The sword cavalry don't cost that much actually. They cost 80 food, 35 wood and 20 metal - and a buildtime of 12. They also cost 2 housing, which has a nominal cost of 30 wood and 9.6 buildtime (by a citizen). Compare with a spearman, which costs 50 food, 50 wood, buildtime of 10, and a cost of 1 housing. The adjusted cost with housing is thus: Sword cav: 80 food, 35 + 30 (65) wood, 20 metal. Spearman: 50 food, 50 + 15 (65) wood. Food is the easiest resource to secure so the extra food doesn't matter much. The wood cost is identical once adjusted for housing (although cav are easier to keep alive, so might be more likely to keep occupying their housing). The metal cost is low, and at that stage of the game metal is easy to secure. Your starting 5000 metal supply will be good for 250(!) cavalry and if maxed out on workers will easily support 3-4 barracks non-stop batching 5 sword cav at a time. With aggressive harassment the critical resource in the early game is really wood as on most maps the easily protected supplies are quickly exhausted. As calculated above, cavalry actually cost the same amount of wood per unit as spearmen and archers. They also only take slightly longer to produce. Thus I think it's not unreasonable to produce armies of cavalry of comparable numbers to soldiers. If you build 5-6 farms around your CC, and put 10 guys on metal, and harvest enough wood, then you can easily run 2 barracks non-stop producing batches of cav, 40 cav is only 4 batches and would take around 3 minutes to complete. Of course in an ideal game you don't just produce 40 cavalry and have them sit on their arses until you charge them into the enemy lines to die like idiots, you rally every batch to the opponents base and have them run in and generally harass and contain the opponent. Weak cavalry should be hotkeyed out and sent to garrison in the CC allowing them to heal up, saving cav in this way can easily accumulate an extra 10-15 in the CC and they'll usually be veterans too so have better stats than fresh cav. I also did some experiments with the sword cav to determine their damage rate in practise. It turns out that sword cav kill buildings (houses) about as quickly as battering rams (per unit of housing - a battering ram is 2.5 times more expensive and kills a building a little faster than 2 cavalry). Their rate of damage to buildings is thus equal to a top-tier siege engine. Of course in practise their rate of damage is higher, because they move so much faster. They're either easier or harder to kill depending on whether the enemy is using hack or pierce. I think that sword cav, used in a dynamic harassing style with the aim of denying woodcutting and ending the game quickly, are clearly a 'killer strat' for Celts. Every civ needs it's killer strategies, and the Celts don't have anything at City level which could qualify as a killer strat. The only really notable City unit is the battering ram, which might be required for cracking open a determined turtle but otherwise won't win any battles. The question is whether at their current cost, sword cav are overpowered, if the celt's killer strat is better than other civs killer strats. Only testing will show this for sure, but I suspect it is so. I suspect that sword cav are too cheap and quick to train for their effectiveness. They may also be available too easily (many RTS games require a stables building, before the high speed high damage harassing unit can be produced). Celts themselves really have nothing interesting at City level, except battering rams and heroes. Other civs have quite good champions, but as a rule I think that champions are more housing effective, but are less cost effective. In other words, they really only come into their own once players are maxed out. Upgrades can matter, but again only once the game has reached a point where players aren't so much resource limited, as housing limited. A good celt should have won the game long before this happens. The roman spear cav which is available at Tier I, has a nearly identical cost to the sword cav, it costs 80 food and 55 wood, 1 housing with a build time of 12s. The only way it is cheaper in that it costs only 1 housing, I would say that 20 metal 35 wood is 'cheaper' than 55 wood because you can just put citizens on metal and forget about it, while wood entails more micro. It deals about half as much damage to units (due to it's very slow attack rate), and much much less damage to buildings. They can certainly be used to harass an opponent's resource gathering activities and are okay in a fight, but are basically worthless for razing buildings. The enemy will have ample time to react and bring in skirmishers/archers. It's basically a far inferior unit and the only thing going for it is you can get it at the start of the game. I think I'd prefer a ranged cav though, especially since it can't realistically kill buildings anyway. Also later-game spear cav, including champions, don't have much to recommend them over sword cav, they can't kill buildings or siege engines quickly and don't really complement ranged units in any way. So all things being equal, a civ with sword cav is stronger than one with spear cav.