This is a truly fascinating bug. If I run a screen as of ten years ago today, 1,171 out of 3,727 stocks have SMA(10,0,#Vol) = AvgVol(10). The others are way off. 90% of the time, when SMA does not equal Avg, SMA is larger than Avg. I tested with FutureSplitFactor, and splits have nothing to do with it. Neither does StockID (I thought perhaps stocks with higher or lower IDs might be more susceptible to this, but no). I also tested to make sure that AvgVol(10) was accurate, and indeed it was. The median market cap of companies with a divergence between Avg and SMA was significantly higher ($1.8B) than the median market cap of companies with no divergence ($620M).