Part 1: http://www.smogon.com/forums/showthread.php?t=51778 (outdated, but still good)
Part 2: http://www.smogon.com/forums/showthread.php?t=65122
Part 3: http://www.smogon.com/forums/showthread.php?t=81313
Part 4: http://www.smogon.com/forums/showthread.php?t=84227
Part 5: http://www.smogon.com/forums/showthread.php?t=84449
Part 6: http://www.smogon.com/forums/showthread.php?t=3447576 best one
Now, onto team building itself. One thing you have to drop is the idea of "new" team. If you're not using a certain move, or a certain pokemon because it would be "copying" someone else's idea and not it being an original creation, that's idiotic. Between the fundamental sets and strats (restbell, curselax, sleep talk) developed by the 00-02 foundational players, and vil/celia/bob/whatever developing most of the defensive conceptual shit, and my developing everything else, there's very little, if any "new" left. The concept of "new" teams revolves around a "new" idea, of which there really aren't any. Changing surf to hydro pump, or cloyster to forretress, doesn't really destroy the core foundational function of the team, so for all intents and purposes, it's still the same team. That's why if I try to "rate" teams, I really just end up giving them a worser version of a team I already made.
As a followup, this means: be flexible. A "team" doesn't consist of 6 pokemon with 4 moves each, it consists of many variations that you can turn to at any time, but still following the same general concept. It should play similarily, but adds unpredictability and flair to the team, something that you could turn to immediately. It can also help with matchups against certain players, and give you a way of adapting to the changing metagame.
Example of the most recent/last real team I'll make (multiple moves are semi-listed in the order of importance):
EDIT: TEAM REMOVED
FUCK YOU GUYS
jk
but not really
Synergies/comments: Also removed.
Team building, and shit. Analysis. In-depth. Try it on your team, and you'll know why it falls short.
Switch patterns: The concept is pretty simple. The best way to get a pokemon in is on an opponent's switch, switching in on a switch if you will. Generally a bulky pokemon starts the chain and probably takes a hit in the process, while the more threatening pokemon ends it. This is what you utilize to bring in the Machamps, the Marowaks, the Eggy, etc. Since it all depends on the opponent's team, there's really no way to list everything. Your team has to be flexible enough such that against almost every team, you should be able to have a 2-3 pokemon relay to get the one you want active. Zapdos -> Machamp is pretty straight forward if you're suspecting Snorlax. Zapdos -> Steelix -> Egg/Machamp if you're expecting Raikou. Vaporeon -> Snorlax -> Zapdos, etc, it all depends on who you're facing and who you want active.
LEARN THE SWITCHBACK! If it's clear one of your pokemon has your opponent on a leash, abuse it. Make it active as often as you can. Spikes Raikou roared in Rhydon? Switch to Suicune, go right back to Raikou. Yes, switch Raikou INTO Rhydon. Don't waste time. Of course, it's always a great idea to keep your opponent in check by throwing out a surf every once in a while. Keep him playing by the books, don't let him get cocky. Probably the most important play and the most beneficial play in the game.
Some quick tips: You should know half this shit "accidentally", but what seperates the good from the great is that I know this shit consciously. Whereas you're feeling around in a match, I have hardcoded shit to turn to. I know exactly what's going on, and I have something pre-made and reliable to turn to when I need it. So while you know the concept of multiple switches to get somewhere, it's about pinpointing those switches and hardcoding them into your arsenal so you can recall them any time, instead of having to relearn it every battle.
Just because a Pokemon sleep talks, doesn't mean it wants to be put to sleep. On the contrary, it doesn't mind sleep as long as it's in control of its own sleep pattern (aka through Resting). Being forced sleep, then relying on Sleep Talk is a huge gamble and should be avoided if at all possible. What I like to do is switch Zapdos in on a damage attack as soon as possible (obviously not to something stupid, like Snorlax or Marowak), then immediately Rest to fall into a sleep pattern. That way, I can status absorb with it while I scout if my opponent has a sleep move. Of course, if it's too late to do that, then obviously letting the opponent sleep a sleep talker is always better than Egg or Steelix or something.
