General Question/curiosity...

Was the watch in the same position for every tests you made? 

Did it rest for a minute or so before every test in order to go back to a calm situation? 

Was it wound the same amount before each test?

Same conditions of pressure (I guess so 😅) and temperature (might be different) ?

From what I understand, most of the time mean deviation should be considered as a mean of the deviation in several positions to be considered of interest and in the same conditions.

And a lot of parameters can influence the mesures you take.

At least, based on what I understand about all this 😅

There is quite a lot of positional variation in most movements. The same movement on a timegrapher can show +1 dial up, -2 crown up, +7 crown down, and so on. On wrist, the position will be constantly changing, giving you an average of the most frequent positions. You will probably never be able to dial a mechanical watch in to 0.1 in every position. Whatever position you left it in overnight probably runs a bit fast. But if you're getting close to 0 on the wrist, that's very good!

Every day is a school day! I am enjoying learning about all of this.

It was on its side last night, so I could see the time throughout the night, so I think tonight I will keep it face up and see how that pans out!

In addition to the positional variations mentioned above, the quality & condition  of the movement might come into play as well.  A cheap movement, or a quality movement in need of servicing, may not keep time as consistently as a well maintained, high quality movement. 

Its all very interesting! It was on its side last night as I mentioned below, so it must be the position that it can run the most free. I am going to try it on its back tonight (position I tested it in) and see how that turns out :)

As the rest here mentioned, gravity is your main factor, which depends on position and altitude. If you want to go real wild, timegraph your watch next time you're on an airplane while flying, see what happens.

How tightly or loosely wound the main spring is also factors in, unless you have a movement that compensates for this. When you have a fresh wound spring it transfers mechanical power more consistently than when it's starting to loosen at the end of its wind.

My Chinese "Winner" watch picked up 5 seconds overnight while I was sleeping, most likely due to my wrist position when sleeping with it.

I found on the timegrapher just yesterday that its accuracy is very susceptible to positional variations, backing what @tempus  said in this thread, i.e., cheap movements.

I've heard that (some?) watches with less wind in the spring run a little fast (due to decreased amplitude?). Seiko says temperature affects accuracy (cold=faster, hot=slower), and that winding daily increases consistency (since accuracy varies depending on the power provided from the mainspring):

https://www.seikowatches.com/us-en/customerservice/faq/mechanical-3

All very interesting points! 

Thank you all for your help! This is why I love this site... Full of fellow nerds 😂🍻

Yeah, gravity does affect the escapement & balance spring. It's the whole reason tourbillon's exist. 

Not to mention, along with all natural forces, gravity is not quite the same all the time. 

The answers regarding gravity are right, but the amount of tension on the main spring can also have an effect, as @AFChris mentioned. I notice that my NHxx movements, when lightly wound but perfectly running, have a lower amplitude and run slightly fast. You have to give them a number of turns from fully unwound beyond the point of when the watch starts running to get the expected accuracy. Temperature may also factor in, but it seems to me that the watch orientation and the tension of the mainspring have the biggest impact.

So it seems to be averaging out around +0.5s per day 👍🏼

This morning I put it on (left face up overnight) and it gained +8s but reduces as the day goes on! Very very happy!!