DM Guidelines
Starting Resources
25 points and 4200 copper should make for a reasonably good value for a campaign that starts with budding heroes. Lower values (especially in points) are unlikely to be very interesting. Higher values are more complicated, and probably not ideal for a first campaign.
Helping the Players Work Together and Against the Enemy
If left to make characters independently, with no input from you or each other, the players may have difficulty coincidentally making characters that have a reason to stick together as a team and accomplish things. By providing a basic premise that all the characters should conform to (like they're all champions of some particular nation or ideal, all come from some particular village that needs protecting, or are all thieves that like working together), you can ensure that they care.
If the characters have a reason to care about their world that they can kind of agree on, this also helps them initiate new adventures when they're not sure what to do next. If you don't have a main quest in mind, giving the characters the motivation to start their own adventures is necessary, but even if you do have a main quest in mind, side adventures are an opportunity for the characters to feel that they are in control of their own destiny, and for you to provide resources and clues that help them directly or indirectly when they decide to return to solving the main quest.
Advancement
You determine how rapidly the point limit of the players increases. To keep the characters who are least dependent on wealth (such as armies and psychics) from overwhelming other characters, players should not gain points while they have less than 100 copper per point.
The upper end of magic equipment is much more fluid; up to 400 copper per point is within the sensible range, but it's actually fine to let people acquire all sorts of valuable devices without any point limit increase.
Point Pool Increase
The simple way to award point pool increases is to do so after battles or sessions, so the players can advance their characters on their own time.
However, for a more Shonen-esque experience, you can award points in the middles of battles. To prevent that from taking a long time, use the same sum each time, and tell the players how big the sum will be so they can prepare.
Encounter Design
In terms of encounter design, the advantage a tactical RPG has over a tactical wargame is that entry into the conflict is not so overly well-defined. An assault on a castle could start with an attempt to scale the walls, or the battle might not start until they're well inside, because they've disguised theirselves as a triumphant returning lord and servants. An ambush in an inn might be detected or avoided entirely.
That's not to say you can't have any fights with clear planned beginnings, but you will be missing out if you don't let the players have some opportunities to come up with ways to arrange battles in their favor.
When the players come up with a clever plan to safely defeat an opponent, it can be very satisfying to watch it succeed or fail catastrophically. But when the plan isn't fantastic, a fair fight is the best kind of fight. Consider calibrating fights such that if they only strategize as well as they usually do, the fight will be challenging yet beatable. If you don't know how well they usually strategize, it's better to guess low for a few encounters; players are usually scared off more easily by losing during their first battle than by winning without enough of a challenge a few times.
Encounter Pricing
In order to maintain some campaign flow, it's better if the players win more than half of the encounters they fight. However, in order to make sure the encounters are fun and challenging, the encounters should not be trivial. Therefore, the point value of an encounter should be somewhere between three-quarters and equal to the combined point values of all players. At the higher end of that, the opposing forces should have significantly less wealth.
You can use higher point values and wealth values, but there should be sufficient advance warning of such a battle that the players can arrange to fight on their own terms or not at all.
You can use lower point values, but the battle won't be very interesting if these opposing forces don't have a significant situational advantage. Forces with no hope of actually harming the player characters should simply be handwaved to defeat, rather than going to the trouble of rolling for initiative, placing out models, and so on: when a 300 point Legend player character is confronted by a handful of 5 point bandits, the player should just say "I kill them" or "I scare them off" or something, and the campaign can continue.
Stealth Encounters
Sometimes the players may wish to infiltrate somewhere and accomplish a task covertly rather than engaging their foes in combat. Encounters like this can work with the players being ostensibly outmatched, but their enemy is spread thin. That is, the total point value of enemies that could converge on their location in 2 rounds or so should not be greater than the combined point value of all players, but the total point value of enemies in the area can be much greater than that.
Running a stealth encounter should be a tense affair, and since stealth is not random, the tension comes from the players not knowing exactly what they're getting into. If the players know exactly where all the guards and servants are and where they're going, which doors are locked, and so on, they can plan their invasion and pull it off without a hitch. That can be a good introduction to infiltration, but for later stealth encounters placing out a few secrets for people to bump into that don't mess things up completely when triggered makes things more exciting (like it turns out the Baron decided not to go to that party after all, or the important door makes a squeak that the player character notices just in time), as does simply not letting the players know the full layout of the building and its patrols in advance. Note that guards should have predictable and sensible patrols when not alerted, because there's nothing particularly tense or interesting about the DM using their knowledge to make the guard check behind the couch where people are hiding.
Expended Resources
The recharge times for spells and items exist primarily as a method to limit their out-of-combat power. If a Legend-tier caster has already expended all of his Legend and Paragon-tier spells, for example, it would be unwise to use his full point value to determine the strength of the opposition. Using a single encounter's-worth of forces to enact a hit-and-run strategy to wear down the caster's spells over several hours, days, or weeks is perfectly valid, however.
Defeat
The players are likely to occasionally find theirselves outmatched. Be sure to have some defeat condition that doesn't involve annihilating the entire party in mind when making an encounter. Ravenous beasts can be willing to feast on some of the fallen as the survivors flee; organized people can attempt to capture as many player characters as possible, and place them into prison. If a prison is built by competent people, it may be necessary to institute some contrivance in order to make escape possible.
Terrain
Cool set pieces are cool. Many abilities are dysfunctional on a completely flat field. The random map generators for The Power Empires Campaign Setting are designed to inspire you to fill your maps with dangerous, useful, or inconvenient features.
Worldbuilding
In order to rationally interact with the world, it must have some meat to it. Although statting out and describing the motivations of every NPC in the world, every cave, and every building takes far too much time, you should at least have some idea of things like social structures, major characters, demographics, architecture, and geography.
