Basis of short-term and long-term estimates
The basis for the short-term estimates is the current situation in the company: the management guidance, cyclical situation, analyst's own assessment of the competitive situation and market. However, the basis for the long-term estimates should be more the long-term growth of the sector in general.
Levelling off long-term growth
All rational company estimates are based on the fact that sales growth levels off near to the long-term GDP growth (about 3%), no matter what the business is. This means that also growth companies become "ordinary" in the long-run — even some software or mobile phones cannot grow more than 5% a year forever. Note that there is even a mathematical problem with high growth rates (higher than WACC), which makes fair value negative with high growth rates.
Profitability
Even more important for the estimates is that profitability should be levelled off to normal levels. Profitability is estimated/changed in the model with EBIT-%, but you should follow profitability with ROI-%or EVA which are much more relevant measures. In our competitive world we have to assume that sooner or later competitors will catch any company in the long run and "abnormal" profits (large positive EVA, greater ROI than 15%) will melt down.
Star companies
Of course there are companies that will not follow this rule during their next 5 years, but it is no sense that we try to "see" it in our 5+ years estimates. Those companies that have grown very rapidly and have been very profitable are for example Nokia and Microsoft. Their share prices have appreciated about 10 000% and 300 000% correspondingly. There is no sense in trying to catch that kind of stars: it is enough if we find that those companies are more than 50% undervalued, because they will grow more that 30% for the next 3 years and remain very profitable.