Forgive me for starting with a bit of a skite but who else but you guys could appreciate the how good it feels to pick 6 winners in 8 races for a POT of 270%. (I did ********** a bit, I backed two horses in one of the races).
I run a computer based system (with a ten year database) which uses little more than well tweaked winning margins, riding weights, and jockeys to generate ratings for horses on all races in WA. No account for barriers, going, preferred distance or course and just a small penalty for older horses returning from a spell.
The funny thing is year after year the system performs badly at some courses (Ascot for e.g, always -10%) and year after year it performs well at others (Belmont nearly always +10 to 15%). Not only that, some courses have pretty consistently good months and bad months. Pinjarra for example Nov +30%, Dec +40%, Jan +25% and the rest of the year crap.
You need a lot of results to work a system on a by month basis (which is why I stick to WA - easily available data) but (if you haven't already) you may want to consider the time of year and course when you run some of your systems. It could explain some runs of outs.
I have a few theories on the whys and wherefores of this up and down performance but I would be interested to hear what you guys feel the reasons for this may be.

Now, I'm just going to look for that thread which asked if you should give up after two winners. Nah, better not, I'd only start skiting about my $12 winner in the last race. :-)