The Sean Posted April 28, 2010 Share Posted April 28, 2010 I decided that I wanted to identify the attributes that most closely correlate to overall value on FM. Originally, I wanted to improve upon the simplistic Microsoft Excel-based scouting tool I created. What I found out was interesting in spots and reinforced by what we would think of each position in others. If you want the statistics I used to accomplish this, I will post that later on. For now, I want to be brief and share the exciting news about which attributes correlate to value on FM. What I can tell you is that for the nine positions I tested, taking the 10 most closely-correlated attributes to the mean, 85 of 90 attributes had a standard deviation from the mean of less than 1 out of 20 (5%). The remaining five had a standard deviation of just over 1 out of 20 (less than 6%). For the non-scientific, this means that these attributes are the ten most essential attributes to defining value on FM. Here goes. Enjoy. Goalkeepers 1. Rushing Out 2. Aerial Ability 3. Concentration 4. Command of the Area 5. Positioning 6. Communication 7. Composure 8. Handling 9. Decisions 10. Agility Outside-Backs 1. Anticipation 2. Teamwork 3. Tackling 4. Acceleration 5. Positioning 6. Concentration 7. Marking 8. Work-Rate 9. Stamina 10. Decisions Sweepers 1. Teamwork T2. Anticipation T2. Balance 4. Jumping 5. Tackling 6. Marking T7. Heading T7. Positioning 9. Concentration 10. Composure Center-Backs T1. Bravery T1. Positioning 3. Marking 4. Anticipation 5. Jumping 6. Heading 7. Strength T8. Determination T8. Tackling 10. Aggression Outside-Mids 1. Passing 2. Stamina 3. Off-the-Ball 4. Determination 5. Pace 6. Crossing 7. Dribbling 8. Technique 9. Anticipation 10. Decisions Defensive-Mids 1. Technique 2. Strength 3. Concentration 4. Creativity 5. Positioning 6. Composure 7. Tackling 8. Decisions 9. Passing 10. Work-Rate Center-Mids 1. Creativity 2. Passing T3. First-Touch T3. Teamwork 5. Technique T6. Decisions T6. Work-Rate 8. Stamina 9. Determination 10. Off-the-Ball Attacking-Mids T1. Flair T1. Off-the-Ball 3. Pace 4. Passing 5. Dribbling 6. Creativity 7. Acceleration 8. First-Touch 9. Technique 10. Decisions Strikers T1. Anticipation T1. Composure 3. Creativity 4. Pace 5. Strength T6. Agility T6. First-Touch T6. Passing 9. Decisions 10. Acceleration Link to post Share on other sites More sharing options...
MelbVictory Posted April 28, 2010 Share Posted April 28, 2010 Good work. I would have thought Finishing would be up tehre for strikers? All else seems about what is accepted as key attributes per posi. Link to post Share on other sites More sharing options...
The Sean Posted April 28, 2010 Author Share Posted April 28, 2010 Finishing was number 11 on the list for strikers and was the last be 'booted off' when whittling the list down to the top-10 attributes by position...and thanks. :o Link to post Share on other sites More sharing options...
The Sean Posted April 29, 2010 Author Share Posted April 29, 2010 First and last time I bump this for anyone's potential interest. Link to post Share on other sites More sharing options...
SFraser Posted April 29, 2010 Share Posted April 29, 2010 How do you mean "value"? Do you mean actual cash value, or do you mean positional value in terms of rating or some other measure? Some of these results are actually quite fascinating and what I personally would consider to be "key" attributes, but I don't know exactly what you are looking for nor doing here, and from what you said I suspect your method of investigation is flawed, despite producing very impressive results. Link to post Share on other sites More sharing options...
The Sean Posted April 29, 2010 Author Share Posted April 29, 2010 How do you mean "value"?Do you mean actual cash value, or do you mean positional value in terms of rating or some other measure? Some of these results are actually quite fascinating and what I personally would consider to be "key" attributes, but I don't know exactly what you are looking for nor doing here, and from what you said I suspect your method of investigation is flawed, despite producing very impressive results. By value, I mean the overall number or rating or ranking we would assign to a player compared to those who also play the same position. I chose the player sample group (25/position) by taking the top cash valued players for each position. As to what I was looking for; I wanted to know what attributes seemed to have the greatest impact on how highly or lowly we would end up rating a player. I didn't want my prejudices of what I believed to be important for that position to affect the outcome of my my study so I simply chose the most frequently identified attributes that FM already lists on the Tactics screen as my starting point. From there, I rated each player based on these attributes; I took the average of each attribute over the tested population and compared it to the mean (in this case, the average "value" rating each player received based on the tested attributes); I used two 3rd-party editors as my control by using the average (of the two) value they assigned the same players at the same position to identify variance between the rating system I created and those already in existence and widely accepted; and finally, I measured the standard deviation from the average of each attribute (over the tested population) to the mean (once again, the average "value" rating) to see which attributes most closely correlated with the expected rating across the entire population. As to what I was doing; I just recently finished reading Soccernomics (Szymanski and Kuper) and they used multiple regression analysis to identify many external factors affecting soccer today. I thought it would be fun to do the same with player attributes in FM. Not having the time or patience to use a sample of thousands, I chose a small sample of 25 per position. Basically, to see if my hypothesis was correct, I decided to look at 25 "great" (high cash value) players from each position and see what attributes they had in common and to what degree. This is a common practice in social science when we want to quantify qualitative data. Finally, as to your suspicion of my methodology; I don't have much to say to that. I would argue that the descriptions I have offered in these posts is consistent with widely accepted practices for finding correlation among data sets. I would be happy to publish my results to a 3rd-party website and post a link if there seems to be some folks looking for that. The only thing I can say to ease the troubled minds is that of course other attributes (besides the ones I have identified) affect a player's value. How well a striker finishes matters. However, according to what I was able to find, it didn't matter as much as these other attributes. I hope these responses answer your questions and allay your suspicions. If not, please demonstrate your own veracity on the subject so that I can better understand where you are coming from. I mean no disrespect, but claiming my inquiry was flawed before you got any responses from me comes across a bit harsh. I published these results for all to enjoy and to encourage questions about them. I am thankful you responded because others likely had the same questions as you. I am happy to dialogue with you and others about these results and any other FM topics. Link to post Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.