Both are viable strategies tbh. The "strong" champs are at a bit of a disadvantage because after a dozen games everyone knows how to deal with them. Show me something like jungle Trundle and I've got no clue where he is at a given time, when he goes b in his first clear, etc. I also might not know how to fight him as well, and certainly not as readily. That advantage can mean a lot if it lets you scrape a lead that exceeds the power of the "meta" champs.
Besides, you assume that picking "popular" champs always works. Trust me, it doesn't. I've seen people pick Nunu in ranked because "well they use him in the LCS so if I do nothing except try to steal dragons that makes me an LCS player right?" Meanwhile enemy Hecarim going 15/0 because he isn't limiting himself by what he sees the pros do. Not everything you see in higher levels of play applies to lower levels of play, and even when it does there's probably something cheesier. That's why if you watch those "pro player in bronze" streams they aren't doing the same cautious farmfest they do in serious games, they're going Riven or whatever and getting fed beyond belief. That's much more useful to learn from, since at least it's a strategy tailored to the elo they're playing against.
Even in the new ranked season, I had to adjust my strategy so I could climb back out of silver. The champs that I did well with in Plat couldn't carry consistently, so eventually I just settled a cheesy jungle Vi with 14/2/14 runes so that I could gank multiple times level 3/4 without going b. Too risky for LCS, maybe even diamond, but nothing's warded so I got ridiculously fed every game. Ended up winning my first 15 games with it and went from Silver 5 to Gold 5 in 2 days.
Tl;DR cheese > maining someone odd and just outplaying fools ~= meta.
