The TAFs are produced by people that do it all day long, and they have access to data which nobody else gets - certainly which nobody else gets unless they pay serious money. So, I doubt that, statistically, anybody is going to beat a TAF for reliability
It would be great if the above statement were true. But I very often look at the TAF, and then look at other information from the Met Office (predominantly MetForm 215, rainfall radar, and METARs from places a weather system is coming from), and see discrepencies between them. And when this occurs, it is invariably the graphical data which is more accurate. I don't know whether the people who write TAFs make use of this same information, but it seems to me they don't.
And while we're on the subject, I'm informed by a source that I believe to be reliable that a station must produce 3 consecutive METARs before producing a TAF, because the information in the METARs is used in deriving the TAF. I am also informed by the same source that the advice given to licensed Met Observers when creating a METAR includes reading the TAF. I'm sure you can see the obvious problem.....
I certainly don't claim to have a "superb understanding of how weather works", but I do claim to have access to a wide range of official weather data from the Met Office, as well as from other sources, and to be able to interpret all of this information, and I believe this gives me a far better understanding of what the weather is going to do in the short term than reading just one source of information (a TAF or anything else) would give me.
But none of this is any help for forecaasting over longer periods.
FFF
--------------