The problem is that most people don't have the 'investor' mindset. People are myopic, impatient, subjective perception of probabilities is heavily biased, fail to consider opportunity costs and expected rate of returns.
For example the principle of optimal selling date: when the expected return on the current value of your asset is surpassed by another one after the transaction costs (if they have the same variance). Simply, if you have the option to switch for another asset with a higher expected return. The grave mistake people do here is taking into consideration the original price they bought the asset at. It doesn't matter! Even if you received it free! Only the expected growth rate matters! They like to tell themselves that 'Oh great! I made 100$ on it! It's enough!'. But if you reinvest that 100$+original price into an inferior asset you just made a wrong decision by selling.
The issue here is how you calculate expected return and the corresponding risk. The investment valuation method most people use here, even without knowing its name, is the 'comparables' approach. However, since the LEGO market is not exactly consistent over time and largely volatile, predictions are mostly inaccurate and people have to rely on their instincts.
But, the essence is still the same. When you want to make a decision about selling or not, you have to ask this question: if my asset has an expected return R, can I find another one with R'>R with transaction costs (shipping, fees, etc.) accounted for? If not, then just keep the set. Already realised return is irrevelant!
For exclusives I found that 2-2,5 years is the sweet spot.