Cumulative Returns Calculation Giving Astronomical Result

Question

I am backtesting a strategy and the cumulative returns get absolutely ridiculous as time goes on.

Here is how I am calculating cumulative return:

pred['cumulative'] = pred.position.add(1).cumprod() - 1

Where position is the pct. return series of the strategy.

pred.position.head()

date
2015-12-30    0.065293
2016-01-04    0.039952
2016-01-06   -0.050252
2016-01-07   -0.009129
2016-01-08    0.018204
Name: position, dtype: float64

As you can see the cumulative returns are off the chart:

The returns however remain relatively constant:

Any help trouble shooting what I am doing wrong would be useful!