My sincere thanks to those of you that have contributed data over at Bearpuncher Labs for Major League Statistics! We’ve got 23 full submissions at this point, which is plenty to do some preliminary research. To reiterate, the workshop file provides a league in Bermuda where each team is identically staffed; all players have 10 in all stats, except for their “Team stat” which is set to 20. There are 36 teams for each of the 36 visible attributes a player possesses.
I took a sneaky peek at the submitted data, and the average league table was something I wanted to check up on. The results give us some food for thought.
+-----------------+--------------------+--------------------+--------------------+--------------------+--------------------+---------------------+--------------------+ | name | wins | draws | losses | gf | ga | gd | points | +-----------------+--------------------+--------------------+--------------------+--------------------+--------------------+---------------------+--------------------+ | Bravery | 17.681818181818183 | 10.545454545454545 | 6.7727272727272725 | 56.22727272727273 | 36.04545454545455 | 20.18181818181818 | 63.59090909090909 | | Heading | 16.761904761904763 | 10.619047619047619 | 7.619047619047619 | 53.666666666666664 | 37.42857142857143 | 16.238095238095234 | 60.90476190476191 | | Work Rate | 15.619047619047619 | 11.238095238095237 | 8.142857142857142 | 48.523809523809526 | 34.857142857142854 | 13.666666666666671 | 58.09523809523809 | | Dribbling | 15.476190476190476 | 11.047619047619047 | 8.476190476190476 | 52.714285714285715 | 39.76190476190476 | 12.952380952380956 | 57.47619047619048 | | Passing | 14.904761904761905 | 11.714285714285714 | 8.380952380952381 | 45.19047619047619 | 33.80952380952381 | 11.38095238095238 | 56.42857142857143 | | Determination | 15.238095238095237 | 10.714285714285714 | 9.047619047619047 | 48.04761904761905 | 37.857142857142854 | 10.190476190476197 | 56.42857142857142 | | Strength | 15.19047619047619 | 10.761904761904763 | 9.047619047619047 | 50.142857142857146 | 38.904761904761905 | 11.23809523809524 | 56.33333333333333 | | Aggression | 14.318181818181818 | 12.772727272727273 | 8.045454545454545 | 47.18181818181818 | 35.81818181818182 | 11.36363636363636 | 55.72727272727273 | | Jumping Reach | 14.285714285714286 | 11.428571428571429 | 9.285714285714286 | 49.76190476190476 | 40.714285714285715 | 9.047619047619044 | 54.28571428571429 | | Flair | 13.952380952380953 | 11.904761904761905 | 9.142857142857142 | 47.04761904761905 | 39.23809523809524 | 7.80952380952381 | 53.761904761904766 | | Natural Fitness | 14.142857142857142 | 11.095238095238095 | 9.714285714285714 | 46.80952380952381 | 38.19047619047619 | 8.61904761904762 | 53.523809523809526 | | Teamwork | 13.619047619047619 | 12.095238095238095 | 9.238095238095237 | 43.714285714285715 | 36.142857142857146 | 7.571428571428569 | 52.95238095238095 | | Penalty Taking | 13.238095238095237 | 12.380952380952381 | 9.380952380952381 | 43.19047619047619 | 37 | 6.19047619047619 | 52.09523809523809 | | Corners | 13.095238095238095 | 11.571428571428571 | 10.333333333333334 | 44.714285714285715 | 40.38095238095238 | 4.333333333333336 | 50.857142857142854 | | Free Kicks | 12.80952380952381 | 12.19047619047619 | 10 | 43.19047619047619 | 38.095238095238095 | 5.095238095238095 | 50.61904761904762 | | Long Throws | 12.571428571428571 | 12.19047619047619 | 10.238095238095237 | 44.23809523809524 | 40.57142857142857 | 3.6666666666666714 | 49.904761904761905 | | Off The Ball | 12.761904761904763 | 10.666666666666666 | 11.571428571428571 | 47 | 43.80952380952381 | 3.19047619047619 | 48.952380952380956 | | Leadership | 12.19047619047619 | 11.523809523809524 | 11.285714285714286 | 41.095238095238095 | 39 | 2.095238095238095 | 48.095238095238095 | | Balance | 