It's generally a bad idea to put absolute price (or volume) data in ones algorithm. Not always bad, but your algorithm highlights a major reason not to - stock splits. VXX is a textbook case.
The algorithm chooses absolute price points to open ($13) and close ($35) a position of VXX. The algorithm runs between 9/20/2011 - 1/28/2018. The problem is there were four 1:4 reverse splits during that time (10/05/2012, 11/08/2013, 08/09/2016, 08/23/2017). Those splits show up on the custom data chart in the algorithm as 4x 'spikes' in the price. However, those 'spikes' also correspond to a 1/4 adjustment to shares held. When the algorithm sells at $35 (believing a big profit) its really selling at a loss because the shares got reduced by 1/4 in the reverse split.
This can be seen in the algorithm 'Transactions Detail' tab. On 7/17/2012 there were 15 shares bought at $12.72. After the stock split on 10/5/2012, there were 3 shares (15 / 4 = 3 because one can't trade fractional shares). Those 3 shares were sold a few days later on 10/9/2012 for $35.96. While it seems like selling above $35 would result in a gain, the shares were actually sold at a loss. The initial value was $190.80 (15 x $12.72) but then sold for $107.88 (3 x $35.96) for a loss of $82.92 (excluding commission costs). This is a little misleading because of the rounding down of the shares but it illustrates the point.
The reason Yahoo prices look so different is that they are 'adjusted' to todays dollars. The prices being displayed in the custom data are actual (unadjusted) prices that one would have seen on a specific trading day. Hope that helps.
