Stats

[detectiv]

I'm relatively new to Smogon and competitive battling. There is one thing strange I noticed about the stats. They are all based on usage and not on results. What I mean is that just because a pokemon is used doesn't mean it is very good. What stats are needed are those that show the success of pokemon at a competitive level. I suggest two such stats:

KO Rate: Basically the number of KO's a pokemon makes for every time it is ko'ed
Damage Rate: The % damage a pokemon deals every time it is ko'ed

Residual damage could be attributed to the pokemon who set up the entry hazard/weather condition/status condition.
Recoil/LO damage should be excluded from the stats.
Please consider my suggestion.

[Fatecrashers]

Why is this necessary when usage is directly correlated to success and how 'good' a Pokemon is? At the competitive level, people don't just carry a dead weight on their team, so if a Pokemon is getting used on almost every team, you'd better believe that Pokemon is a success!

Furthermore, your suggestion of a KO rate and Damage rate as a form of measurement is overly simplistic. Consider a Baton Pass team, how would the kills be attributed? All to Stored Power Espeon? Would Ninjask then be considered a failure, even though it is very adequate at its sole role of Speed-passing? What about a full stall team? Would all the kills be credited to the Tentacruel that set up Toxic Spikes? What about the Wish-passing Blissey that halts special threats and supports the whole team, but struggles to do any damage in return? How would her success be measured?

Then, even if you do come up with some formula to encompass all possible battling scenarios and team formats, how would it all be implemented? How could you possibly extract your desired results from the data? It is like measuring a person's hours spent at work versus measuring a person's time spent daydreaming at work; though they both measure productivity, one method is simple while the other is absurdly hard. Measuring success via usage may be considered simple, but it's true and it works. Any other formula would likely be too convoluted to be feasible.

[detectiv]

I understand all your concerns to my suggestion.
Firstly usage stats are good-I'm not denying that. However it is true that usage stats can be influenced by fads and previously usable pokemon that are no longer competitively viable. Also few 'banned' pokemon have been #1 in their respective tiers. While I do not propose that stats from results will be 100% accurate, they could be interesting when used in conjunction with usage.

It is true that 'non-hazardous' support pokemon will not be credited in these kind of stats. However, other kinds of pokemon could still be compared. Especially two pokemon that fulfill similar roles in a team can be compared.

To your last concern about implementation I cannot comment. Though I'm okay with computers I do not know exactly how the pokemon server works. Since it is possible to save battle logs I assumed an application could be made to read from this data.

[Fatecrashers]

For tiering purposes, usage stats is the simplest and easiest to understand method. They are updated monthly and are supposed to be reflective of metagame changes, so if a Pokemon has been banned or fallen out of favour, the effects of this will all show up in the next stats update.

Any attempts at determing what Pokemon is the best for what role, as well as what similar roles exist, especially using something other than usage stats, is going to be very subjective and arbitrary. I just don't see this working as a programmable application.

As the feasibility and potential benefits of this suggestion are very suspect, I'm going to lock this thread. Thanks for your suggestion however, and don't hesitate to bring us any other good ideas you may have in the future.