We revise the simplest ARED stock market model, where heterogeneous beliefs on the future prices of a risky asset have first shown to be responsible of wild (chaotic) price fluctuations. Two often unrealistic scenarios, namely traders allowed to supply shares into the market and market clearing realized at negative prices, are here prevented by limiting traders' demands to nonnegative values and considering more realistic price predictions. The numerical analysis confirms that chaotic price fluctuations are expected when the intensity of traders' choice among the available price predictors is high.