The spellsword is the build everybody tries early and most people quietly abandon around level 20, usually because they spread themselves so thin that the sword feels weak AND the spells feel weak. I've run this build to the end more than once now, and the trick isn't picking the right spells. It's deciding, up front, what you're NOT going to invest in.
Pick a lane: melee primary, magic support
The mistake is treating it like a 50/50. A genuine 50/50 spellsword is a level-60 fantasy, not a level-20 one. Until then, commit to a weapon hand as your damage and let magic do the jobs a sword can't.
For the first 30-ish levels I run sword (or mace, if you want the armor-ignore) in the right hand and treat the left as a toolbox: Flames or Sparks for softening, Healing when things go sideways, and a ward when you smell a Destruction mage coming. One-handed perks are cheap and front-loaded. Armsman alone scales your whole game, and Bladesman/Hack and Slash give you crits and bleed without much investment. That's maybe five or six perks doing the heavy lifting.
The stat split
Here's where most guides hand-wave. My rule: pour into Health first, then alternate Magicka and Stamina depending on what your magic is actually doing. If your left hand is Restoration and Alteration support, which is what I'm recommending, you don't need a huge Magicka pool, because those spells are cheap. Heavy investment in Magicka only pays off if you're throwing expert-tier Destruction, and that's a different build (and frankly a worse one in melee range, where the enemy closes before your big cast lands).
Concretely, for the early-to-mid game I go roughly Health-heavy with a lean toward Stamina, because power attacks and carry weight matter every single fight, while my spells are background utility. A 3:2:1 Health:Stamina:Magicka feel works well. You can always nudge Magicka later once Enchanting closes the gap (more on that below).
Restoration and Alteration are the real spine
Destruction gets all the attention, but the two schools that make a spellsword survivable are Restoration and Alteration. Restoration gives you in-combat healing and wards, and the Respite perk turns healing into stamina regen, which directly feeds your power attacks. Alteration is the unsung hero: Oakflesh through Ebonyflesh is flat armor on demand, and with the Mage Armor perk an unarmored or light-armored spellsword can hit the 567 armor cap with a single spell. That's how you justify wearing robes (and the Magicka regen they often carry) instead of full plate.
Both schools level off cheap, frequent casts, so they grow naturally as you use them. You're not grinding. You're just playing.
Enchanting bridges the magicka–stamina gap
This is the perk tree that makes the whole thing cohere. Once you take the Insightful Enchanter and Extra Effect perks, you can put magicka-cost reduction for Restoration and Alteration on your gear. Head, chest, hands, ring, amulet. Stack four pieces of "Restoration/Alteration costs X% less" and those schools effectively become free, which means you stop dumping levels into Magicka and keep feeding Health and Stamina. That's the resolution to the tug-of-war: don't out-spend the problem, enchant it away.
The other Enchanting win is a Fortify One-Handed weapon enchant, so your sword hand keeps pace into the late game without dragging perks out of magic. With a decent Smithing or fortify-potion loop you end up with a one-hander that hits like a two-hander and a left hand that costs nothing to run.
If you only take three perk trees seriously, make them One-Handed, Restoration, and Enchanting, and let Alteration ride along for free.
So here's my actual question for the thread: how do you personally manage the magicka–stamina tug of war moment to moment? Do you lean on a Stamina-favored stat spread and enchant your magicka costs to zero like I do, or do you keep a bigger Magicka pool and pick spells that justify it? I'm curious whether anyone's found a split that doesn't feel like a compromise.