Which model can predict the weather most reliably?

Being sick of inability of any models to forecast the weather more or less reliably even couple days out, I really wonder about your opinion on this topic.

I understand that weather prediction remains as tricky business as it has been dozens years ago.
Long range forecast sucks in its nature (except a general statement about the global warming, however debatable by many), and it does not matter who makes it (Henry, Phillip, Brett, Snowstorms or Torontoweather, or Farmer's Almanac - no personal offense please!).

But it seems that even short range prognosis is not much better.

So my question is, which model and for how many days out is the most reliable one in terms of weather forecasting. Say, Ruc gives 100% accuracy within 2 days, while ECMWF gives 80% accuracy as far as 4 days out.

Something like that. From that moment I will be using only that model and only as far out as it forecasts accurately.

Otherwise all such "forecasting" is not much than "reading the tea leaves"

Alternatively, I may go back to the days I was a kid, and predict weather by looking at the type of clouds, direction of the wind, color of sunset and behavior of my cat and other animals around. By the way, quite reliable forecasts were made that way!

Regards,

Your favorite website for model forecast

Hi,

I use mainly these tree:
http://www.arl.noaa.gov/READY_animations.php
http://coolwx.com/cgi-bin/getptype.cgi?reg...amp;field=ptype
http://weather.unisys.com/gfs/misc/gfs_cum_prec.html

Can you suggest better/more informative ones?