Not Premier League proven

Not premier league proven, a phrase used by pundits, fans and insiders to suggest that a particular foreign player may not cut it in the British Premier League. Like all pieces of conventional wisdom, i wanted to take it on and separate myth from truth.

Too long, didn't read version:
Most players need some time to adapt, but they do more than ok if they get time on the field. "Not premier league proven" is overstated.

Long, tedious yet complete version:
I started out by looking at every single transfer for offensive players in the last five seasons that are coming to the British Premier League for the first time and looked at their output in Goals and Assists in accordance to the minutes they played and compared it to the last season they had before coming to England.

Why offensive players? Offensive players (Center forwards, Wingers and Attacking Midfielders) have the job of producing results in the last third of the pitch, fans know what they want out of them, a center forward that doesn't score is a scourge (Torres) instead of your average defender or midfielder who are not expected to come up with goals. Stats like "key passes" or completed pass % in the last third are great but extremely hard to come across for non insiders, and are very much non existent for unpopular leagues. Unless someone here has an OPTA account we have to do with goals, assists and minutes played.

Besides, i'm comparing how they did their last season before arriving and not overall productivity. Unless some factor (such as a tough Premier League) changes up their game in a significant way, these players are expected to produce such as were doing up till that moment.

Without further ado, the data for the last 5 seasons:
(Note: Goal Legacy crops up images, i decided to leave the links instead)

Season 2010/2011
https://2img.net/r/ihimg/photo/my-images/861/1011n.jpg/

Season 2009/2010
https://2img.net/r/ihimg/photo/my-images/9/0910z.jpg/

Season 2008/2009
https://2img.net/r/ihimg/photo/my-images/138/0809o.jpg/

Season 2007/2008
https://2img.net/r/ihimg/photo/my-images/11/0708he.jpg/

Season 2006/2007
https://2img.net/r/ihimg/photo/my-images/526/0607f.jpg/

Age = the age the player had that season
GpA = short for Goals plus Assists
Minutes = The minutes that player had that season on all competitions with his club, not including friendlies or international appearances
Minutes played p/GpA = The minutes they player had on the pitch divided by the "GpA" number. The number is the average amount of minutes that player took to score or get an assist.
GpA p/game = Their "Minutes played p/GpA" number used as a divisor by 90 (90/Minutes played p/GpA). This number is the average that player performed on any given match over 90 minutes, as an example a .5 "GpA p/game" means that player scored or assisted once every two games.
Performance % = The comparison of the players "GpA p/game" of his debut Premier League season compared to his last "foreign" season by a simple rule of three. For example a 68% means that player only managed to reproduce 68% of his offensive game to the Premier League. Any number over 100% means that player did even better that his last season and "overproduced" what was expected of him.

Bolded percentages are the teams average. All stats taken from transfermarkt.de. Red numbers are the season average.

Taken the whole 5 seasons average we get 123% for performance, which would suggest the average transfer overproduced by 23% on their debut season... However there was still much disparity, some transfers were intended as experienced players who were expected to bring in short term results, other were youth players who were in for the long haul. Some players had fantastic seasons which brought up the average up too much.

I decided then to separate the players into those who had some good time on the pitch and those who didn't, using 1000 minutes as my magic number to decide. Why 1000 minutes? It's a good simple round number that equals around 11 matches, i arbitrarily decided that was a good measure to decide if a player had been offered a decent amount of minutes and a real chance to crack into the team.

After that i rounded up and separated the players further by grouping them into a range of "performance percentages". The range goes from 0% to 300+%, i put the numbers into a graph.

Here are the numbers and graph for those players over 1000 minutes played:
https://2img.net/r/ihimg/photo/my-images/546/overry.jpg/

Numbers and graph for under 1000 minutes played:
https://2img.net/r/ihimg/photo/my-images/705/undermd.jpg/

Conclusions and closing Remarks:
Players who get over 1000m perform by an average of 78% compared to their previous year. Take into account a transfer which includes a change in country, language, culture, weather, friends and a difference in tactics or pace and you can see where the under performance comes from. A number close to 80% would get a pass in any school and suggests players do a remarkable job in fitting in into their new surroundings. The phrase "Not premier league proven" is thus another piece of conventional rhetoric which is blown out of proportion and not very useful.

It is however a very different story for players who get less than 1000m. Most are youngsters, flops or squad players who from very early on are expected to under perform. Their performance average is a pity full 9% compared to their last year. It would be interesting to know if the lack of playing time makes them perform poorly or the fact that they are under performing makes them get less minutes. The answer is most probably a cross between the two.

To finish i just want to say this project took me MONTHS to finish, i easily poured around 20 hours from idea to the finish line. In any project with this many variables there is bound to be mistakes, i know i included a pair of players who had already been in the premier league before in the 10/11 season. You are more than welcome to point them out but please do not chastise me too hard for them or discredit the whole thing.

