PStats

From Hattrick
INTERWIKI

· English · Magyar ·

Original Info

PStats Index[edit]

Antecedents[edit]

PStats (Acronym of PeasoStats) seeks to be an appreciation coefficient for the efficiency of a team in Hattrick, depending on the obtained qualifications in a match, and to bring the results as close as possible to what the game engine considers to be this team’s efficiency.

This coefficient does not try to be a tool to solve the score in a match, but tries to be an indicator of the exact efficiency of a team and a statistic to show which the evaluation tendency is.

The reference point that has been taken to make this research is the statistic section: My games > Best played games > All Team. There, we can find a classification of the 15 games in which HT considers that the team has had the best efficiency, and compares it with percentages. A consideration that has been taken is if these statistics really reflect what the engine considers to evaluate a game, but we have supposed that this is correct, as if it was not, the statistics would be useless and would give wrong information.

The HatStats index, due to its simpleness and universality has become the start point. For not initiate, HatStats is based on the numerical equivalence of the Hattrick qualifications in every zone of the field. i.e. 1=Disastrous (very low) and increasing 1 by 1 every level (very low, low, high, very high), so that we have 4 points for a full level and 80 for every zone of the field. As HT gives 3 qualifications in defense and attack and only one in midfield (MF), this one is multiplied by 3. Adding all these 7 marks, we get a range from 9 to 720.

The basic problem in HatStats is its equability when comparing the specific weights in every zone of the field, as the general feeling is that players and scores in the games make us feel that not all lines have the same weight and cannot have the same consideration, as in a standard tactic like 4-4 -2 it cannot have the same value a winger (where only one player plays) than an inner (where 2 players usually play).

Another advantage of this new index is that, as it is studied with percentages to each one of the obtained qualifications, its absolute weight indicates its global team potential, able to translate it into HT qualification terms.

Percentages[edit]

These antecedents brought us to test with different values, using HT statistics of “My best games” as a reference point and the score comparison, we arrived at these conclusions:

   * Midfield has a higher qualification in HT than in HatStats.
   * Attack has a higher qualification than defense.
   * Central attack and defense is more important then the one in the sides.

These conclusions, numerically, are:

   * MF has a specific weight of 46%
   * Attack has a specific weight of 32%
   * Defense has a specific weight of 22%
   * Left side has a specific weight of 30%.
   * Central zone has a specific weight of 40%.
   * Right side has a specific zone of 30%.


Model validation[edit]

Now, we had our dearest formula, but we had to test it empirically. We asked a high number of users from different divisions and different way of playing to give us their HT statistics and we started to process all data we received.

This testing was based on the PStat and HatStats index of every match they sent us, its conversion into a percentage like the one in HT and comparing its differences (we will not deepen and not publish more specific data, as we have a privacy policy with those users ho shared their data with us).

And exactly, the results we got made us see we were on the right way. As an example, here you have some of the final results:

Average of the 15 matches

   * PStats medium guess percentage: 99.51%
   * HatStats medium guess percentage: 97.7%
   * PStats minimum deviation: 99.68%
   * HatStats minimum deviation: 98.6%
   * PStats maximum deviation: 99.21%
   * HatStats maximum deviation: 96.25%

Single matches

   * PStats Maximum deviation: 98.4%
   * HatStats Maximum deviation: 91.3%


Conclusions[edit]

We will not talk about consequences or repercussions about these data and index, as probably we are not the right ones to do it, due to being involved in this project. But we ask you all to open a discussion so that we can deepen on this results we are posting.

We know it is not a discovery, as it is a confirmation of a general feeling that most of the players in HT have. But the goal of our labor has been to evaluate and put down in numbers what had never been written down before.

We also know it is not a discovery, as some users that some users have informed that they have been working in similar percentages, but we have thought it was a nice idea to show it to all the HT World, so that they can have a tool to help them more and more to take their own decisions.

We want to thank all these users who kindly have given us their data, without whom this study would have been impossible.


This study has been developed by MacAre and Peaso.

English translation by Sowjon

See also[edit]