Look ahead. You should never be playing "for the moment", then you're just feeling around. Stop feeling around. You should at the very least, have the next 5-10 turns played out in detail in your head, combined with a rough idea of what you're working towards for the next 50-100 turns. Stop feeling around and waiting to see what your opponent will do. You should know what you're doing, that way you maintain the advantage in a match. Seize tempo, seize control. Make him your bitch.
You should always know what you're doing, what you're working towards. You should never be blindly attacking, or just hitting to get a CH. There are almost always better alternatives (relying on 7 straight thunder hits is better than relying on a CH). Have a strategy. A pokemon's worth on a team changes depending on the opponent, in some games, it makes sense to sacrifice Snorlax, in other games, it doesn't. It depends. How quickly you find out depends on how quickly you can form that strategy.
Stop relying on Snorlax. If it dies, it dies.
Sometimes the best time to use Explosion is when your opponent has low HP. GSC players go to great lengths to save their pokemon, take advantage of that. Explosion is best used when your opponent feels safest. Use it when facing a steel/ghost pokemon. Unpredictably, not stupidly though. It's risky, but that's some of the best moments to catch the opponent. The most frequently played "good" explosions are generally on the switch explosions. Risky, but definitely worth it more times than not. Explosion is risky by definition anyway.
Playing an offensive team is about applying pressure. That means, doing something as often as you can. That means, minimizing the number of switches you're making. Minimizing the number of turns you're "fast asleep". Pokemon that don't need to switch out as often are great (which is why I love Zapdos). If you're dishing a hit, switching, dishing a hit, switching, you're doing it wrong. Stop wasting time, get to the point. Stop losing to leftovers.
Work impulsive intuition into your game. It's the best random number generator out there. And yes, it gets better over time.
What qualifies as "something" is an attack that matters. Hitting Raikou with Zapdos and not capitalizing, or on a team that CAN'T capitalize it is stupid. It's doing nothing. It's using PP to do nothing. On the other hand, that same attack with the purpose to put it in 2shot range for Vap is doing something. The key here is purpose. Why are you doing what you're doing?
Play the numbers. Disregard females, acquire currency. Do what makes sense statistically. Being on the wrong end of something in your favor is nothing to be ashamed of. Multiple options? Pick the best one statistically. Pause and do the math if you have to. Then blame it on the door/porn pop up when your opponent complains.
Reality is an illusion that occurs due to the lack of alcohol.
On prediction and Thunder:
Which then brings me to the topic of "necessary" prediction. All/most of predictions should lie in this subgroup. A prediction is a gamble by definition. If there's no reward, there should be no reason for you to do it, that is, unless, there's no disadvantage either, then you're just clicking. A prediction should always accomplish something, be it a kill, a status, or just merely improving your position, it should never get you nowhere. In GSC, far more attacks amount to nothing, because almost everything can be recovered from. So really, when you "predict" with your Ttar and tbolt my resttalking Suicune, unless I was in position to die, I'll send it in over and over. You can feel like you're "outplaying me by predicting it", but really, you're not. Of course, in the instance where I decide to go to Lax instead, you'd be shit outta luck. It's rare that I predict in a game I can just win by just playing it through naturally (unless I'm trying to prove something), there's no reason to. Prediction merely shortens the game, at the cost of "guaranteed victory".
Prediction is merely the journey, and there is no journey without a destination. rediction is always working towards something, so if you're winning mostly those unnecessary/inconsequential predictions, you're not winning anything. I'll let you have those more times than not, because you'll gain very little, whereas I only have to catch you once to put a pounding in. And quite frankly, that's not really worth it. Don't do it.
And the way to look at Hydro Pump is: anything is better than 0%. That's to say, if Hydro Pump gives you a chance at killing something you normally wouldn't, then by definition it's a more than viable move. Let's say you had a move, let's call it Ultra Pump, that's a 30% accurate 300 base power water-move. "Statistically", it's weaker than both Surf and Hydro Pump, but in practice, it's a 30% chance for Vaporeon to OHKO Zapdos versus a 0% chance for Surf/Hydro Pump to get that same kill. It's a 24% chance for Vaporeon to 2-shot Snorlax (with Hydro Pump turn 2) versus 0% for Hydro Pump/Surf. The only reason I ever use Surf over Hydro Pump is for PP reasons, else, Hydro Pump is plain better. It nets otherwise impossible kills, and you can't put a numerical value on that.
