I have recently done an intense of deconstruction of rpg, going from dumbing it down to its simplest but essential elements (dps/level up) to rebuild it piece by piece, mechanics by mechanics. I should thanks people who help me understand management games (I see you in this thread, hello!) because it was essential to understand everything else.
First, in game term what is strategy? Strategy is choosing the right planning to achieve a goal (different from policy/politics which is choosing the goal/ethics and from tactics which is about executing the plan). Planning mean thinking over time the sequence of action that lead to the goals.
But if the plan is already known (matching the right plan to the right situation) it's kinda boring, that would be a first class of "strategy", not very demanding, relies on memory and pattern matching, basically a puzzle at best, weakness system are just that. But for a lot of player that's enough. However challenge decrease over time towards zero as the player get real experience with each situation, and it can converge pretty fast if the player is accustomed to similar games with similar mechanics ... until it go through the boredom ceiling (jrpg tend to do that for some people). Mastering the initial variation of the set of gameplay is fun until it sink in and became automatism and the experience start like a grind (hope the scenario has you immersed by then
Roguelike is a genre often lauded for giving player more strategical RPG experience, it does so by upping the uncertainty of situation (opposed as randomness like jrpg encounter) through scarcity and tension (rogue like use permadeath, it is super effective!), randomly generation prevent situation that repeat themselves too often and actual bugs can turn into feature by giving unexpected elements and situations. Because uncertainty is high as death mean a reset of the experience (but manageable) the player is constantly evaluating the pursuit of options through risk/reward assessment. Rogue like work because the player swing a lot between power and weakness, as she swing he goes through an emotional roller-coaster where every random opportunities serves as a "haha!" (i'm smart!) or a "hoho!"(bad luck), she change her tactics to reflect that. Maybe she get too confident and reckless vs carefully exploring unknown items while in high hp or maybe she get kamikaze vs sneaky while in low hp, that's not the same experience. It works because there is a definite goal to achieve and a definite metrics to judge efficiency of action, actually situation are always "clear", the uncertainty being a known unknown we can assess.
While a planned experience in a game (exacerbate in jrpg) rely on a constant influx of new "fair" situation by tossing content, roguelike ensure variation by endless permutation of existing content.
But what would make a real intricate strategy? There is two main things, complexity and depth:
Complexity is the breadth of option, the more complex, the more it is hard to keep track of all variable, the more focus and attention must be bring to decipher data. Complexity does not necessarily translate in intricate strategy, but complexity can give the illusion of strategy and actually please the player by making him feel smart and provoke a sentiment of mastery if he is capable of simply reading the data. Whole game are entirely about the deciphering of the huge spreadsheet to find the hidden optimality of the system, more than once there is an uproar when designer streamline system because of the unnecessary and technically poorly balance. For a lot of player deciphering IS the gameplay, it's true that's a kind of puzzle, it is a gameplay. Most dissension therefore is just a matter of discussing what is the focus of the game, is it the understanding of the data or the manipulation of the data, simulationist tend to love complexity for the sake of it, even if the underlying system is simple, that's how they immerse themselves in the game.
Depth is about anticipation, it reveal itself through time, how much ahead you must anticipate, generally the more step in the future the more depth, and as depth increase the more difficult it is to keep track of accuracy/breadth of data and the more bet you have to do in choosing the right path. Intricate depth is about long term advantage over short terms, it's about deceiving your opponents by making decision that would be stealthily critical. Depth can be poorly balance, slippery slope is depth gone wrong, when you realize it's long locked in a state and there is nothing to do, similarly when a system is too complex that is essentially random as nothing can be predict with enough accuracy. Generally long terms decision can be created through production and management systems. Management is about smartly using resource, production is about resource that grow over time. And you can have one without the other, mana system without way to replenish can be seen as management, hit based super gauge in figthing game is more production that build opportunities (options) than a management (as such it's more a pacing system). Production and management build depth by interlocking opportunities, efficiency, options through feedback loops, positive and feedback loop can swing advantage over time, starting weak and growing stronger (or the reverse) or acting cyclically, a player can totally deceive his opponent by appearing weak and grow godly advantage over time either by buffing himself or debuffing the opponent or both. But management and production is not the only way to create depth, positioning is great too.
