Where do I get the most accurate weekend predictions?

You don’t.  The first thing you should know about predicting box office is that no one really knows anything.

Here are some prediction sites, and their accuracy over the four months between 15 January and 15 May 2012.  With one notable exception, they are all about 70-80% accurate. There isn't really a lot of difference between them either, even though some have the advantage of making their predictions on Friday while others give out theirs at the beginning of the week. You can't single any one out as "the best".

Predictor	Accuracy
Entertainment Weekly	78.2%
Box Office Follower	77.9%
Los Angeles Times	76.8%
Box Office Mojo	76.7%
HSX Spin Factor Blog	76.6%
Hollywood Reporter	76.2%
Box Office Theory	76.1%
FilmGo	75.8%
Coming Soon	75.2%
boxoffice.com	74.8%
Variety	73.5%
Box Office Guru	73.4%
Box Office Prophets	72.8%
MTC	72.3%
RS	59.2%

Now, here are the same figures for the top ten most accurate players on EAP God, an opening weekend prediction game, over the same period.

User	Estimates	Accuracy
helmetz	45	79.43%
BigInJapan	45	78.96%
JDolphin	40	78.49%
karspov	45	78.46%
misterglass	45	78.20%
RogerMore	43	77.10%
islanderbst	41	76.58%
Oleg Max	42	76.33%
slappydavid	45	76.27%
JCRITTER	45	74.63%
Sassy	43	74.16%
BionicMoron	45	73.31%
jackfoley	45	73.13%
badams	42	72.74%
jessomine	45	72.71%

Notice how they are all about 70-80% accurate?  The same as the movie predictors?  I’m on that list, and I don’t put any thought into my EAP God predictions beyond a casual number that fits my vague ideas about how a movie will do.  I’m some random dude sitting at a desktop and I'm making better predictions than Variety.

This says to me that, when it comes to predicting how much a movie will make, no one really knows anything.

Instead of seeking out the site that make the best predictions, you should seek out the sites that have the best analysis.

There are a lot of sites that cover box office, and they basically fall into three groups.

The trackers:  Trackers make their predictions based on movie audience surveys.  They ask questions like are people aware of the movie? Are they aware if we give them a hint? Are they interested in seeing it? Would it be their first choice?  Tracking is done by companies like RS, which sell the data, and internally by movie studios and distributors, like MTC.  They don’t post on a website, but you can find their information weekly on the HSX Forums.

The media: The media, like the Los Angeles Times, The Hollywood Reporter and Variety get their predictions from their sources inside the studios and distributors.  They generally don’t add their own analysis but they give insight into what the people behind the movie (and their competitors) are thinking.

The analysts: These are sources that provide their own analysis.  Some are professional, like EW, boxoffice.com, Box Office Mojo, Box Office Guru and Coming Soon.  Others are fansites, like Box Office Follower and Box Office Theory.  This is where you go for some independent thoughts on the important questions for picking the openers – Who is the target audience? Is the target audience interested in the movie? Is the target audience responding to the marketing? – and so on.  All of these are worth checking out, and some of them – particularly boxoffice.com and BOT – will back up their thoughts by analysing publicly available data like Fandango and Flixster.  Others, like Coming Soon, will typically have seen the movie before it comes out, and will share their thoughts on how it will do.

So rather than finding the site with the best numbers, pick the sites with the best analysis, read them and then decide what you think.