Another way to look at Hydro Pump is an 80% accurate "necessary prediction". Those are insanely good odds. With those consistent 100% moves, you're relying on prediction (aka guessing), which is another form of inconsistency. You can say that it's something in "your control", but it's not, it's 50% in your opponent's. So really, unless that chain of successful predictions is greater than 80%/70% in odds (and there's no way you can say that definitively), Hydro Pump/Thunder is FAR better than Surf/Thunderbolt relying on prediction in those instances.
Having said all of that, I never use Hydro Pump. The two water pokemon that matter and learn it (Starmie and Vaporeon) both outlive what the 8 pp move has to offer. Actually, Vaporeon can get away with it in some games (e.g. vap vs vap matchups), but Starmie just has no business using the move. It doesn't kill anything ever.
Thunder, though, is a whole 'nother story. My rule of thumb is: Zapdos gets Thunder/HP Ice and Raikou gets Tbolt/HP Water. Zapdos is far tankier than you can imagine, and Ice hits the very important Exeggutor [along with ground types etc, but that's beside the point]. Zapdos can also go toe to toe with shit like Steelix, not necessarily threaten it, but I've found it's better to hit Egg for SE and Steelix for neutral than Steelix for SE and Egg for NVE. Thunder is better than tbolt in this case because of para [and damage rolls, sorta]. 60% to lax is frightening, and enough to pull off 4HKO combos, OHKO on starmie is irrelevent, para on lax/raikou/others is golden. Raikou is pretty much a stall pokemon, and second best special wall in the game. I'm a firm believer that Raikou shines best as a phazer with spike support, and because of that, HP water just makes more sense. Whereas Zapdos' aim is to kill Pokemon, Raikou's aim would be to scare pokemon and own them with roar predictions. Because of that, it's better to threaten something like Steelix/Rhydon, who can end your fun on the spot, than to hit something like Egg for X amount of damage, who doesn't really force you to do anything at all.
Part 2: http://www.smogon.com/forums/showthread.php?t=65122
Part 3: http://www.smogon.com/forums/showthread.php?t=81313
Part 4: http://www.smogon.com/forums/showthread.php?t=84227
Part 5: http://www.smogon.com/forums/showthread.php?t=84449
Part 6: http://www.smogon.com/forums/showthread.php?t=3447576 best one
Now, onto team building itself. One thing you have to drop is the idea of "new" team. If you're not using a certain move, or a certain pokemon because it would be "copying" someone else's idea and not it being an original creation, that's idiotic. Between the fundamental sets and strats (restbell, curselax, sleep talk) developed by the 00-02 foundational players, and vil/celia/bob/whatever developing most of the defensive conceptual shit, and my developing everything else, there's very little, if any "new" left. The concept of "new" teams revolves around a "new" idea, of which there really aren't any. Changing surf to hydro pump, or cloyster to forretress, doesn't really destroy the core foundational function of the team, so for all intents and purposes, it's still the same team. That's why if I try to "rate" teams, I really just end up giving them a worser version of a team I already made.
As a followup, this means: be flexible. A "team" doesn't consist of 6 pokemon with 4 moves each, it consists of many variations that you can turn to at any time, but still following the same general concept. It should play similarily, but adds unpredictability and flair to the team, something that you could turn to immediately. It can also help with matchups against certain players, and give you a way of adapting to the changing metagame.
Example of the most recent/last real team I'll make (multiple moves are semi-listed in the order of importance):
EDIT: TEAM REMOVED
FUCK YOU GUYS
jk
but not really
Synergies/comments: Also removed.
Team building, and shit. Analysis. In-depth. Try it on your team, and you'll know why it falls short.
