Help me out. The direction of what, seriously. You have a bunch of individual averages based on a wide variety of data, compiled with extreme variances..... let's say, hypothetically, some had 3% some had 10% - some had likely voter, some just adults and some had 500 people and some had 1500.... some +5 dem and some +7 dem... and some only had 5% independents: how can the average of anything, especially polls, tell you anything.... anything at all.
Hell, in the business world I would never try to average test market results.
You have a bunch of individual averages based on a wide variety of data, compiled with extreme variances..... That's the entire point of aggregation like RCP does.
If you have a diversity of polls with different methodologies and track each of them over time you can see movement trends.
If almost every poll, regardless of it's bias, is moving in one direction over the course of several weeks that's significant.
Does the average lead mean much? No. But the direction of sentiment does.