12.136363636363637 | 11.181818181818182 | 11.681818181818182 | 41.5 | 40.22727272727273 | 1.2727272727272734 | 47.590909090909086 | | Long Shots | 11.857142857142858 | 11.714285714285714 | 11.428571428571429 | 40.666666666666664 | 40.19047619047619 | 0.4761904761904745 | 47.285714285714285 | | Pace | 11.666666666666666 | 11.619047619047619 | 11.714285714285714 | 41.714285714285715 | 43.333333333333336 | -1.6190476190476204 | 46.61904761904762 | | Stamina | 11.333333333333334 | 12.142857142857142 | 11.523809523809524 | 39.38095238095238 | 41 | -1.6190476190476204 | 46.14285714285714 | | Technique | 10.571428571428571 | 11.761904761904763 | 12.666666666666666 | 38.80952380952381 | 42.476190476190474 | -3.6666666666666643 | 43.47619047619048 | | Composure | 10.136363636363637 | 10.681818181818182 | 14.181818181818182 | 37.18181818181818 | 43.5 | -6.31818181818182 | 41.09090909090909 | | Vision | 9.619047619047619 | 11.238095238095237 | 14.142857142857142 | 37.857142857142854 | 46.04761904761905 | -8.190476190476197 | 40.09523809523809 | | Crossing | 9.857142857142858 | 10.428571428571429 | 14.714285714285714 | 35.95238095238095 | 45.142857142857146 | -9.190476190476197 | 40 | | Tackling | 8.857142857142858 | 12.80952380952381 | 13.333333333333334 | 33 | 40.523809523809526 | -7.5238095238095255 | 39.38095238095238 | | Finishing | 9.761904761904763 | 10.095238095238095 | 15.142857142857142 | 35.80952380952381 | 45.23809523809524 | -9.42857142857143 | 39.38095238095238 | | Anticipation | 9.136363636363637 | 11.954545454545455 | 13.909090909090908 | 35.13636363636363 | 43.04545454545455 | -7.909090909090914 | 39.36363636363637 | | First Touch | 8.80952380952381 | 11.380952380952381 | 14.80952380952381 | 34.142857142857146 | 45.523809523809526 | -11.38095238095238 | 37.80952380952381 | | Acceleration | 9.136363636363637 | 10.045454545454545 | 15.818181818181818 | 37.59090909090909 | 49.72727272727273 | -12.136363636363633 | 37.45454545454545 | | Concentration | 8.571428571428571 | 11.19047619047619 | 15.238095238095237 | 34.095238095238095 | 45.76190476190476 | -11.666666666666664 | 36.904761904761905 | | Marking | 8 | 10.857142857142858 | 16.142857142857142 | 32.714285714285715 | 45.19047619047619 | -12.476190476190474 | 34.85714285714286 | | Positioning | 7.238095238095238 | 11.619047619047619 | 15.666666666666666 | 31.904761904761905 | 46.666666666666664 | -14.76190476190476 | 33.333333333333336 | | Agility | 7.090909090909091 | 10.954545454545455 | 16.954545454545453 | 31.136363636363637 | 48.95454545454545 | -17.818181818181817 | 32.22727272727273 | | Decisions | 5.428571428571429 | 8.571428571428571 | 21 | 24 | 51.76190476190476 | -27.76190476190476 | 24.857142857142854 | +-----------------+--------------------+--------------------+--------------------+--------------------+--------------------+---------------------+--------------------+
I wasn’t really expecting there to be such a smooth gradient of point averages. Bravery may well be an underappreciated stat, as it was last year. The most eye-catching thing, though, is Decisions. The team with 20 in Decisions is by far the worst team in the league, and significantly worse than the next worst team, over a large amount of data, 805 match days for each club.
I had similar results last year but ultimately threw the data out because I feared I was not meticulous enough in ensuring a level playing field. I have no such fear this year and stand by the data. Now, when Decisions is that far off the pace of the rest of the teams, it got me wondering, would a team with 10 in all attributes and 1 in Decisions outperform this team with 10 in all attributes and 20 in Decisions? And so I began an experiment in private.
The initial data says yes, that’s absolutely the case. Decisions is a broken attribute.