Basically positioning is being in an advantageous state by navigating the graph of all states, with some state distant of many step, the beauty is that positioning is at best relative to the opponent own positioning. Positioning can be any state, spatial positioning or weakness type to class or weapon type vs armor. The way you traverse the state space is also a way to deceive your opponent, going too clearly for good position might broadcast the intent, by taking convoluted path (especially if it makes you look weak to the opponent) is like the icing of the cake of depth, but that's assuming the state landscape allows it! Intricate positioning allow for tricks and is always highly situational, it is the very definition of outsmarting your opponent. Intricate positioning is also tricking your opponent into a less favorable state position, basically cornering/trapping him, especially when he thought he was making good move all along, as good position can be turn into situational weakness, of course if the cost of moving is pointless it might swing advantage to strongly to actually generate true depth.
Depth and complexity alone does not make a good strategy design, there is a bit more. The first thing is that all player don't see strategy under the same light, and there is a lot of idealization going on, especially when they think confuse their favorite aspect for effect it may not have. Mannerism for example tend to favor any kind of complexities (even superficial) for strategy. Good competent strategy design allow for improvisation, ie do not lock down the player on a failing state long time ahead, and would not need abstract mechanics like rubber band to fix himself. Player are keen to choose path of lest resistance, there can be obvious path with optimal advantage, less obvious path must still be open, and at best even in desperate situation there must be a way out even if it is difficult to pull out. The solution is generally to have multiple way to win that is not dependent on advantage, only supported by them, chess is a great example of that as many pieces does not guaranty victory, the right piece at the right place can still do all the works. Strategy can be artificially augmented by increasing "obfuscation of the state of possibility space", basically the more complexities in reading and option offered and the more step of depth through feedback loop and state movement the more analysis,, focus, attention and synthesis it is required to retrieve essential information to decide the right move, a small goban has reduced possibility space to allow beginner to practice go. For me complete information system with competent strategy is the finest strategy because it really emphasis outsmarting your opponent as your moves is laid bare for him, hidden information tend to have blind bet and uncertainty favor "averaging" ie preparing for all eventualities instead of improvising and adapting to the situation (unless you have random map). Another element of competent strategy design is simply trade off, it simply encourage .
Given this analysis, can arpg have strategy? My answer is a fat YES! but you have to design for it, action tend to favor short terms decision with short terms consequence, it mean that strategy is shallow, another things is that the future is uncertain in a bad ways and negate informed decision, it devolve in blind specialization OR averaging, you don't have a mean to anticipate meaningfully. Aspect of management might increase strategy by shifting approach but is generally negligible because it's too linear and base on optimum (buy max positions to quaff, get best armor and high stat weapon), access to items being streamlined for short terms and instant gratifications (loot, pet, shop, portal), but those convenience are there to mask the lack of meaningful alternative. Modern focus on overarching aspect like main class stat, dps, skill spamming with cooldown, created to replace clunky system like potion quaffing only emphasis short terms. Basically spam skills, quaff potions, loot, sell (portals/pet), equip best, buy potion repeat. The element of strategy is not absent but minimal in aspect of mob control, except good mob control does not feedback into consequence other than health consumption. First play are basically random encounter as there is little to anticipate future, which is often mean just have best weapon and max potion since it's a linear power scaling, and in case you can anticipate the linear scaling just mean rambo through the weakness, there is no trade off.
Roguelike get away with more focus with survival and meaningful uncertainty, which encourage trade off in the way you manage available resource and information to take advantage of situation. For example potion has known effect ... except you don't know the association of color/effect with each playthrough, finding a potion might occur in various situation with wildly different stakes, meaning discovering the effect is a clear decisive actions: imagine you are chased by a level 10 dragon, too bad you are a level 5 bard and you are low on health and low on chanting mana that could tame the dragon, what's left is a unknown black potion, will you quaff the potion expecting good things to happen or toss it on the dragon expecting it is a poison, you know it's not blindness (already the red potion) nor it is invisibility (pink potion), maybe you can try to run into fog of war expecting another creatures distract him but that might have you kill if you aggro both of them or maybe you can hide behind a tree, then let him burn the tree and then use the fire to have fire arrow that do much more damage that will buy you a bit more of time?
