Historical walk-forward backtest of the Framler top-decile sleeve. At each rebalance, the universe is sorted by composite score, the top 10% are bought equal-weighted, held for the forward window, then re-ranked. Compared against buy-and-hold SPY as a baseline. Survivorship- corrected (delisted tickers retain their final-day return). This is what Stockopedia and Zacks publish — same methodology, our universe and engine.
Walk-forward backtest. At each rebalance date, the engine scores the surviving universe using the live PRIOR_WEIGHTS factor weights (no future leakage), sorts descending, takes the top decile equal- weighted, and holds for the forward window. Forward return for each ticker comes from cached daily closes (apex_historical_prices). Sleeve return = arithmetic mean across the decile. SPY return for the same window is the baseline. Cumulative compounding starting from $1.
Refresh: weekly via backtest-weekly.yml GH Actions cron (Sun 04:00 UTC). Results persist to apex_backtest_runs.
← Track record overview · Engine methodology → · 9 live invariants →