Hi--it might be a combination of two things:
(a) Live sound isn't inherently symmetrical in the way some people seem to expect. Think of what happens when a drum is struck: There's a sudden positive increase in pressure, followed by restoration and some ringing--but nothing else reaches the level of the initial attack. Thus the net "area under the curve" will be greater on the positive side than on the negative side.
(b) Some microphones, some cables, some preamps and some recorders invert the polarity of whatever they record. There's a widespread notion that "absolute phase" isn't audible; some equipment manufacturers and recording engineers aren't careful about it. And polarity inversion can easily occur when you connect balanced signals to the unbalanced inputs of some piece of equipment.
If you're recording sounds of type (a) on equipment of type (b), you can very well get waveforms like the ones you're seeing, even when there isn't much DC offset in the long term.
Inversion of signal polarity is definitely audible to some people with some sounds on some playback equipment. It's clearly not as important as some other characteristics of recordings (e.g. "did you remember to turn the recorder on?" is more important in my book), but it's not a total non-issue, either.
--best regards
P.S.: Looking again at that waveform, I'd guess that it's in inverse polarity overall--but I'm also struck by the near-constant limit of the excursions in one direction. So maybe something else really is going on here other than (or in addition to) inherent asymmetry of the "live" sound pressure levels.
tailschao, do you have access to a tone generator? You might connect your preamp and recorder just like they were, and with their gain settings just as they were. Feed some microphone-level test signals into the preamp at various levels, record the signals, and look at the resulting waveforms. If they're not fully symmetrical, something is distorting. It would be good to rule that out if you can.
As an example, the attached jpeg image shows the result of asymmetric "soft clipping" of a sinusoidal signal. This corresponds to a significant amount of distortion that includes odd-order harmonics, which are generally more audibly disturbing than even-order harmonics of the same amplitude.