One key is in realising that the game seems to end with one fairly large railway. So, most things you do need to be targeted at getting shares in or stations on that eventual line. I've won both ways; I think both are equally viable, but the shares route seems riskier. Shares can pay off really big if you can merge in a railway you are the majority owner of (getting paid off), and then use those shares as the basis of a majority position in the new line, in essense scoring your shares twice (much the same logic as in Acquire, although the mechanisms are slighly different). The risk is that if you don't control the merger, you can end up not scoring your shares at all. This risk can be somewhat mitigated by diversity, but as always, lowering your risk also decreases your potential payout.

Stations don't score at mergers, but they can score incrementally through the Railways Hubs. I had great success in my first few games by getting stations on the LNWR and scoring up to $15-$20 early just by driving it to all the Hubs in the vicinity. In general, though, you don't seem to get dramatic incremental scoring with stations; but when you do, it can be huge (I've seen $20 in one move by connecting the LNWR to a Railway Hub late in the game).

There seems to be little point in taking goods, especially early. I've won 5 of the games I've played recently and never have had more than 3 goods disks (and they were all taken very late in the game to guarentee they scored). Certainly,  I've seen very low-scoring games (winner having around $36), and perhaps the goods would have been more attractive; but, I think what makes the game low-scoring is intense competition for stations. So if one person were to go after goods instead, the game score would go up, making goods less attractive. So I dunno. I'm going to try one game where I go after goods primarily, but the total available score is only 36 + bonuses, rarely enough to even be close.

All the railways are not created equal. I like the LNWR (nice fairly central location, and proximity to a lot of cities and especially hubs) and the SER (constricted early options makes it easier to get stations on) for the first move. The NER and the orange make good alternates because they are centrally located (and so can merge with whatever becomes big) and can easily build up a bit of value before merging.