To answer this, you need to see that there exists “randomness” that is not due to the butterfly effect, and then how that phenomenon will be affected by time travel.
chaos and determinism
Just because the random process is “complex” doesn’t mean it becomes nondeterministic. The proper phenomenon here is Chaos, and this is what leads to the so-called “butterfly effect”. If your arrival caused some air molecules to bounce differently than they had in the original timeline, and this eventually made the lottery choose different numbers, that is an example of the butterfly effect and thus not what you are wondering about.
So first, you are asking whether there even exist nondeterministic effects. In pure Newtonian physics, every action can be computed from the previous history, and everything plays out like clockwork. A deterministic system will always produce the same output from a given starting condition or initial state.
nondeterminism, or genuine quantum randomness
As it turns out, quantum mechanics does indeed have genuine randomness. The observable measurement will be chosen at random with a weight based on how close the quantum state is to each possible eigenvector. To give a concrete example without getting too deep into the quantum mechanics, consider what happens when a single photon at some random polarization angle reaches a polarizing filter that passes vertical and rejects horizontal. What if the actual light has an angle that’s not exactly vertical or horizontal, but 15° off vertical? Well, this particular photon will have a 6.1% probability of being horizontal and 93.9% chance of being vertical.
There is no cause due to prior state. There is no reason, and no way to know the result ahead of time. This is genuinely random and nondeterministic, and this is not just a limit to our knowledge. It can be logically shown that the result literally comes from nowhere. Real-world experiments back this up.
So, ultimately the bouncing of the balls in a classic random lottery machine will indeed bounce in unpredictable ways, being truly nondeterministic and not just chaotic. A modern lottery machine will probably combine traditional bounding balls with a micro-controller that uses a quality random event generator to control the precise timing of the paddles and other movements. This will, if implemented as intended, put the ball draw definitely in the genuinely random category.
Now will quantum randomness play out the same way in a new timeline? That’s the question.
branching timelines keep happening
Note that this behavior itself may be the source of having multiple timelines! The Many Worlds Interpretation shows how the apparent choice of a specific value (e.g. horizontal or vertical polarization in the example above) is due to a superposition of everything including your brain’s particles with each possible outcome.
So if this is happening, there is no “choice” and you will split into two new timelines, one of which sees H and one sees V. The “you” asking this will only experience one of them, and that is randomly chosen. So it is “no different than before” in the sense that the particle went both ways in the original timeline, and it goes both ways in the new timeline. But which resulting daughter timeline from this new split you ride into may very well be different.
So from your point of view, the random events will be freshly chosen in the new timeline, and will not necessarily follow the same as in the original timeline.
Experiments over the last few decades, including practical engineering efforts into building a quantum computer, have shown that decoherence is correct, and we have things like delayed choice experiments and even quantum erasers, which suggest that the separate timelines caused by every random wavefunction decay are real.
is that the answer you wanted?
But maybe there’s more to it than that. Maybe the possibilities encoded in the entanglement of all the particles in the environment somehow get pruned back and there is a single “real” timeline, and this path is persistent so the timeline will try and recover the same path when it’s replayed. This is used to great effect in many science fiction stories where the timeline has an inertia of sorts and history is difficult to change.
Consider another concept, that of the quantum handshake. This has to do with the source and future destination of an interaction that form a transaction between the two points in time. You can read about this in John G. Cramer’s Alternate View column in Analog magazine, where I first learned of it.
In some science fiction accounts, going back in time will nullify the transactions and allow them to reform in newly-chosen ways. Under this model, quantum randomness will also be different.
But what if they are not all reset? If the other end of the transaction occurs later than when the traveler departed, maybe that end will hang onto its transaction, and the side of the transaction that the traveler experiences a second time in the new timeline will be the same. This idea more readily supports an overwritten timeline though, but you can make it branching by saying that the state is copied of the entire timeline, past and future, not just the point before the time at which the traveler arrives.
recap — quantum randomnes resets
To summarize, the easy answer that works hand-in-hand with real physics as we understand it, given the branching timeline paradigm, is that quantum randomness will be reset and will play out independently in the new timeline.
To do otherwise will take some contrivances, both in describing new physics for the timelines and in preventing the “many worlds” from making it appear that they have reset. And why do we need multiple mechanisms producing timelines that work in different ways and fight each other?
Furthermore, once anything changes, such as the specific collision of two atoms in the lotto balls, then the specific quantum events that even exist will be different in the new timeline. The butterfly effect (due to your presence, or any partial tendency to play out the same) will pull the rug out and you won’t have the same quantum events to replay, from that point forward. So making things work the same way on the new timeline really would require a higher order fate to make things restore to the same large scale state, even with the low-level details (entropy) completely randomized.