Switch patterns: The concept is pretty simple. The best way to get a pokemon in is on an opponent's switch, switching in on a switch if you will. Generally a bulky pokemon starts the chain and probably takes a hit in the process, while the more threatening pokemon ends it. This is what you utilize to bring in the Machamps, the Marowaks, the Eggy, etc. Since it all depends on the opponent's team, there's really no way to list everything. Your team has to be flexible enough such that against almost every team, you should be able to have a 2-3 pokemon relay to get the one you want active. Zapdos -> Machamp is pretty straight forward if you're suspecting Snorlax. Zapdos -> Steelix -> Egg/Machamp if you're expecting Raikou. Vaporeon -> Snorlax -> Zapdos, etc, it all depends on who you're facing and who you want active.
LEARN THE SWITCHBACK! If it's clear one of your pokemon has your opponent on a leash, abuse it. Make it active as often as you can. Spikes Raikou roared in Rhydon? Switch to Suicune, go right back to Raikou. Yes, switch Raikou INTO Rhydon. Don't waste time. Of course, it's always a great idea to keep your opponent in check by throwing out a surf every once in a while. Keep him playing by the books, don't let him get cocky. Probably the most important play and the most beneficial play in the game.
Some quick tips: You should know half this shit "accidentally", but what seperates the good from the great is that I know this shit consciously. Whereas you're feeling around in a match, I have hardcoded shit to turn to. I know exactly what's going on, and I have something pre-made and reliable to turn to when I need it. So while you know the concept of multiple switches to get somewhere, it's about pinpointing those switches and hardcoding them into your arsenal so you can recall them any time, instead of having to relearn it every battle.
Just because a Pokemon sleep talks, doesn't mean it wants to be put to sleep. On the contrary, it doesn't mind sleep as long as it's in control of its own sleep pattern (aka through Resting). Being forced sleep, then relying on Sleep Talk is a huge gamble and should be avoided if at all possible. What I like to do is switch Zapdos in on a damage attack as soon as possible (obviously not to something stupid, like Snorlax or Marowak), then immediately Rest to fall into a sleep pattern. That way, I can status absorb with it while I scout if my opponent has a sleep move. Of course, if it's too late to do that, then obviously letting the opponent sleep a sleep talker is always better than Egg or Steelix or something.
Look ahead. You should never be playing "for the moment", then you're just feeling around. Stop feeling around. You should at the very least, have the next 5-10 turns played out in detail in your head, combined with a rough idea of what you're working towards for the next 50-100 turns. Stop feeling around and waiting to see what your opponent will do. You should know what you're doing, that way you maintain the advantage in a match. Seize tempo, seize control. Make him your bitch.
You should always know what you're doing, what you're working towards. You should never be blindly attacking, or just hitting to get a CH. There are almost always better alternatives (relying on 7 straight thunder hits is better than relying on a CH). Have a strategy. A pokemon's worth on a team changes depending on the opponent, in some games, it makes sense to sacrifice Snorlax, in other games, it doesn't. It depends. How quickly you find out depends on how quickly you can form that strategy.
Stop relying on Snorlax. If it dies, it dies.
Sometimes the best time to use Explosion is when your opponent has low HP. GSC players go to great lengths to save their pokemon, take advantage of that. Explosion is best used when your opponent feels safest. Use it when facing a steel/ghost pokemon. Unpredictably, not stupidly though. It's risky, but that's some of the best moments to catch the opponent. The most frequently played "good" explosions are generally on the switch explosions. Risky, but definitely worth it more times than not. Explosion is risky by definition anyway.
Playing an offensive team is about applying pressure. That means, doing something as often as you can. That means, minimizing the number of switches you're making. Minimizing the number of turns you're "fast asleep". Pokemon that don't need to switch out as often are great (which is why I love Zapdos). If you're dishing a hit, switching, dishing a hit, switching, you're doing it wrong. Stop wasting time, get to the point. Stop losing to leftovers.
Work impulsive intuition into your game. It's the best random number generator out there. And yes, it gets better over time.
What qualifies as "something" is an attack that matters. Hitting Raikou with Zapdos and not capitalizing, or on a team that CAN'T capitalize it is stupid. It's doing nothing. It's using PP to do nothing. On the other hand, that same attack with the purpose to put it in 2shot range for Vap is doing something. The key here is purpose. Why are you doing what you're doing?
