The normal curve can be centered wherever the data has the highest frequency.
For example, one could consider the net worth of a sample of people who are 20 years old. Some would have positive net worth but many would have student loan debt that would make their net worth negative (they had more debts than assets). If this corresponded to enough of the sample the normal curve could be centered around a negative value and have a negative mean.
In short, yes, a negative mean value is feasible with a curve which is normally distributed.
It simply means that the values and frequency for the data you are analyzing had enough negative values that the mean was negative. If you did not expect such a result it could be that a highly influential negatively valued observation is skewing the mean. Or it could mean that there is error in how the data was entered. But it is not necessarily an error. It could simply be that your data had more negatively valued observations than positive.
Thanks for your insightful answer. I now have confidence in the negative value.
The negative values appear most in the data set after I standardized the data to zero mean and a unit standard deviation. Originally, the data set were naturally occurring events. I have two set of data that I was comparing; the same number of samples. Their means values are extremely small, but while one is positive the other is negative.
Of course, your distribution will always centered towards the value with the highest frequency. So it is very possible and i have experience it so many times.