SHORT ANSWER
No bet-sizing tactic can change the house edge on a certified RNG slot. The RTP is fixed per spin regardless of bet size. What bet sizing can do is control session duration, bonus exposure, and the absolute scale of wins and losses. The only defensible "strategy" is choosing a bet size that keeps you in play long enough for the variance to work in your favour.

The myth: bet size affects RNG outcomes

A widespread belief holds that increasing your bet during a losing streak — or decreasing it — can influence the RNG. It cannot. The RNG generates outcomes independently of bet size. The per-spin probability of triggering a bonus, landing a Wild, or hitting any specific symbol combination is identical at 0.10/spin and 100/spin.

What does change with bet size: the absolute value of wins and losses, the cost of buy feature entries, and how quickly your bankroll depletes. None of these affect the underlying probability distribution.

Tactics that are mathematically defensible

Three bet-sizing approaches have legitimate rationale, not because they change the RTP, but because they manage bankroll mechanics sensibly:

Flat betting at a sustainable stake. Choose one stake you can play 200+ spins with and stay there. This gives the RTP the most spins to normalise and gives the bonus the most opportunities to trigger. It is unglamorous but it is the most statistically sound approach.

Stepping down after significant losses. If your bankroll drops below 50x stake, reducing your bet to rebuild spin count is defensible. You are extending session duration to give yourself more chances at a natural trigger. This is the opposite of the martingale — it conserves rather than accelerates.

Stepping up from a position of strength. If a good bonus win has put you significantly ahead, raising your stake uses profit rather than original bankroll. You are not chasing losses — you are parlaying a win. This is the only psychologically clean version of bet-raising mid-session.

TACTICMATHEMATICAL EFFECTVERDICT
Flat bettingMaximises spin count per bankrollRecommended
Step down after lossExtends session, same RTP per spinDefensible
Step up after win (profit)Increases absolute win potential from profitAcceptable
Martingale (double on loss)Accelerates bankroll depletionHarmful
Increase bet after droughtRaises absolute loss rate, no RTP benefitHarmful

The "high-variance streak" framing is the problem

The question of what to do "during a high-variance streak" assumes variance streaks have some predictive quality — that knowing you are in one tells you something about future spins. It does not. A losing streak is only visible in hindsight. During the streak, you have no way to know whether the next spin continues it or ends it. Bet-sizing tactics that respond to perceived streak momentum are responding to a statistical illusion.

The only meaningful pre-session decision: Choose your stake based on your total session bankroll. Use a 100:1 ratio as the baseline — that is, your session budget divided by 100 gives you your maximum stake per spin. Set that number before you start. Change it only if your bankroll fundamentally changes during the session (significant win or loss), not in response to short-term spin sequences.

BOTTOM LINE
No bet-sizing tactic maximises ROI in Fist of Destruction because ROI is determined by RTP, which is fixed. What bet sizing does is determine session duration and absolute win/loss scale. Flat betting at 100:1 of your bankroll is the most defensible approach. Raising bets during a losing streak is the most harmful. Everything else falls somewhere between those two poles — closer to the harmful end the more it responds to recent spin history.