It's very tempting, when the players provoke some form of conflict, to say "well, I said there would be a thousand soldiers in this town, so that's what you have to fight", but if you want to be able to do that sort of thing, you should consider reasonable encounter design when deciding how powerful various forces in the world are.
Introducing the World
The less players know about what's going on around them, the less able they are to sensibly strike off in their own direction, rather than bumbling about looking for something to do. The beginning of a campaign, the entry to a new territory, or when the players seem lost, is an excellent time to spring action on them to deal with immediately, especially the sort of action that introduces antagonists, reveals clues, or otherwise makes them aware of resources to make use of or problems to solve.
Major NPCs
Having a few major NPCs who try to help, hinder, or work at angles with the players can spice up the campaign significantly. From a tactical RPG angle, it's good to have some major opponent characters for the players to be fighting against. If the players could ever want to fight against a character (or the character could ever want to fight them), that character shouldn't be significantly stronger than the entire party. If the character needs to be far too powerful for the players to defeat at their current point limit (like the God-King of a continent when the party is two players with a 50 point limit), then you need a very solid justification for not having that character go out on a player-slaughtering trip. Teleportation exists in this game, so "It takes too long" doesn't work very well when the God-King can send out minions with portals to locate the players, then step through, kill everyone, and step back a minute later.
For major villains that the players are trying to foil, it's perfectly reasonable to have the villain level at a rate comparable to that of the players: after all, the villain is also a proactive adventurer of sorts.
Choosing Gem Colors and Psychic Bloodlines
The choice of which gem colors and psychic bloodlines exist in your world is significant, and should be done with some thought. Each color makes some problems easily solvable and opens up new possibilities for challenges; here are a few examples:
Orange brings time skip, making it entirely possible for people from the ancient past to suddenly appear and mess with things.
Pink provides a plethora of soul-altering effects that make it entirely possible for a person to be completely taken over and put to new purposes. In combination with White, it enables raise dead, making death merely a speedbump to achieving one's goals.
Purple supplies lots of teleportation and extradimensional space effects, making it easy for people to transport large forces in small packages, set up intercontinental (or intergalactic) passages, and easily bypass the walls of dungeons.
White fills the world with undead minions, and trace bindings opens up the possibility of locating the root causes of magical problems.
Blue brings the joys of steampunk to everyone.
Green provides a variety of cave-mining options, and earth travel renders no cave safe from intruders.
Black allows people to venture into a second realm, the astral realm, and summon creatures from it. If the astral realm exists in your world, you'll have to work out its geography and nations as well.
Projection allows limited teleportation and portals, as well as hijacking people to get into secure locations. It also allows interaction with the astral realm.
Telepathy allows for mind control and memory wiping.
Slaughter lets someone temporarily die and come back, and inflict serious damage on the scenery without carrying any weapons.
Force lets people fly and move things at a distance.
Shapeshifting lets people disguise theirselves convincingly as objects and other people.
Celerity lets someone ride another creature from a distance.
Imaginary Horrors allows the psychic to unleash an unending tide of nightmarish abominations.
Extra Advice
Handwaving
Sometimes it is more efficient to fast forward past a scene with an obvious resolution than to play it out, like a battle that all players are sure is trivial. Even challenging fights can have stretches of rounds where the players don't have any new decisions or rolls to make, such as a battle where both sides have depleted their very long ranged attacks, and now need to run forwards for a minute.
Note that this is not a tool for steamrolling over disputes: if someone thinks that the consequences are uncertain, then you should play it out so that you may educate them or become embarrassed about your miscalculation.
Making Rulings
Sometimes people may wish to attempt something that is not explicitly laid out in the rules. It is primarily the DM's responsibility to find a ruling that seems reasonably plausible to everyone, although the DM is encouraged to accept suggestions. Keep in mind that this is a game that subscribes to a level of realism where characters can punch through steel walls and become gigantic without eating a forest, so it is not strictly necessary to adhere to the physically possible in all rulings.
Refluffing
"Refluffing" is the practice of taking an existing game element, and changing the name or description (the "fluff", as it were) without significantly changing the mechanical content of that element. It is a popular technique, because it allows players to introduce themes they find interesting without risk of unbalance.
It's far from a perfect technique, however. Partial refluffing, where one character or group uses refluffed material, is a lot like giving those characters free disguises that they cannot remove: For example, if someone wants to play a battle musician, making it possible for them to beat people up with their instrument, one could refluff a massive hammer as an oboe, but then naive spectators would falsely assume that the musician was unarmed. This problem can be solved by either making the refluffing apply everywhere (so almost everybody knows that oboes are dangerous) or by making the refluffed characters famous (so almost everybody knows that this musician can beat people to death with their instrument).
Retiring Characters
Sometimes it may not be practical to bring certain types of characters on certain kinds of adventures; a dragon can't very well go adventuring in a dungeon, now can it, so the dragon player should temporarily retire the dragon and spend those points on something else for a while. Other times a player may simply get tired of a character for some reason, and want to play something else — while it's good to keep players excited about their characters, preserving continuity helps keep players excited about other players' characters, so you may want to consider getting them to keep playing with those characters for one or two more sessions, possibly with an eye open for the possibility of a dramatic exit.
The retired characters presumably continue doing something while they're retired from play: perhaps people would like to run a side-session with them eventually, but most likely you should just narrate what it was they did when it comes up. It's not a good idea to make drastic or terrible things happen to characters who were temporarily retired for reasons not entirely under the player's control, unless that player gives you express permission, but characters retired by choice are a perfect target for plot hooks.