Bushnell Elite 3200 5-15 Review

Hattrick - Hatstats or Ratings?

Laiho-NH was talking yesterday with regards to whether the Hattrick game engine to work with Hatstats or evaluations that the public report of every game. As you know

Hatstats proceed by transforming the skill areas in the field of numerical values \u200b\u200bby assigning a "disastrous low" value of 1, to "disastrous low" value of 2 and so on (for a table see http: / / wiki.hattrick.org / wiki / HatStats ).

Reasoning with the evaluation leads to intermediate values \u200b\u200bassigned to the scissors on the description, so "disastrously low", which corresponds to range from 1.00 to 1.25 is assigned the intermediate value of 1.125, a 'disastrous low' 1375, and so on, forming this table:

Reasoning with one or the other is not the same thing, the progressions are very different, such as "insufficient very high" is four times the "tremendous very low" with the hatstats (20 is four times 5), instead of evaluations, the ratio is 5875 / 2215 = 2.76 times.

In terms of possession of a team that changes its value Midfield, while Team 2 is fixed to "tremendous very high" we have these differences with the reasoning or the Hatstats Ratings

that you chart

show that the calculation with the evaluations than with Hatstats, overstates the holding of a team's midfield for values \u200b\u200blower than the Team 1 Team 2 and back to underestimate the values \u200b\u200bof upper midfield.

E Hattrick engine with which of the two modes right? At the suggestion of the same
Laiho-NH to go and see the values \u200b\u200bhave actually made the field, I took un campione di 30 partite a caso (10 di X serie, 10 di IX e 10 di VIII) con l'unica condizione che il possesso tra primo e secondo tempo non variasse oltre il 5%.

Questi i dati raccolti da Hattrick:

e queste le stime con le Hatstats

e con le Valutazioni (per calcolare le quali abbiamo dovuto metterci nei punti estremi dei rispettivi range per calcolare un minimo e un massimo di possesso per il Team 1)

Se restringiamo il confronto alla differenza tra le stime con le Hatstats e le stime con le valutazioni rispetto ai valori di Possesso reali vediamo subito come ragionare con le evaluations consistently leads to an underestimation of possession for the first team when the same midfield is superior to that of Team 2 (in red) and back to an overestimate of the midfield when the team is less than 1.

Moreover, more than the midfield is higher or lower, the greater the difference between estimated and actual values \u200b\u200bwith valuations. If you rearrange the sample for increasing values \u200b\u200bof possession and you chart a course of two estimates we see that:

against a trend of the errors of estimation with Hatstats quite small and around 0%, l ' performance error of estimate with ratings (in red) is clearly higher and lower values \u200b\u200bfor sovrastimante midfield sottostimante Team 1 and for high values \u200b\u200bof the same. As seen
I feel I can say that the game engine based on the reasoning Hatstats.

PS. take a look at ' CONTENTS of the blog, there are several items that may be of interest.




Andreace (team in Hattrick ID 1730726)

This work is licensed by Andreace under a Creative Commons Attribution-Noncommercial 3.0 Unported License . Cioé questo lavoro può essere liberamente copiato, distribuito o modificato senza espressa autorizzazione dell'autore, a patto che l'autore sia chiaramente indicato e la pubblicazione non sia a fini commerciali.