Max-win caps,
ranked.

Max-win cap is the ceiling. The probability of hitting it on any given spin is, by definition, vanishingly small — typically one in tens of millions of spins for the upper-tier slots. The math:

Mental (NoLimit) — 100,000× max-win cap
  Hit probability per spin    ≈ 1 in 25 million (NoLimit's published math
                                modelled in their game-info panel)
  Spins per hour, $1 bet       ≈ 800-1,000 base game; ~30-40 with
                                bonus-buys at 100× cost
  Cost to "guarantee" one hit  ≈ $25-100M+ in turnover

Sweet Bonanza (Pragmatic) — 21,100× max-win cap
  Hit probability per spin    ≈ 1 in 8 million (estimated from
                                Pragmatic's variance model)
  Cost to "guarantee" one hit  ≈ $8-15M in turnover

Gates of Olympus (Pragmatic) — 5,000× max-win cap
  Hit probability per spin    ≈ 1 in 100,000 (the most accessible
                                upper-tier hit in this lineup)

That's the long-run reality. The streamer hits you see clipped are the variance tail — the right end of a probability distribution that has a much fatter left side (hundreds of bonus-rounds that pay 5–50× on a 100× buy). The clip is real. The path to it is expensive.

Where the math actually matters for the typical player: the cap is a useful proxy for the slot's upside variance. A 5,000× slot can't print a 50,000× hit no matter how lucky you are. A 100,000× slot can — at the cost of a higher base-game grind to get there. The cap encodes "volatility you can't see" into a single comparable number.