A large trading firm, particularly one that had the capability to trade cheaply enough to take advantage of this effect would never do this simply because 50%+ of the return of a single stock is determined by the return of a common market factor. If you take a position in a long only stock portfolio, you may own many stocks but statistically, particularly if the portfolio is only held overnight, it's a large bet on the direction of the market plus some noise. So you are taking a massive binary bet that is going to go wrong anytime the market opens down. This is why most high Sharpe ratio equity trading strategies involve short selling even though equities have a strong upward bias. If you eliminate the effect of the market factor, you have a portfolio that is made up many statistically independent bets.
For example, if you take a bet that is expected to make +/-1000 dollars with an equity portfolio, and it's long only, it's like tossing one coin to determine whether you make 500-700 dollars of the 1000 and then tossing coins repeatedly to determine the outcome of the remaining 300-500 dollars one dollar at a time. If the portfolio was market neutral, i.e. long and short in equal amounts, it's like you toss the coin 1000 times, each coin toss determining only 1 dollar of the outcome. If you believe the coin is biased even somewhat in your favor, as it usually is when you are trading a high quality strategy, the second option is far better.
Given the kind of risk-adjusted returns achievable by a firm that could trade cheaply enough to take advantage of this effect by other means, even using relatively simple well-known strategies, doing this would hurt their risk adjusted return and the returns they could generate on their risk capital. So the odds that there is a large long-lived firm doing this to mark their book and causing this effect is likely close to 0.
Others have mentioned several explanations for the effect. To this I will add one more, which I think is at least, if not more likely than the rest. Retail day-traders, some of whom are larger than people realize have a long bias, ie they are far more likely to hold a stock long than short sell it and they tend to close out their positions at the end of every day.
Seems like the authors have an axe to grind more than a point to make. The paper reads like the kind of certainty a completely uninformed but (markets are made up of evil people) true-believing crusader has. To me it seems like the kind of situation where someone doesn't yet know enough about a subject to know how little they know, in this case buttressed up by the broken credentialing system we call our education system.
The claim in my opinion is complete nonsense.
Edit: Fixed grammar.