are the higher potential ability will be (and vice versa) - are:
Technique 0.49
Anticipation 0.47
First Touch 0.45
Composure 0.45
Passing 0.43
Good players invariably have good technique. This should be the first attribute you look at (ignoring position specific attributes for now). Anticipation, first touch, composure and passing round out the top five. Sometimes when you watch the match engine (ME) it's easy to see the effect of pace and acceleration on a match because you can see that some players are physically faster than others - and it's easy to assign that fact to the reason why you scored or conceeded. How do you observe technique? How do you measure composure?
Michael Lewis in his great book Moneyball, a story of revolution in baseball scouting, reported that young pitchers are measured on the speed of their fast ball rather than control or disguise because all you need is a speed gun. Control and disguise are much harder to measure, but ultimately far more important if you want a successful career in major league baseball.
So maybe when you think the ME is biased towards fast players what you are really doing is judging what is easy to perceive rather than what is really important. Just like going to a football game in real life you spend most of the time watching the ball - which means you miss the off the ball movement or anticipation of a striker. It's easy to witness a composed finisher in action - but how does composure manifest itself in your midfield general?
Out of interest the five attributes you should completely ignore when evaluating a regen because they have no bearing on potential ability are:
Influence 0.18
Natural Fitness 0.16
Long Throws 0.15
Flair 0.13
Aggression 0.09
You will often find regens with 17+ for long throws and natural fitness - and I think this analysis proves, if nothing else, that those attributes are meaningless in judging the quality of the player.
Saturday, 23 August 2008
What to look for in a young striker
I've been playing with a data set from an FM08 save game. I thought it would interesting to interrogate the data to see if I could find anything of interest to give managers an edge in FML.
The first random thing I looked into was which attributes are most highly correlated with potential ability for a striker. Or, which attributes should I look for if I am trying to gauge the potential value of a young striker.
A simple correlation test between all attributes and potential ability revealed the following attributes as having the strongest relationship with potential ability:
1 Technique 0.60
2 Dribbling 0.57
3 Finishing 0.54
4 Passing 0.53
5 Long Shots 0.52
The number relate to the strength of correlation with 1.00 being totally positively correlated and -1.00 being totally negatively correlated. Although correlations can't explain causality they can prove a relationship. Which means by examining those five attributes for a player you can guess their potential ability - which in FML you either not know because you don't have the Judging Potential skill, or you do have it but it is expressed on a 1 to 5 star scale with a degree of inaccuracy built it. The higher these five figures are, the higher the potential ability will be.
Those attributes that have no relationship with potential ability include Influence, Aggression, Flair, Natural fitness and Long throws. The lesson here is regens with high values in these attributes could be complete busts.
The test was run on about 12,500 players so it should be robust - however the save game the data was extracted from the first week of the game and includes no regens.
The first random thing I looked into was which attributes are most highly correlated with potential ability for a striker. Or, which attributes should I look for if I am trying to gauge the potential value of a young striker.
A simple correlation test between all attributes and potential ability revealed the following attributes as having the strongest relationship with potential ability:
1 Technique 0.60
2 Dribbling 0.57
3 Finishing 0.54
4 Passing 0.53
5 Long Shots 0.52
The number relate to the strength of correlation with 1.00 being totally positively correlated and -1.00 being totally negatively correlated. Although correlations can't explain causality they can prove a relationship. Which means by examining those five attributes for a player you can guess their potential ability - which in FML you either not know because you don't have the Judging Potential skill, or you do have it but it is expressed on a 1 to 5 star scale with a degree of inaccuracy built it. The higher these five figures are, the higher the potential ability will be.
Those attributes that have no relationship with potential ability include Influence, Aggression, Flair, Natural fitness and Long throws. The lesson here is regens with high values in these attributes could be complete busts.
The test was run on about 12,500 players so it should be robust - however the save game the data was extracted from the first week of the game and includes no regens.
Subscribe to:
Posts (Atom)