In this version I'm rebalancing 10 minutes after market open rather than 10 minutes before close. This makes sure you're closer to your latest information (yesterday's close), and makes sure you actually get filled (the log said most of your orders weren't filled).
Another thing is that you might want to order every day even if you're already holding some of the symbol you want to own, because otherwise you might have a tiny amount of cash sometimes (leftover from unfilled order or ETF was too expensive), but this has just a minor influence.
Anyway, it doesn't seem to help much at all. Note that the paper is from 2013 so you might need to re-configure the momentum lookback period k. Or it's just another rubbish paper.
NB: I also set slippage and commission to 0 in the attached backtest, because that's what the paper does as well (yeah...).