Play the numbers. Disregard females, acquire currency. Do what makes sense statistically. Being on the wrong end of something in your favor is nothing to be ashamed of. Multiple options? Pick the best one statistically. Pause and do the math if you have to. Then blame it on the door/porn pop up when your opponent complains.
Reality is an illusion that occurs due to the lack of alcohol.
On prediction and Thunder:
Which then brings me to the topic of "necessary" prediction. All/most of predictions should lie in this subgroup. A prediction is a gamble by definition. If there's no reward, there should be no reason for you to do it, that is, unless, there's no disadvantage either, then you're just clicking. A prediction should always accomplish something, be it a kill, a status, or just merely improving your position, it should never get you nowhere. In GSC, far more attacks amount to nothing, because almost everything can be recovered from. So really, when you "predict" with your Ttar and tbolt my resttalking Suicune, unless I was in position to die, I'll send it in over and over. You can feel like you're "outplaying me by predicting it", but really, you're not. Of course, in the instance where I decide to go to Lax instead, you'd be shit outta luck. It's rare that I predict in a game I can just win by just playing it through naturally (unless I'm trying to prove something), there's no reason to. Prediction merely shortens the game, at the cost of "guaranteed victory".
Prediction is merely the journey, and there is no journey without a destination. rediction is always working towards something, so if you're winning mostly those unnecessary/inconsequential predictions, you're not winning anything. I'll let you have those more times than not, because you'll gain very little, whereas I only have to catch you once to put a pounding in. And quite frankly, that's not really worth it. Don't do it.
And the way to look at Hydro Pump is: anything is better than 0%. That's to say, if Hydro Pump gives you a chance at killing something you normally wouldn't, then by definition it's a more than viable move. Let's say you had a move, let's call it Ultra Pump, that's a 30% accurate 300 base power water-move. "Statistically", it's weaker than both Surf and Hydro Pump, but in practice, it's a 30% chance for Vaporeon to OHKO Zapdos versus a 0% chance for Surf/Hydro Pump to get that same kill. It's a 24% chance for Vaporeon to 2-shot Snorlax (with Hydro Pump turn 2) versus 0% for Hydro Pump/Surf. The only reason I ever use Surf over Hydro Pump is for PP reasons, else, Hydro Pump is plain better. It nets otherwise impossible kills, and you can't put a numerical value on that.
Another way to look at Hydro Pump is an 80% accurate "necessary prediction". Those are insanely good odds. With those consistent 100% moves, you're relying on prediction (aka guessing), which is another form of inconsistency. You can say that it's something in "your control", but it's not, it's 50% in your opponent's. So really, unless that chain of successful predictions is greater than 80%/70% in odds (and there's no way you can say that definitively), Hydro Pump/Thunder is FAR better than Surf/Thunderbolt relying on prediction in those instances.
Having said all of that, I never use Hydro Pump. The two water pokemon that matter and learn it (Starmie and Vaporeon) both outlive what the 8 pp move has to offer. Actually, Vaporeon can get away with it in some games (e.g. vap vs vap matchups), but Starmie just has no business using the move. It doesn't kill anything ever.
Thunder, though, is a whole 'nother story. My rule of thumb is: Zapdos gets Thunder/HP Ice and Raikou gets Tbolt/HP Water. Zapdos is far tankier than you can imagine, and Ice hits the very important Exeggutor [along with ground types etc, but that's beside the point]. Zapdos can also go toe to toe with shit like Steelix, not necessarily threaten it, but I've found it's better to hit Egg for SE and Steelix for neutral than Steelix for SE and Egg for NVE. Thunder is better than tbolt in this case because of para [and damage rolls, sorta]. 60% to lax is frightening, and enough to pull off 4HKO combos, OHKO on starmie is irrelevent, para on lax/raikou/others is golden. Raikou is pretty much a stall pokemon, and second best special wall in the game. I'm a firm believer that Raikou shines best as a phazer with spike support, and because of that, HP water just makes more sense. Whereas Zapdos' aim is to kill Pokemon, Raikou's aim would be to scare pokemon and own them with roar predictions. Because of that, it's better to threaten something like Steelix/Rhydon, who can end your fun on the spot, than to hit something like Egg for X amount of damage, who doesn't really force you to do anything at all.