Andy, As one of the fellow spongs who didn't look for or find the magic formula, I feel I might be in a better position to commiserate. I've been playing at this perl golf for several years now, and sure, I've learned some tricks along the way. But mostly I play to the level of my abilities. If I come up with a good algorithm, I usually do well. In this case I did OK. In other contests I've done very badly.
I've taken a peek at your final answer, and it is short. There is nothing that strikes me as obvious as to where you can remove characters. Therefore, and I think this is the important bit, it's a bad algorithm.
I haven't had the time to look through all your solutions, from your first entry to the last one, but how many different algorithms did you try? Me, I tried it four or five different ways until I stumbled upon one that seemed shorter. Heck, I thought I was a freaking genius when I thought of y/IVXLC/XLCDM/ for multiply by ten. And having spent some time post-mortem with Ton's magic formula, I believe that I'd have only saved 3, maybe 4 (at a stretch) characters with it. The other parts of Ton's (and everyone who beat me) winning entry were the majority of the difference between him and me.
Of all the entries that beat you into 28th place, how many used the magic formula? Not as many as you seem to think. There are certainly a number of 'in the know' people (Yanick, Michael Wrenn, tybalt89, Honza - me!) who didn't use/find the magic formula. There are also people (experienced golfers, too) who did use it and didn't get within 40 characters of Ton.
Take a step back and think about how you sound by suggesting that your final position is far away from where you would have finished. Because it's probably not true.
p.s. sorry this is waaaay late in the day - someone only just pointed out this thread to me - perhaps if I'd kept an eye on perlmonks, I'd have done far better!