When does my fidget spinner spin longer? Just wondered if it has an impact on duration of spin if I hold it horizontally or vertically. After a few tests, I could not figure out which was actually longer - beside the fact that I can't put the same power on it twice.
 A: Without knowing the details of the construction of the spinner it is not possible to give a definitive answer.
The spinner can be modelled as a single rigid disk which spins around a central axis, supported by some 'bearing'. 
The work done against friction when the disk rotates through the same angle vertically vs horizontally depends on the torque due to friction at the bearing. The disk will be supported by different contact surfaces in each case. Assuming the coefficients of friction are the same for both contact surfaces then the resultant normal and friction forces will be the same in both cases. However, the normal and friction forces will be spread non-uniformly over these surfaces, at different distances from the axis of rotation, so the resultant torque may be different in each case. Without knowing the details of how the disk is supported by the bearing in each case, it is not possible to say which case results in less energy per revolution being lost due to friction.

It is not very difficult to determine experimentally in which case friction losses are greater, if you have access to a stroboscope. 
Set the spinner going then gradually increase the flash frequency of the stroboscope until the spinner appears to have stopped. Record the frequency and time. When the spinner has slowed, reduce the frequency of the stroboscope and repeat the measurement at several intervals, again noting frequency and time. It does not matter if you start from different spin speeds in the two cases.
Plot frequency against logarithm of time for the two orientations of the spinner. You should find a straight line in both cases, but with a different slope. The steeper line indicates a faster rate of slowing down. 
A: If you spin your fidget spinner horizontally, the weight will be symmetrically distributed around the center of mass and the rotation axis. It should therefore spin for a longer period of time in comparison to doing it vertically.
But if you're looking for the perfect conditions to spin your fidget spinner forever, those would be spinning it:


*

*in free-fall or in space : gravity causes friction in the rotation axis;

*in vacuum : there is also the friction between the spinning arms and the air.


In those conditions you could accelerate the rotation up to a really high speed and your fidget spinner would spin forever. :)
A: I hope these theoretical factors answer your question:
Since the spinner is only a device that transfers energy from kinetic to inertia back to other forms, observing particular issues may help ableit negligbly:
The spinner is based upon an alloy ball race to minimise friction. It is usually a sealed type bearing. All motion is subject to the laws of thermodynamics and therefore working with the laws instead of resisting them will have an impact but probably not noticeable to human perception. Expecting an extra second of spin time is asking a bit much... but here goes anyway:
Mechanically, the race supports the bearings until they are set in motion in which case they are forced to behave counter to natural behaviour by rotatational motion instead of towards the Earth's centroid of gravity. This means that friction is developed between the internal metal components and this causes an electromagnetic field to develop molecularly. Since the energy input is greter than the electrical resistivity, motion ensues.
As the bearings race, their centrifugal force tends to push them outwards from the race which increases the disatance which they have to travel. A lower machine tolerance and harder material for a ball race would therefore icrease te tendency for the bearing to spin longer which is why the grade of steel isvery important.
Inertia in the presence of a massively larger body will induce a counter force so directional rotation is important too. Orienting the device so that it rotates vertically is the first option, the next option is to orient it laterally so that it accurately aligns with gravitational pull.
If you really want the spinner to spin longer, you need to reduce the friction even more in the ball race. Understanding that spinners are essentially bearings, adding a second bearing inside the first will improve the friction reduction marginally but, this will have an impact on actual spin time since both bearings will absorb the energy input proportionally.
Thermal conductivity makes a big difference in that it reduces the radius of the ball races and reduces the tolerance between bearing balls and the race, lifting performance time substantially.
Lubricating the ball race with superfine graphite will also contribute to longer running times and this is often available as pencil lead but not of the polymer type.
Increasing the density of the bearing balls while reducing their weight would also improve spin time. This obviously is unfeasible.
Indications are that rotating anticlockwise on an east-west linear orientation is probably going to be faster... except you will probably not be able to measure it with your estimation even with all the above advice.
A: I'd like to make a couple of observations related to the orientation of the fidget spinner, which could potentially affect the duration of spin.
First, it seems that ball bearings used in most fidget spinners have symmetric grooves in the inner and outer races, optimized for radial loads. We can argue that, for such bearings, the normal pressure on the balls from the races will be higher in the horizontal orientation than in the vertical orientation, since in both cases the same (vertical) load (=weight) has to be supported, but in the horizontal orientation, only a fraction of the normal force will be vertical.

So, if we assume that the rolling friction is proportional to the normal force, we'll have to conclude that, given the same load per ball, the rolling friction in the horizontal orientation will be greater than the rolling friction in the vertical orientation.
Of course, the load per ball would not be the same between the two orientations, since, in the vertical orientation, the load is not evenly distributed: the balls on the top will carry all the load and, therefore, will contribute most of the friction losses.
So, the next question is whether fewer balls carrying the same total load will produce more or less of the total friction losses. According to some sources (e.g., [this Wikipedia article][2]), the friction or resistance coefficient for the rolling friction, at least for some metal pairs, decreases as the load increases. If we assume that this is true for typical ball bearings used in fidget spinners, we'll have to conclude that the friction losses in the vertical orientation should be lower than in the horizontal orientation.
Of course, there are a number of other factors that could affect the duration of spin. For instance, we can suspect that, in vertical orientation, some energy could be lost as the balls are moving up and take up more load or some energy could be lost due to slight vibrations caused by the load asymmetry. But, those factors depend on the quality of the ball bearings and, therefore, are more difficult to quantify.
In summary, it appears that, in the vertical orientation, a fidget spinner should have smaller losses and therefore should spin longer.          
