There seems to be some misunderstanding about VP on this and various other sites. Quite a few people appear to think that it is a 'system', which can be judged by how well you perform using it. It is not a system, but a piece of analytical software that generates things like moving averages on the basis of neural network prediction.
Blaming the software for poor performance would be like blaming eSignal or Sharescope for the same thing. All these programs analyze price feed and generate indicators, among other things. How you use them is up to you. Finding a reasonable way to work with VP is less hard than I expected, and it is now something I always use, though with other programs/indicators as well. In my view it adds a lot.
Worth mentioning two interesting threads - one which doubts that VP uses intermarket data at all, the other that you could get the same results with conventional MAs based on single market data. For each market in VP you can find the other markets that are used in intermarket analysis, and I think it very unlikely that the VP folk are telling a deliberate lie, particularly since training neural networks using intermarket data is not enormously difficult, and those who use the program will have noticed that each time new EOD data is provided the calculations take a fair amount of time. That does not suggest to me that VP is just using single market data.
For FX, in particular, the intermarket basis may be slightly obscured by the fact that so many FX pairs are highly correlated, and rather obviously these are the ones used in training. So the suggestion that you could get the same results without VP could be true. I have looked at various 'projections' of MAs into the future and am so far unable to replicate VP predicted MAs. Maybe I haven't tried hard enough, but if you can't replicate, then VP is unquestionably offering something new, though as I said already, it's up to you how you use it.
To my mind, anything that offers a new kind of indicator, particularly one that claims to be 'leading' not 'lagging' is worth taking a good look at. There aren't too many of these around - MACD divergence, pivots and fibs to some exent. Can't think of too many others.