+--------------+--------------------+--------------------+--------------------+--------------------+--------------------+---------------------+--------------------+ | name | wins | draws | losses | gf | ga | gd | points | +--------------+--------------------+--------------------+--------------------+--------------------+--------------------+---------------------+--------------------+ | Decisions 1 | 18 | 13.285714285714286 | 6.714285714285714 | 54.42857142857143 | 37.57142857142857 | 16.85714285714286 | 67.28571428571429 | | Decisions 6 | 16.571428571428573 | 13.714285714285714 | 7.714285714285714 | 54.142857142857146 | 38.857142857142854 | 15.285714285714292 | 63.42857142857144 | | Decisions 10 | 17.142857142857142 | 11.285714285714286 | 9.571428571428571 | 51.285714285714285 | 38.857142857142854 | 12.42857142857143 | 62.714285714285715 | | Decisions 2 | 16.857142857142858 | 12 | 9.142857142857142 | 52.142857142857146 | 38.42857142857143 | 13.714285714285715 | 62.57142857142857 | | Decisions 3 | 16.571428571428573 | 12.571428571428571 | 8.857142857142858 | 52 | 37.142857142857146 | 14.857142857142854 | 62.28571428571429 | | Decisions 4 | 16.714285714285715 | 12.142857142857142 | 9.142857142857142 | 47.57142857142857 | 36.285714285714285 | 11.285714285714285 | 62.28571428571429 | | Decisions 9 | 15.857142857142858 | 12.857142857142858 | 9.285714285714286 | 51.285714285714285 | 41 | 10.285714285714285 | 60.42857142857143 | | Decisions 5 | 15.285714285714286 | 13.857142857142858 | 8.857142857142858 | 50.714285714285715 | 37.714285714285715 | 13 | 59.71428571428572 | | Decisions 8 | 15 | 11.571428571428571 | 11.428571428571429 | 49.714285714285715 | 44.857142857142854 | 4.857142857142861 | 56.57142857142857 | | Decisions 7 | 14 | 14.428571428571429 | 9.571428571428571 | 51.142857142857146 | 42.285714285714285 | 8.857142857142861 | 56.42857142857143 | | Decisions 12 | 14 | 13.142857142857142 | 10.857142857142858 | 47.857142857142854 | 42.285714285714285 | 5.571428571428569 | 55.14285714285714 | | Decisions 11 | 14.428571428571429 | 11.571428571428571 | 12 | 47.714285714285715 | 42.142857142857146 | 5.571428571428569 | 54.857142857142854 | | Decisions 13 | 12 | 11.571428571428571 | 14.428571428571429 | 40.57142857142857 | 46.142857142857146 | -5.5714285714285765 | 47.57142857142857 | | Decisions 14 | 10 | 12 | 16 | 37.714285714285715 | 49.42857142857143 | -11.714285714285715 | 42 | | Decisions 16 | 8.285714285714286 | 13.142857142857142 | 16.571428571428573 | 38.57142857142857 | 53.285714285714285 | -14.714285714285715 | 38 | | Decisions 17 | 8.857142857142858 | 11.428571428571429 | 17.714285714285715 | 34.57142857142857 | 50.142857142857146 | -15.571428571428577 | 38 | | Decisions 15 | 8.571428571428571 | 11 | 18.428571428571427 | 34.285714285714285 | 51 | -16.714285714285715 | 36.714285714285715 | | Decisions 20 | 6.285714285714286 | 12.714285714285714 | 19 | 31.142857142857142 | 52.285714285714285 | -21.142857142857142 | 31.57142857142857 | | Decisions 18 | 5.857142857142857 | 12.428571428571429 | 19.714285714285715 | 31 | 52.142857142857146 | -21.142857142857146 | 30 | | Decisions 19 | 6 | 10.714285714285714 | 21.285714285714285 | 31.142857142857142 | 57.57142857142857 | -26.428571428571427 | 28.714285714285715 | +--------------+--------------------+--------------------+--------------------+--------------------+--------------------+---------------------+--------------------+
This is pretty straightforward. The teams with 10 Decisions or less are all in the top 10. The teams with 15 or more are all at the bottom of the table, and by a significant amount. The team with 1 in Decisions averages 3 times the number of wins as the teams with 18-20, and 3 times fewer losses. The team scores nearly 75% more goals, while conceding 39% fewer. Further recorded data indicates that teams with lower Decisions control the possession game, and complete a higher percentage of passes. That should not be. The common-sense explanation for decision-making ability would manifest as “If you’re not good at making decisions, you’ll give the ball up more.”
Compare that, though, with this:
The top 6 sides in pass completion ratio possess Decision attributes of 1 through 6, in that order.
And the teams with high decision-making ability are really bad at keeping possession of the ball.
Note that the bottom 7 teams in possession have the 7 highest Decisions attributes in the league, and are at the bottom 7 positions in the league table at the end of the season.
Clearly, something is not right.
This data file is also available on Steam Workshop, if you wish to contribute data to the experiment please do! The Bearpuncher Labs page is open and the amount of data requested is significantly less than Major League Statistics.
11 Comments
Manrico Bugeja (@manricob) · December 16, 2017 at 12:49 pm
as you rightly say,”Clearly, something is not right.”.. and there I was, giving decisions such importance! Thanks for the work
bluesoul · December 16, 2017 at 5:44 pm
Thank you, friend. There’s more experiments getting started with a more real-life scenario that should be more definitive than using the stat in isolation.
davidwolf06 · December 18, 2017 at 12:26 am
hey can you share the results in another download format i.e. media fire or something similar?
