INTRODUCTION
a drop in skill I mentioned last May in the first article I had written on the planning of training http://acandio.blogspot.com/2010/05/hattrick-tool-di-pianificazione.html . Here I will try to address the issue in a more organic, acnhe taking into account the recent significant changes have occurred on this topic.
The following discussion is based on a sample is not great, I filtered hundreds of players, thousands and filtering them clearly in a broader time frame, the results may improve, but currently do not have time to conduct a search so expensive. The conclusions that follow will therefore be indicative and not conclusive, but from the perspective of "better an egg today than a hen tomorrow," I decided to put down as arrangements which currently, which should be enough, however, to outline a general framework.
The following discussion is based on a sample is not great, I filtered hundreds of players, thousands and filtering them clearly in a broader time frame, the results may improve, but currently do not have time to conduct a search so expensive. The conclusions that follow will therefore be indicative and not conclusive, but from the perspective of "better an egg today than a hen tomorrow," I decided to put down as arrangements which currently, which should be enough, however, to outline a general framework.
THE SKILL OF CALI
Skill Cali I can be of two types:
- for HIGH SKILL
- for ETA '
The change announced at 'internal editorial of June 21, 2010 "Training faster next season and become operational at the beginning of the season 43 (7 August 2010) marked a watershed in the drop in skill:
" Age will become less important for the drop in skill - therefore, cali di skill dovuti all'età rallenteranno. Comunque, i cali di skill legati alle skill per abilità sopra Titanico saranno più veloci (più alta la skill e più veloci i cali)"
Quindi, schematizzando, abbiamo questa situazione:
" Age will become less important for the drop in skill - therefore, cali di skill dovuti all'età rallenteranno. Comunque, i cali di skill legati alle skill per abilità sopra Titanico saranno più veloci (più alta la skill e più veloci i cali)"
Quindi, schematizzando, abbiamo questa situazione:
il 7 agosto sono stati modificati i cali e i cali da alta skill sono passati da essere leggeri (in giallino, fattore "1") a pesanti (in color pesca, fattore "2") viceversa i cali da età sono passati da pesanti (fattore "3") a leggeri (fattore "4").
Il punto è che nei giocatori over 30 molto forti gli effetti si sommano e i cali sono dovuti sia all'età che all'alto skill level prior to August 7 by the sum of factors 1 and 3 in the table above, from 2:04 hours. Therefore necessary to analyze separately the different factors to estimate effectively decreases. We try.
from Cali HIGH SKILL
In this article published on the site DAC http://croack.altervista.org/it/cali-divini-ricercheconcluse-115 Danfisico was brilliantly successful in obtaining a formula that could give an account of the values \u200b\u200bof the losses for goalkeepers:
Lawful at this point ask yourself two questions:
[(Skill - 17.5) ^ 2] / 1000
Lawful at this point ask yourself two questions:
- formula works for other skills?
- as was modified after August 7?
to answer these questions we need to look closely at the sample that we collected: monoskill in defense, attack and directed Hattrick and listed in the following analysis of their salary history at Alltid.
To avoid the overlapping of the high skill factor and the age I have seen players in the 24/28 year-end. Younger players do not have very high levels of technical skill for reasons of length of training and older players are starting to suffer the heat factor of age.
E 'need to highlight the precise date of their birthday, because only their birthday manifested through their salary, their level of skill. Knowing when it is their birthday, we can calculate how many weeks have lapsed since the anniversary date of August 7 and then know how many have suffered declines of skill in the new regime of greatest falls from high skill.
This is the model where the data will be shown
To avoid the overlapping of the high skill factor and the age I have seen players in the 24/28 year-end. Younger players do not have very high levels of technical skill for reasons of length of training and older players are starting to suffer the heat factor of age.
E 'need to highlight the precise date of their birthday, because only their birthday manifested through their salary, their level of skill. Knowing when it is their birthday, we can calculate how many weeks have lapsed since the anniversary date of August 7 and then know how many have suffered declines of skill in the new regime of greatest falls from high skill.
