⚠ SUSPENDED BY OUR OWN REVIEW
We took our own headline backtest down.
This page used to show a walk-forward “top-decile sleeve” equity curve with a large cumulative return and an alpha figure against SPY. A source-level review of the reconstruction — prompted by an external audit — found it failed our own honesty bar in three separate ways. We publish our record so it can be checked; that has to include the part where the check fails.
1. The SPY benchmark was never loaded
Every stored weekly run carried spy_available: false — the benchmark price series was missing from the historical table since the page launched, so the baseline rendered as a flat 0% and every “vs SPY” figure, including the headline alpha, was computed against nothing. The engine flagged it; the page ignored the flag.
2. The methodology text overstated what was ranked
The page said the sleeve was ranked by the live 13-factor engine weights. The reconstruction actually ranked by a single momentum proxy, because most of our factors cannot be reconstructed point-in-time before our live history began. A momentum-decile backtest is not the Framler engine and should never have been labeled as one.
3. The survivorship claim was not implemented
The page said delisted tickers retain their final-day return. In the code, a ticker whose price series ended mid-window was silently dropped from the sample instead — the stated correction did not exist.
What has to be true before the sleeve returns: a real, verified SPY total-return baseline in the data; labeling that states exactly what is ranked (a momentum proxy is not the engine); and delisting handling that matches whatever the page claims. Until then, the honest evidence is the live record, which needs no reconstruction:
the pre-registered bet,
the graded track record and
the Bitcoin-anchored ledger.
The withdrawn figures are retained internally and in this page's public edit history — withdrawn, not erased. No other page cited them.