GamePress

Automatic Moveset-Evaluation

The move-Tables are now public for two days. There is a formula for cycle damage, for attacking and for defending. So evaluation for the known Game-Press-System could be done quite fast, automatic, maybe with this approach:

Highest Damage moveset: A
Each 4% less damage: one note worse.
Same Type of Quick Move / Charge move on Attacker: +0,5 note
Different Type of Quick Move / Charged move on Defender: +0,5 Note
Flexibility bonus for both, short timed Quick and Charged moves on Attacker.

This can be done very easy with a sheet. Manual correction can be done later, but with this proposal, many questions about quality of movesets could be obsolete.

Asked by Bruno Brezel8 years 2 months ago

Report

Answers

It is a good idea to automate this to speed up things. Frankly, I'm glad Gamepress take their time to provide accurate info and doesn't choose the Niantic way of releasing stuff (publish something broken and then fix it and hope someone didn't notice they got it wrong the first time).

Additionally, none of those spreadsheets factor in type advantages and metagame, which is what separates Gamepress' ratings from the mere damage/battle calculations that most other sources provide.

If people are eager about ranking their movesets, they can get hold of a spreadsheet and do their own analysis, or ask other people on here who have. This is a Q&A section, and I'd rather see some activity and discussion on here than reducing the number of discussions because they would be "obsolete" :)

Amen

and as mentioned by Arac2 recently, some charge moves are hard to dodge (as oppose to prior Gen 2) and Gamepress will take that into consideration maybe.

amen, brotha

This. Purely quantitative lists could be automated and updated without much issue. However, the real value in Gamepress is the qualitative analysis. For example, think of the top tier attacker/defender lists - just going by DPS (and ignoring the metagame) you'd expect Arcanine and Charizard to be ranked much better attackers, and Rhydon and Golem to be ranked much better defenders. In each case, their utility is to some extent neutralized by the prevalence of good type counters (read: Vaporeon).

If you're so desperate for the "best" rankings right meow, then why not go and calculate them yourself? You can even pull the DPS and EPS data from the Gamepress site. I, for one, don't have a problem with waiting for a few days/weeks for the more thorough qualitative analysis to come out though.

Honestly the only ranking I check is the best type attacker ranking, I consider defender and attacker rankings mostly useless for anything other than "generic" attackers, for the most part when you attack a gym you choose the ones that have the most advantage against the defenders, therefor, you choose the best attackers of a certain type that counters the defenders present in the gym. For defense, until the gym placement system is changed, the only valid ranking is and will be Max CP minus legendaries.

Most people aren't this logical. They think their 3500 Dragonite with SW/DP is going to scare me away. "Bish I drove 10 km for this gym, I'm taking this mf gym down!"

cycle DPS has a formula and is relatively straightforward to calculate. It does, however, have certain limitations:
1. It only makes sense for the Spamming Attack Style; that is, never dodge and launch charge attack as soon as possible. The "cycle DPS with dodge" is hard to calculate because it depends on the specific matchup.
2. It assumes neutral damage. Gengar's Shadow Claw + Shadow Ball is No.1 DPS, but Dragonite's DT+O on another Dragonite is obviously superior, due to the x1.25 factor. Yet it is wrong to adjust only by *1.25 for mix type move set, such as ZH+SoB. The multiple is somewhere between 1~1.25.
3. A cycle is often incomplete; there must be some energy left (or wasted due to overcharging), especially for 1-bar charge moves.
4. It didn't account for the energy gained due to HP loss, which is also case-dependent.
5. DPS is not everything! Sometimes we need KD (Kill Death ratio, how many enemies can this mon take down before fainted). A move set with lower CD will increase dodge performance and yield higher KD, which isn't reflected in the cycle DPS as well.
Therefore, cycle DPS analysis should only be a "glance" at the move set, or "expected performance". If the difference is within 5%, then other factors may play even more important roles.

A next-level analysis is done by battle simulation. There are a few websites available for conducting a specific matchup battle simulation. Putting the accuracy aside, since such simulation is too case-specific, one may need to conduct a lot of sims before drawing any statistically reliable conclusion.

Therefore, the only way left to evaluate a move set effectively is to try it out in the game. Practices beat theories! That's why I'm most interested in reading posts where trainers share their experiences using different move set.

It only makes sense for the Spamming Attack Style; that is, never dodge and launch charge attack as soon as possible. The "circle DPS with dodge" is hard to calculate because it depends on the specific matchup.

How was this taken into account by gamepress? My proposed formula takes it into account with rating faster movesets higher.

It assumes neutral damage. Gengar's Shadow Claw + Shadow Ball is No.1 DPS, but Dragonite's DT+O on another Dragonite is obviously superior, due to the x1.25 factor. Yet it is wrong to adjust only by *1.25 for mix type move set, such as ZH+SoB. The multiple is somewhere between 1~1.25.

STAB is a property of the attacker, not of the moveset, so it is already calculated into circle damage.
I agree, likely matchups can be taken into account evaluating a pokemon, but not a moveset. Giving a bonus to a same type quick- and charged attack, i already take into account, that attacker is carefully chosen for each machup.

A circle is often incomplete; there must be some energy left (or wasted due to overcharging), especially for 1-bar charge moves.

Yes, that is why i give a bonus to faster movesets.

It didn't account for the energy gained due to HP loss, which is also case-dependent.

This is not neccesairy. It is a property of a pokemon (its hp), so it does not prefer a specific moveset on the same pokemon.

DPS is not everything! Sometimes we need KD (Kill Death ratio, how many enemies can this mon take down before fainted).

Defense on same pokemon is same. KD-Ratio is directly dependend on kill speed, thus on DPS.

A move set with lower CD will increase dodge performance and yield higher KD, which isn't reflected in the circle DPS as well.

I give a bonus for faster moves.

Therefore, circle DPS analysis should only be a "glance" at the move set, or "expected performance". If the difference is within 5%, then other factors may play even more important roles.

All facts you mention above are already taken into account by my formula, if relevant.

A next-level analysis is done by battle simulation. There are a few websites available for conducting a specific matchup battle simulation. Putting the accuracy aside, since such simulation is too case-specific, one may need to conduct a lot of sims before drawing any statistically reliable conclusion.

Maybe, this could be done with the most likely opponents. Maybe this would refine my table.

Therefore, the only way left to evaluate a move set effectively is to try it out in the game. Practices beat theories!

Higly subjective anecdotical knowledge, only reliable with big amount of data. Yes, i can tell you how bad my moveset is, i fainted with a cloyster against a dragonite!
What i maybe do not tell is, that i failed to dodge dc twice because of cold fingers, or how big dragonite and cloyster were exactly. Next try, maybe my cloyster wins - did i already mention, that before change to cloyster i applied a solar beam with my exeggutor who preceded it in my attackers lineup and had few hps and much energy left?.....