Testing 2 More Oil ETF

After I mentioned the study I did on oil ETF, Phil at Phil's Stock World mentioned 2 other ETFs meant to track oil prices - OIL and DBO. I ran 2 quick models and it appears that they suffer the same problem as USO. Here are the results. OIL first:

Even though this tracks well through the big ups and downs of 2007, lately the model sees a divergence. And it is confirmed by the relative performance of OIL and the oil futures over the same period:

Not sure what gives. This seems to be affecting all the oil ETF. How about DBO:

The model tracks incredibly well the first 700 trading days, but since then, we see the same divergence. In terms of relative performance, DBO does better than all the other ETFs. 

But still, 2010 is tough! I would like to see a good explanation from someone on this problem. It's irritating to see a model works with over 90% accuracy for 3 years go haywire in the one year.