Don't worry. You have a statistics major turned data science major here to break it down for you.
This year in particular on the Republican side polls have been placed in higher importance due to the fact that it determines which stage the candidates will be on for the debates. It is important to make the top 10, but every candidate who made the top 10 is doing the one very crucial thing right now: consistently polling decently.
I check RealClearPolitics every single day to watch the polling numbers. At this point, it's not that important to be trending up or getting concerned when your candidates poll numbers are trending down. As long as they are maintaining some sort of consistent polling, then they are doing fine because he or she is keeping his or her name on the map.
Three-quarters of primary voters will not choose their candidate until about a month out from the election in their state. Therefore, you can start looking at and worrying about trends at the beginning of January for the states of Iowa, New Hampshire, South Carolina, and Nevada. Until then however, I would say you should more likely pay attention to the favorable-unfavorable rankings and the percentage of people who would never vote for the candidates.
If you still want to look deeper into current polls, it is important to keep in mind the statistical errors of these polls. When a reporter announces something along the lines of "Donald Trump increases his lead by another 3 points" or "Jeb falls 5 points in the latest poll," take it with a very large grain of salt. Being this far out from an election, the polls being conducted are not going to have the same precision as before the general. Therefore, they tend to sample less people, which results in a higher margin of error, sometimes upward of 5%.
Just think about that for a second. You have all these candidates on the Republican side polling about 5%. If there is a large error in the polling for whatever reason, they could possibly actually have 10% support or >1% support. That would either put them at the front or back of the pack.
The lesson here is that it is important to slow down and not consider every newly released poll as certain, not to be challenged polling percentages of the candidates.
Finally, don't believe every poll that you see. Here's a good recommendation: if it's a reputable enough poll to be used in the RealClearPolitics average, then it was most likely scientifically conducted properly. If not, I personally do not trust the poll, unless the organization who conducted the poll has a reputable history in the field.
I bring this up due to the recent polls after the debate. People from all over the internet pointed to the Drudge Report poll as proof that Donald Trump won the debate. This poll was not scientifically conducted in any way. It was literally Drudge Report readers who voted in the poll, which automatically has a significant bias in the sample if the population you are looking for is Republican primary voters.
If there is any reason to suspect bias in a poll, throw it out. If it was conducted by a website, throw it out. If there is a significant backing of a candidate by an organization, throw it out. If you click on the poll to view the actual data and it directs you to a Twitter post, throw it out. (Yes, that did happen yesterday.)
The only online poll I would even possibly consider, though examining with much more scrutiny and caution than others, is a YouGov poll, which have shown to be fairly consistent with reputable polls. However, there are still serious doubts over possible biases still present.
Overall Lesson: Don't worry about polls so much. Not every poll is a make or break. Question every polls findings.
If you are looking for reputable sites in order to look at and analyze polling data, I recommend the following: