PDA

View Full Version : A new way to improve the analysis of the data of each heore.



TheLastPandaa
06-11-2017, 06:12 PM
When the devs explained their system to gather the data I was amazed by the amount of inaccuracies they will have to check. The system they use does not hold to gather any information worthy of consideration. The reason is the learning curve each hero has as well as the muscle memory that players must develope with each one in order to be competitives. It is not a good idea to put players of rep4 in the same bag as players of rep28 because it is clear that they do not play the same way and any data that can be collected in statistical terms would be flawed from the beginning since the number of Rep1,2,3, 4... players is probably much greater than rep27,28,29,30 players.


Ubisoft would have to gather the information in at least three different sections to make any data worth of consideration:
* rep 01-9 (new players with that heroe)
* rep 10-19 (experienced players with that heroe)
* rep 20-30 (masters in terms of mechanics and muscle memory with that hero)

Having different sections of analysis will allow the devs to gather data and evaluate it more precisely, so that the corrective measures they take will be much more appropriate to the reality of the game.

With this method, in each section devs should collect all possible information such as the movements most used, the percentage of success that these movements have, etc. The same information should also be taken into account, but in the opposite direction. That is, what movements are the most used against them, the percentage of success that these movements have, etc.

The information gathered in each section would have to be analyzed independently of the rest of the sections and depending on the result of the analysis, compare the data with the other sections.

If, for example, the devs reach to a conclusionthat within section 2 there is a movement with a disproportionate success rate compared to the rest of movements, that information should be compared with the other two sections and see if that index is maintained in those sections or not.
Is that movement a problem for the players who are in the first and third section? Depending of the answer to this question the mesures taken would be diferent, but what i truely this is that with all this system any balance decision between heroes will be much more efficient than it currently is.

Gastgrinder
06-11-2017, 06:20 PM
Useless
I play with my rep 5 Zerker much better as with my rep 30 warden.

Netcode_err_404
06-11-2017, 06:26 PM
Useless
I play with my rep 5 Zerker much better as with my rep 30 warden.

You don't miserum enough then

bmason1000
06-11-2017, 06:37 PM
Useless
I play with my rep 5 Zerker much better as with my rep 30 warden.
Which is just as useful, honestly. His suggestion is about studying repetion and success rates, your routine and success rate as a rep 5 is still indicative of routine and success rates amongst rep 5s. That level is reoresentive not of your skill with that character but of your experience with that character which is what he's suggesting be observed. Commonalities amongst experience levels, not skill levels. It sounds like he's heavily implying experience is equal to skill which i would believe is the case for the majority. Obviously there are anomolies and other factors involved but in a general sense that is probably true.

Gastgrinder
06-11-2017, 06:43 PM
Much does not change in the high rep area.
Someone with much experience plays in the low rep area quite differently as a noob, which plays for the first time for honor
It makes no sense to split it into rep sections

AzureSky.
06-11-2017, 06:49 PM
When the devs explained their system to gather the data I was amazed by the amount of inaccuracies they will have to check. The system they use does not hold to gather any information worthy of consideration. The reason is the learning curve each hero has as well as the muscle memory that players must develope with each one in order to be competitives. It is not a good idea to put players of rep4 in the same bag as players of rep28 because it is clear that they do not play the same way and any data that can be collected in statistical terms would be flawed from the beginning since the number of Rep1,2,3, 4... players is probably much greater than rep27,28,29,30 players.


Ubisoft would have to gather the information in at least three different sections to make any data worth of consideration:
* rep 01-9 (new players with that heroe)
* rep 10-19 (experienced players with that heroe)
* rep 20-30 (masters in terms of mechanics and muscle memory with that hero)

Having different sections of analysis will allow the devs to gather data and evaluate it more precisely, so that the corrective measures they take will be much more appropriate to the reality of the game.

With this method, in each section devs should collect all possible information such as the movements most used, the percentage of success that these movements have, etc. The same information should also be taken into account, but in the opposite direction. That is, what movements are the most used against them, the percentage of success that these movements have, etc.

The information gathered in each section would have to be analyzed independently of the rest of the sections and depending on the result of the analysis, compare the data with the other sections.

If, for example, the devs reach to a conclusionthat within section 2 there is a movement with a disproportionate success rate compared to the rest of movements, that information should be compared with the other two sections and see if that index is maintained in those sections or not.
Is that movement a problem for the players who are in the first and third section? Depending of the answer to this question the mesures taken would be diferent, but what i truely this is that with all this system any balance decision between heroes will be much more efficient than it currently is.

Rep just say play time with that hero, sure you can say that more experience with the hero = more skill but its not like that

The best analysis as a high skill ceiling fighting game should be making analysis based on the data of top level players, you cant nerf or buff based on low level or normal play because you get things like LOL were all the chars are really weak.

Strength and weakness that should be the balance, heros should be allowed to have good moves that they can mix up, kind of like raider (IMO best balanced character in the game)

bmason1000
06-11-2017, 06:54 PM
Useless
I play with my rep 5 Zerker much better as with my rep 30 warden.

Also, I'm not arguing for or against the efficiency of his analysis.

TheLastPandaa
06-11-2017, 07:38 PM
At no time have I said that rep means skill, but if we include in the system an additional model of analysis, it can only benefit everyone. It is not a question of excluding the existing, but adding something new to the system to improve it. Im talking about including one more factor of correction to take into account so devs can reach more validad conclusions according to the obtained data.

Do you remembers the statements that the Devs made months ago that the problem with the PK were their side dodge atacks? Such conclusions would be avoided, or at least mitigated by including the corrective factor I propose, or asimilar one. Im just giving an idea here.


More detailed measurements mean more precise conclusions leading to more appropriate measures to balance the existing heroes.