Thanks!
bluesoul · December 18, 2017 at 4:19 am
What exactly are you looking for to be shared? I’m certain I can provide it, just let me know what you’re after.
Hess · December 18, 2017 at 1:33 am
Which makes me wonder if Decisions isn´t just: How often do I take risky plays? High decisions = High risky plays
bluesoul · December 18, 2017 at 4:15 am
The more I consider, the more it would line up with what I’ve observed in this and the proceeding article. If Decisions actually correlates to likelihood to attempt a risky pass, but a direct correlation rather than the inverse it would logically be, it would explain the increase in both possession and pass completion ratios.
davidwolf06 · December 18, 2017 at 11:50 am
hey sorry i meant the table in a format so that the numbers are aligned ect as it is difficult to understand the two tables above.
also can you provide the save game or database which you got these results?
Sarcastic Turtle · March 18, 2018 at 11:44 am
It’s your tactics dude you have to adjust your tactics accordingly.
Football Manager Expert #237553 · January 31, 2019 at 12:23 pm
This is a FLAWED research on so many levels so i had to post this as a true FM fanatic for future reference. Based on what OP has posted here and on reddit, it seems that OP used a fixed 105 CA/PA for all the players while changing the variable stat as mentioned above. Unfortunately, FM uses a system of weighted attributes. MEANING that certain stats like DETERMINATION and AGGRESSION and FLAIR take up 0 CA cost. BRAVERY is also a low cost CA. Also depending on the natural position of the player, certain stats will be more weighted for certain positions. For instance, TACKLING might take lot of CA for a defender but very little for a forward. There are multiple natural position variation to the equation as well as footedness. And the way FM works via editor is that when you set a FIXED CA, and change the stat as mentioned above, it will calculate whether the edited stats are equal to FIXED CA and if not, will either INCREASE or DECREASE all stats EQUALLY, and show IN GAME after some games (also shown as rounded INTEGERS not fractions) IN CONCLUSION, just from looking at the OP’s list I can see that it’s generally similar to LOWEST CA COST TO HIGHEST CA COST LIST, just as I’ve explained. So what this actually means is that the team with 20 DETERMINATION will have vastly higher stat than the team with 20 ACCELERATION (even if not actually shown in game due to FRACTIONAL STAT system-explained below) IF, OP wants really REALLY want to find out if stats have a difference, then he must take into account that every VARIABLE stat increase will alter the CA, all players must only know one natural position, and figure out the exact CA number for every single position which will all be different too-this is very time consuming. There is also the consideration of fractional stat in FM meaning that players don’t actually have integer stats, but attributes are only SHOWN to be integers. meaning the player could have 20 in one stat and 9.6s or 10.4s in all other stats and have vastly different CA, but will be shown as having the exact same stat IN GAME, which is one 20 and all 10s. So OP will need to figure out High/low CA for the same stat use the median CA. This will result in teams with different CAs but if the research is finding out which stat is the best, than that’s how it should have been done. On a side note, this is the reason why decision 1 team won against decision 20 team too. Decision is highly weighted stat, which would have resulted in overall lower general stat for the higher decision team at a fixed CA. meaning something like 20 decision 8.6 in all stats versus 1 decision and 11.4 in all stats (in game 20/9 vs 1/11-btw this is an exaggeration not actual) so it could seemingly be looking 1-2 points insignificant but VERY SIGNIFICANT almost 3-4 difference in actuality. And obviously 20 decision cannot overcome even 1-2 difference in ALL OTHER ATTRIBUTES. that’s about a difference between one level of league. OH forgot to mention that attribute of 1 does not equal crap in game either or 20 as perfect. all stat numbers are converted and added to the in game engine number that crunches out unknown number. that’s why you see some 1/1 finish/composure cracking at goals and scoring and 20/20 player miss a sitter. all players seem to have some innate ability at all point range, meaning that 1 decision and 20 decision could actually be 81 and 100 in the game engine for all we know.
psychotic · February 9, 2019 at 12:53 am
Yeah. That changes the results of the test: Instead of Decisions being broken in the match engine, it’s just that Decisions is the most worthless stat in the game. It’s so overvalued in the CA calculation algorithm that players with Decisions 1 will be better, because their CA will be distributed in more useful stats.
psychotic · February 9, 2019 at 12:57 am
That just means Decisions isn’t necessarily broken in the match engine, it’s just so overvalued in the CA algorithm that it’s worthless.