Comments are more than welcome. What do you think of the whole thing?

Bonus Round:
If you want to predict how your clubs new transfer will do in the coming season use these functions:
Over 1000m (players you believe will be starters):
F(x)=[90/(Minutes Played last season/GpA last season)]*.78

Under 1000m (youngsters or bench players):
F(x)=[90/(Minutes Played last season/GpA last season)]*.05

The number you get will be the predicted "GpA p/game" stat. Fun!

Must of taken forever to compile all this lol....Good job

Fascinating numbers. And great job! Now one "only" needs to actually draw conclusions from this work. I shall ponder a bit about it.

Thank you both, it did indeed take a very very long time .

There is actually a ton of stuff i can do with the stats that i have already started planning, like how useful rotation is.

yourself bro

Anyone else??

I think it might be interesting to break this up in the source leagues, to see players of which league adapt best. From when I scanned the tables, I first looked a bit more closer to the BL players, albeit there weren't too many of them), but there it looks as if it depended on the season, too. In one season the BL imports looked pretty abysmal, while in others things looked better.

chinomaster182 wrote:
Taken the whole 5 seasons average we get 123% for performance, which would suggest the average transfer overproduced by 23% on their debut season...
Players who get over 1000m perform by an average of 78% c

It is however a very different story for players who get less than 1000m. Most are youngsters, flops or squad players who from very early on are expected to under perform. Their performance average is a pity full 9% compared to their last year.

Very interesting, but I see a contradiction with your numbers. I don't think it's possible that the total add up gives a 123% in performance while the breakdown between >1.000 minutes and <1.000 minutes give 78% and 9%. Something must be wrong there.

Impressive. It's going to take me awhile to digest all of this though.

free_cat wrote:

chinomaster182 wrote:
Taken the whole 5 seasons average we get 123% for performance, which would suggest the average transfer overproduced by 23% on their debut season...
Players who get over 1000m perform by an average of 78% c

It is however a very different story for players who get less than 1000m. Most are youngsters, flops or squad players who from very early on are expected to under perform. Their performance average is a pity full 9% compared to their last year.

Very interesting, but I see a contradiction with your numbers. I don't think it's possible that the total add up gives a 123% in performance while the breakdown between >1.000 minutes and <1.000 minutes give 78% and 9%. Something must be wrong there.

what i did there (78% and 9%) was get an average out of the performance range with the most players. 66.66%-100% for the first and 0%-33.33% for the second.

the average just divided by two camps was still too high to represent the normal transfer (guys like pienaar with 1100% brought it up).

i think it would be interesting if i showed you guys the list of players on both sides, however im typing on my phone right now.

thank you everyone for your kind comments

Alright, as promised here are the names, minutes and percentages for the players. First those over 1000m:

Name Minutes Performance %
Marouane Chamakh 2656 131%
Jermaine Pennant 2569 134%
Moussa Dembele 2227 111%
Javier Hernandez 2634 72%
Pablo Barrera 1330 18%
Victor Obinna 2493 173%
Luis Suarez 1102 46%
Rafael Van der Vaart 2719 112%
Edin Dzeko 1314 68%
David Silva 3833 83%
Mario Balotelli 1720 80%
Asamoah Gyan 2161 89%
Nikola Zigic 1886 88%
Jean Beausejour 1328 235%
Aleksandr Hleb 1413 244%
Elliot Grandin 1558 37%
Maxi Rodriguez 1211 62%
Yuri Zhirkov 1779 59%
Diniyar Bilayaletdinov 2109 246%
Björn Helge Riise 1341 53%
Guillermo Franco 1376 274%
Alessandro Diamanti 2033 75%
Chung-Yong Lee 2754 76%
Ivan Klasnic 1579 56%
Aruna Dindane 1922 210%
Frederic Piquionne 3059 106%
David Hoillett 1278 0%
Nikola Kalinic 2055 50%
Jozy Altidore 1575 255%
Christian Benitez 2268 42%
Andrey Arshavin 1193 211%
Samir Nasri 3543 50%
Albert Riera 2645 142%
Carlos Villanueva 1189 41%
Robinho 3474 73%
David Di Michele 2177 67%
Roman Pavlyuchenko 2180 90%
Jonas Gutierrez 2443 70%
Amr Zaki 2283 109%
Djibril Cisse 2270 78%
Marc Antoine Fortune 1514 294%
Nani 2543 171%
Claudio Pizzaro 1251 48%
Florent Malouda 2520 52%
Andriy Voronin 1583 86%
Fernando Torres 3582 238%
Ryan Babel 2534 154%
Eduardo 1851 63%
Carlos Vela 1179 368%
Steven Pienaar 2984 1125%
Roque Santa Cruz 3585 360%
Tuncay Sanli 2391 83%
Martin Petrov 3305 190%
Olivier Kapo 2022 122%
Salomon Kalou 2855 75%
Andriy Shevchenko 3423 72%
Julio Baptista 1778 156%
Dirk Kuyt 3728 30%
Bernardo Corradi 1930 119%
Damarcus Beasley 1238 106%
Dimitar Berbatov 3841 91%
Obafemi Martins 3780 94%
Benni McCarthy 4025 171%
Shabani Nonda 2147 70%
Niko Kranjcar 1453 9%
Antonio Valencia 1383 100%
Carlos Alberto Tevez 1963 86% (est)

