A two-portfolio, RSI-guided momentum system that alternates between long and short levered ETFs. Each “machine” chooses a single asset from a pool based on 4-day momentum, with direction determined by RSI comparisons between asset groups. Weights (80% and 30% in the two machines) and a drawdown-based selector pick the final lineup, rebalancing about every 15 days. It targets amplified moves across tech, biotech, semiconductors, China/EM, and bonds, using 3x leverage and inverse/short bets.

Two main sub-strategies (machines) run in parallel and each has two possible directions on any rebalance: - Direction decision: compare momentum of two asset groups using relative strength index (RSI). If RSI of the left asset group is greater than RSI of the right group (specific pair varies by machine), the machine goes long a pool of bullish bets; otherwise it goes short a pool of bearish bets. - Asset pool and selection: each machine maintains a pool of candidate ETFs. From that pool, the system ranks assets by their recent momentum (moving-average return over a short window, here 4 days) and selects the single asset to carry for the next period (the pool is cash-equal-weighted internally before selection). The selected asset is then fully funded for that machine’s slice of capital (i.e., the machine allocates 100% to that one instrument for the next window). The long pools are loaded with leveraged long bets (e.g., TECL, URTY, TMF, YINN, EDC, LABU, HIBL, TARK, SOXX, etc.). The short pools use inverse or bear bets (e.g., PSQ, SARK, TMV, DRV, LABD, SH, TYO, JDST, etc.). - Risk-based selection between machines: a final filter looks at a risk metric (maximum drawdown) over a lookback window, and the best machine (top 1) is selected to determine overall allocation for the rebalance. The two machines have fixed weightings (e.g., 80/100 and 30/100, depending on variant). - Rebalance cadence: roughly every 15 days, with the direction and asset for each machine updated according to the rules above. This produces a dynamic, momentum-driven tilt between a broadly bullish, high-beta/levered set of bets and a broadly bearish, inverse/short set of bets, guided by RSI comparisons and short-horizon momentum screens.

High upside potential: out-of-sample annualized return ~43% vs S&P ~24%, driven by RSI-guided momentum in leveraged bets. 15-day rebalances with a drawdown filter seek alpha beyond the S&P, though higher drawdowns.