A universe selection based algo I made for 2016 would seem like the best algo ever, achieving 10000% in 2016 or so. Upon closer inspection, certain data is very suspicious. I suspected splits initially but after trying to look them up I don't think they are the cause. Example case:



So for some reason, in the QC data, TVIX jumps from 11.825 to 269.75. There was *supposedly* a 1 for 25 split 08/09/2016, but that's a different date.

I'm trying to figure out because this is completely taking my backtest results into fantasy land.