Even a game like darksoul has meaningful decision, will press your luck while packed with soul and backtrack to unload them, would press your luck moving forward or go to the bonfire to replenish at the cost of respawning enemy on the sector? Strategy is not just on encounter but layered on navigation and exploration too, as long it is informed and meaningful, does it have consequence, does the advantage shift.
My playthrough in skyrim is also pretty meaningful, I have spend all my point into stamina and skill point in sneaking, too handed and marksman. With the autoleveling i'm a glass canon (no armor, no shield, no skill armor) and every step is fill with danger, the more level I get the more meaningful sound and position get, I'm overpowered, enemy go down fast but so do I. If not careful almost anything kill me in no time, that's where the gamedesign shine, it's full of tell, sound, animation became intense alarm and not being cornered mean I anxiously monitor escape route, I have a big stamina bar so running is an actual option as much as punching in the face, but I must be sure that my hammer connect or might die the next instant, since not all hit stagger enemy, keeping the right distance is very important to not fall under a retaliation, even being on horse is not safe so i must plan carefully how I dismount the horse (or run for my life without taking time to check up any real danger)! It really feel glorious. Skyrim is really flexible, you can go power ranger and build a non stoppable tank even cranking difficulty, or you can choose the right skill that change the game forever (never heard of the illusionist run)? It's all up to you.
To contrast, in torchlight my archer was overpowered and didn't care as I maxout her dex skill through slot, gem, weapon and skills, once I got the bouncing arrow I was unstoppable and little everything else wasn't very meaningful, I didn't even bother leveled up the pat which was a non stop mule as I would send it town as soon he got back selling stuff, I never run out of anything and only died twice until the last level. It match most my experience with (a,w or j) rpg in general, specialization tend to overpower you or totally break you. Only once in a while there is an encounter that is a bit more puzzly than other but once you figure out it's a matter of spamming the right spell in the right sequence, secret of mana I finish the game once with all magic max out but on the replay I only need the tier of leveling and even that weak no boss could even scratch me. I wasn't even thinking too hard (btw atb + moving, love, p2).
Despite being dumb down for kids, pokemon has a pretty good system with all the right limitation and allow you to anticipate your run while giving you plenty way to customize, also depending on your need (for exemple winning or capturing) it totally change your approach (and it's case by case, you don't deal with each pokemon the same way which also depend on your team in your deck and their personal stats). Because situation are always changing (capturing a rare low level pokemon when you just build a team packed with the kind of ultra laser meant to destroy rival team demand a surprising kind of finesse and planning, having a "weak" pokemon in the team in case of also mean it's not an asset in case of stronger opponent sudden apearition).
FF13 tried "something", the stagger system does enforce a minimal strategy (you need to weaken enemy by filling the stagger bar with weak attack before pilling combo once staggered to have a nice feedback loop that do more and more damage under a pressure time as the stagger bar slowly empty itself back) but it makes the game even more tedious by forcing a clear pattern and making any battle puzzle that are just too long and too frequent (I stop playing after 10h of tutorials ...). The swing between losing and winning was a bit too much, you either crunch them or they crunch you, but once you figure out the encounter pattern it wasn't much a challenge, just grinding and padding, plus the stagger system overshadowed all other option (even invoke, square U bummer). However there is some nice idea with the paradigm system for action game.
In conclusion, strategy is possible even in action game, it's not just about pausing, it's about having the tools and option to think ahead, feedback loop and resource management can greatly help, but having a lot a varying and shifting abusable situations is more interesting. In action game, planning is more likely to be layered on multiple gameplay (battle, encounter chaining, navigation and risky exploration, leveling) but only if it is possible to make informed decisions and the world actually react to these decision in predictable but complex way. Linear scaling is the bane of strategy.