A weighted coin flipping in the air is neither heads nor tails until observed. That we can't observe the property of the coin (its weighted bias) while it's in the air flipping, doesn't change the coin's properties. In fact, we wouldn't be able to measure the bias from a single observation. Yet, if I constructed the coin to be precisely heavier on heads versus tails according to some ratio, it would have that property the whole time, no?
To determine if the coin is biased you would need to repeat the experiment large number of times and see if you get experimental outcomes that do not agree with the expected values given by law of large numbers.
This is not how things are in quantum mechanics though. Uncertainty principle is an intrinsic property of quantum systems. It's not the case that the act of measuring somehow disturbs the system and changes the measured values. It's that values are undetermined at all until they are measured. And even then there is inverse relationship between how precisely you can know momentum of the particle vs its position. The more you know one the less you can know the other.
Though realism may be a tempting viewpoint what we know about modern physics makes it a difficult one to defend. For example, according to the principles of quantum physics a particle has neither a definite position nor a definite velocity unless and until those quantities are measured by an observer. It is therefore not correct to say that a measurement gives a certain result because the quantity being measured had that value at the time of the measurement. In fact, in some cases individual objects don’t even have an independent existence but rather exist only as part of an ensemble of many.