Now less than 1000m:
Name Minutes Performance %
Eidur Gudjohnsen 547 116%
Hatem Ben Arfa 171 102%
Bebe 339 N/A
Mauro Boselli 827 11%
Robert Pires 389 61%
Apostolos Vellios 34 0%
Magaye Gueye 241 172%
Ruben Rochina 120 0%
Matt Derbyshire 828 65%
Rodrigo 717 N/A
Sergey Kornilenko 182 0%
Gabriel Obertan 498 133%
Mame Biran Diouf 95 106%
Andriy Voronin 384 43%
Landon Donovan 890 62%
Stefano Okaka 482 67%
Eddie Johnson 116 0%
David Elm 531 104%
Eidur Gudjohnsen 546 200%
Ilan 573 248%
Fabio Daprela 752 235%
Marcelo Moreno 730 8%
Quincy Owusu-Abeyie 395 118%
Yildiray Bastürk 45 0%
Elrio Van Heerden 63 0%
Boudewijn Zenden* 615 71%
Amr Zaki 230 0%
Kamel Ghilas 765 38%
Geoffrey Mujangi Bia 118 0%
Stefan Maierhofer 276 35%
Manucho 907 35%
Ricardo Quaresma 179 192%
David N'Gog 603 274%
Jo 588 87%
Savio Nsereko 271 174%
Walter Lopez 114 0%
Diego Tristan 834 172%
Giovani Dos Santos 577 23%
Peter Lövenkrands 644 273%
Ignacio Gonzalez 371 0%
Xisco 380 42%
Marvin Emnes 610 55%
Hugo Rodallega 971 42%
Ariza Makukula 294 48%
Riga Mustapha 542 0%
Ebi Smolarek 249 124%
Juan Carlos Menseguez 350 36%
Daniel Braaten 528 227%
Maceo Rigters 212 0%
Alfonso Alves 788 66%
Rolando Bianchi 843 114%
Geovanni 435 342%
Valeri Bojinov 52 0%
Emmanuel Villa 978 57%
Mile Sterjovski 694 22%
Rade Prica 211 74%
Mauro Zarate 616 101%
Daniël de Ridder 672 130%
Garry O'Connor 880 150%
Henrik Larsson 914 31%
Christopher Eagles 95 783%
Emile Mpenza 837 156%
Lee Dong-Gook 321 0%
John Carew 906 168%
Clint Dempsey 316 51%
Vincenzo Montella 475 195%
Duoala 237 0%
Julius Aghahowa 314 0%
Kepa 177 61%
Tamas Priskin 914 80%

See a pattern?

rwo power wrote:I think it might be interesting to break this up in the source leagues, to see players of which league adapt best. From when I scanned the tables, I first looked a bit more closer to the BL players, albeit there weren't too many of them), but there it looks as if it depended on the season, too. In one season the BL imports looked pretty abysmal, while in others things looked better.

Yeah i was definitively thinking about something like that, the most surprising thing though? Looks like it doesn't really matter much where they come from. Seems like quality is quality.

chinomaster182 wrote:

free_cat wrote:

chinomaster182 wrote:
Taken the whole 5 seasons average we get 123% for performance, which would suggest the average transfer overproduced by 23% on their debut season...
Players who get over 1000m perform by an average of 78% c

It is however a very different story for players who get less than 1000m. Most are youngsters, flops or squad players who from very early on are expected to under perform. Their performance average is a pity full 9% compared to their last year.

Very interesting, but I see a contradiction with your numbers. I don't think it's possible that the total add up gives a 123% in performance while the breakdown between >1.000 minutes and <1.000 minutes give 78% and 9%. Something must be wrong there.

what i did there (78% and 9%) was get an average out of the performance range with the most players. 66.66%-100% for the first and 0%-33.33% for the second.

the average just divided by two camps was still too high to represent the normal transfer (guys like pienaar with 1100% brought it up).

i think it would be interesting if i showed you guys the list of players on both sides, however im typing on my phone right now.

thank you everyone for your kind comments

Instead of adding up the % and then dividing by the number of players in each group, what you should do is divide the total minutes played for the total GpA of all the players and then get the % performance compared to last seasons total minutes played divided by last seasons GpA for all the players.

That would make much more sense.