Community » Hattrick Press

Safi Experiment

Written by Yndy_

How many specialists do you have in your team and how many goals do you score due to specialties? What is the optimal specialist strategy for a top team?

Safi Experiment was set up to attempt to find a specialties strategy superior to the ones commonly employed. Now it has finished and its time to look at the results.

According to the Unwritten Manual (17035223.1) (the creation of which I had contributed somewhat) and the fantastic studies of EngelRS4, the occurrence of goal special events is directly linked to the number of specialists in the team.

As you all know, building a team with specialties gives you an advantage over opponents and it's an advantage which is only paid for once at acquisition of the player since you don't pay a bigger salary for specialists. Best hattrick managers have long figured out that besides a skills strategy, they need to consider a specialties strategy when building their team. The most frequent strategy has been the "many Headers" strategy, with teams having up to 9 Headers in the team to maximize the chance of the Corner+Head goal event with up to 55% chance of occurrence per match.

Probably the second most used strategy was the "Maximize Quick & Head strategy" where a team would have 3-5 Quick players on Forward and Winger positions and 5-7 Head players on Midfielder, Winger and Defender positions. On average this strategy should lead to 30% chance of Quick event, 30% of Quick pass event and 35% chance of Corner + Head Event. I have experienced with this strategy on my main team Chimica Târnăveni (183500). In 174 league matches in Global seasons 49 to 60 (Romanian seasons 36 to 47), my team scored 13 Quick rush goals, 25 quick + pass goals and 61 Corner + Head goals with ratios of 7%, 14% and 35% of total matches and an average of 0.57 SE goals per league match.

These numbers are below the theoretical probabilities mainly because the EngelRS4 studies researched events of both teams per match and my data represent my team's goals per match. Still an extra 0.57 goals per match due to specialists was a pretty good result.

But what if an even better strategy is there? What if we truly maximize the number of specialists of one kind?

I chose to do a real experiment by maximizing the number of Unpredictable specialists in my second team Olympique Club de Safi (2015802), while also pursuing a long shots optimization strategy. What needs to be considered is that the conversion of events to goals for Unpredictable events is much lower than the conversion of Quick and Head events but I have covered this bias by comparing the actual number of goals (i.e. successful events). Also in order for the Unpredictable pass event to materialize, I had to train some passing for my team, which hindered the optimal training regime for a long shots team.

The statistical results were obtained with a team with 11 Unpredictable players over 70 league matches in Global seasons 60 to 64 (Morocco seasons 40 to 44) and my team scored 20 long pass unpredictable goals, 15 special unpredictable goals and 10 solo unpredictable goals. My team also conceded 3 goals due to negative unpredictable events. Scoring ratios relative to the number of games were 29%, 21%, 14% and -4% for a total net of 0.6 SE goals per league match.

I’m not satisfied with the sample size, it’s too low and a significant number of matches were played against weak or bot teams, but since I’m not extending my Platinum supporter, Olympique Club de Safi will be closed and this experiment will end.

So I will try to conclude on the basis of the existing data and the reasonable conclusion is that the “All Unpredictable” strategy appears to be marginally superior to the "Maximize Quick & Head” strategy. In the Safi Experiment, the results were only a little better but honestly the sample size is not large enough in order to claim the superiority of the tactic.

If you like the article, feel free to click "Like" and/or leave a comment. You can discuss the article on the forum: (17110586.1).

You can also read and comment this article on our FB page: (https://www.facebook.com/notes/hattrick-press/safi-experiment/1512540342151071/).

Previous article: Meeting the teams with best ratings from Hattrick International (20073).