Did I ever say I predict it based on MAL score? No. I think you just misunderstood what I wrote.
I basically just analyze the first manga chapter and make a vague first prediction. That's not only based on my own opinion, but also based on things others might criticize or like about it. Although I often change my opinion after the second or third chapter as one chapter often isn't enough for that. For example Hungry Marie had a really good first chapter, but everything afterwards was terrible which ultimately led to it being cancelled. Which is also why I only made vague predictions after the first chapter from that point on.
Of course bias might cause me to predict wrong, which is why I read through other peoples opinions on reddit, MAL and take the MAL score into consideration just to see if others think similarly or different. So far the majority always shared a similar opinion.
The only exception so far is Act-Age where I'm still torn on whether it survives or not. Although it seems that many like it, there are still lots of factors that could cause a cancellation. Having not read past chapter 3 because there are no Edited Mangas makes it especially hard. Also taking into consideration that Act-Age might not fit into WSJ makes it even harder. For now I think it survives, but I might be wrong depending on how ch.4+ are received, since I haven't read them.
But there you have it, I don't just randomly predict based on MAL scores. If I did I would've been wrong in Hungry Marie's case because it had better ratings than BokuBen at the start before it dropped down.
Of course Japanese readers might think differently, but so far that hasn't been the case. And of course there will be times where I predict wrong. But so far I've been right 14 times (5 of them survived, 9 got axed).