I don't think I have the time and energy to develop a 1 to 1000 system for you, but here is what I would start with if I was designing something. All my tiers will have 42 guilds in them (a beautiful number which is divisible by both 6 and 7). The smaller the number, the more "fair" the match-up, but the more that you'd face the same guilds week after week. On PS4 where I play, the ideal number is probably 7 or 14 at the top of the leaderboard, but could be larger further down. I can't really say.
In the first week, all guilds that opt in would be ranked by whatever is available, let's say total trophies, and then sorted into tiers of 42 (or whatever number is chosen) guilds. Within that 42, every guild is randomly assigned 6 other guilds from their tier to battle against that week. They do their battles, points are assigned however and at the end of the week, those 42 guilds are ranked. The top 1, 3, 5 or whatever are promoted to the next tier for the following week and the bottom same number are relegated to the next lowest tier.
For rewards, only the top tier is eligible for "epic" rewards, but each tier would have a nice reward for finishing in the top 3, and each tier's reward structure would be slightly more favourable than the one below. The important thing is that the rewards for every tier but the top one are modest enough that there is no incentive to tank your results so that you can get easier matchups the following week. Personally, I'd probably keep gems/gold/traitstone rewards epic in the top tier as an incentive, but spread the new troops around as the reward for winning any tier (with the consequence being that you have to compete in a tougher tier the next week). With the promotion/relegation system, guilds would tend to find their level over time, and it would be possible for a growing/developing guild to work their way up.
This would be a really complex system to manage because guilds can choose to opt in or out week-by-week, meaning that the promotion/relegation system might not result in the same number of guilds being available for each tier. You also have a problem with what to do with new guilds or guilds opting in for the first time once the system is running. I think you would need to have a developer acting as guild war commissioner to move guilds around and populate the tiers, which is probably why they went with the system they did.