This is the model where the data will be shown
begin his defenders are just two:
Danfisico always, to answer the second question "how is amended after August 7? "had suggested that the old formula [(Skill - 17.5) ^ 2] / 1000 become
[(Skill - 14,85)^2]/1000
Precisando che " per ora il valore migliore da usare sembrerebbe 14,85 in attesa di una futura analisi con più dati ".
Proviamo a vedere:
Nel primo caso con 17.086, secondo la formula il giocatore subirebbe 16 cali di entità variabile (maggiore il primo e via via inferiori gli altri) per un totale di
un totale di 0.077 non lontano dallo 0.079 che osserviamo. Nel secondo caso con la formula di Danfisico avremmo un calo totale di 0.141 di pochissimo inferiore allo 0.146 stimato sui valori di skill al compleanno.
Riassumendo in the table:
continue with the directors, which are most abundant, 14 players.
see listed in the blue drops that occur in full (16 weeks) in the new regime of heavy declines in yellow when the drop is mixed (some 'new and a bit of a fall' to drop old) and peach when all the observed decrease is due to loss of the old type:
see listed in the blue drops that occur in full (16 weeks) in the new regime of heavy declines in yellow when the drop is mixed (some 'new and a bit of a fall' to drop old) and peach when all the observed decrease is due to loss of the old type:
We also see the 15 attackers, before moving on to a table of analysis:
Now for the table analysis rearranges the data collected so easier for a comparison of assumptions and observed data:
Seeing the values \u200b\u200bof the differences we can answer the first question we set ourselves above, that "the formula works for the other skills? "and the answer is yes: there are major differences between a skill and the other, so I feel I can say that, as noted above, ALL SKILL LOSSES SUFFERED BY THE SAME HIGH SKILL.
Now, let's focus on new dips and eliminated from the values \u200b\u200bfor only the old regime falls. It also eliminates the two outliers with ID 176499548 e 188016298 che probabilmente hanno avuto una piccola quantità di allenamento. Fatto questo procedo a riordinare livelli di skill e differenze tra valori ipotizzati e osservati nella realtà in un grafico a dispersione per vedere se vi è una correlazione tra i valori.
Now, let's focus on new dips and eliminated from the values \u200b\u200bfor only the old regime falls. It also eliminates the two outliers with ID 176499548 e 188016298 che probabilmente hanno avuto una piccola quantità di allenamento. Fatto questo procedo a riordinare livelli di skill e differenze tra valori ipotizzati e osservati nella realtà in un grafico a dispersione per vedere se vi è una correlazione tra i valori.
Una leggera correlazione c'è in effetti: si può osservare come all'aumentare della skill la formula di Danfisico tenda leggermente a sovrastimare i cali effettivi subiti dal giocatore. Forse la formula sarebbe migliorabile rendendola un po' meno esponenziale di quanto sia tutt'ora e aumentando leggermente i valori base, ma ne vale la pena?
Guardiamo i dati: su 39 casi solo 2 volte (5% del campione) l'errore supera 0.010 cioè 1 centesimo di skill a stagione, con un errore medio di 0,0014 , cioè 14 decimillesimi di skill a stagione . Direi c'è da essere più che soddisfatti, ringraziare e fare davvero i complimenti a Danfisico per essere riuscito a trovare una formula così efficace.
RIASSUMENDO , per quanto visto sopra
RIASSUMENDO , per quanto visto sopra
- TUTTE LE SKILL sembrano soffire di IDENTICI CALI DA ALTA SKILL
- Il CALO settimanale da ALTA SKILL sembra aggirarsi attorno a
[(Skill - 14,85)^2]/1000
Infine posto una tabella di riferimento veloce, se un vostro giocatore ha un certo livello di skill, quale perdita ci si può aspettare in una stagione hattrickiana (16 settimane)?
PS. dai un occhio all' INDICE del blog, ci sono parecchi articoli che ti potrebbero interessare.
Andreac (team ID 1730726 in Hattrick)
Andreac (team ID 1730726 in Hattrick)
This opera by Andreac is licensed under a Creative Commons Attribuzione-Non commerciale 3.0 Unported License . Cioé questo lavoro can be freely copied, distributed or modified without the express permission of the author, provided that the author is clearly stated and the publication is not for commercial purposes.
0 comments:
Post a Comment