Thomas Markov’s answer does a good job of showing the probabilities, but doesn’t really explain the ‘why’.

To actually understand this, you have to understand how the normal advantage/disadvantage rules work.

Rolling 1d20 has an equal chance of each value from 1-20 coming up. Rolling 2d20 produces a triangular probability curve with a peak at 21. Rolling 2d20 and then ignoring one die also produces a triangular probability curve, but with the peak at 20 (for advantage) or 1 (for disadvantage).

With your system though, things get a bit strange. Equal amounts of advantage and disadvantage actually produce not a flat uniform distribution like is the case with the standard rules, but instead a classic Gaussian distribution (bell curve) peaking at about average values. This is because in your system, advantage is more likely to eliminate low rolls than high ones, and disadvantage is symmetrically more likely to eliminate high rolls than low ones, so most of the time what you have left will be about average.

Where it gets really complicated though, is when you have unequal numbers. You still get a Gaussian distribution, but it’s an asymmetrical one with a different positive and negative standard deviation and the average is not right at the halfway point.

This, in turn, leads to two specific mathematical aspects to your approach that are non-obvious to most people without a background in statistics:

### Both advantage and disadvantage have diminishing returns.

If you have one source of disadvantage, then adding one source of advantage shifts your most likely roll from 1 to 10-11, effectively adding 50% to your chance of success. Adding a second source on top of that only shifts it to 14, which is only effectively about 20% extra. Adding a third then brings you to roughly 15-16, only a 5-10% better chance. The same works in reverse as well, and this feeds into the next point.

### Any source of disadvantage makes is exponentially less likely to roll a 20 overall, and any source of advantage makes it similarly less likely to roll a 1.

If you have six sources of advantage and none of disadvantage, you have a whopping 30.17% chance of rolling a 20. Adding _just one source of disadvantage_ drops that all the way to 5.72%, roughly one fifth of what it would be without that one source of disadvantage. This is _huge_ for combat, because a natural 20 is a guaranteed hit, and a natural 1 is a guaranteed miss.

----

Overall, whether this is an issue for you or not is up to you. It reduces overall randomness within the game though (you are more likely to see average rolls more frequently), which is generally a factor in making the game shift more towards abstract strategy than realistic tactics, but it does not reduce it _that much_.

That by itself would not be enough for me to dismiss such a rule outright, but the overall complexity it introduces (touched on in other answers) and the impact it would have on game pacing.