Every Monday, the system produces a market brief: what could move this week, which way it currently leans on each stock, and why. I publish the claims before the market has answered them, then score them when their time windows close.
The full record stays public. I do not delete misses, rewrite old calls, count conditions that never happened, or take credit for a stock simply rising with the rest of the market.
The file is a compact update on what the system is seeing, what it has already committed to, and how its older calls turned out.
Earnings, policy decisions, reports and other events that could matter this week, with the possible outcomes written down before they happen.
A simple long, short, split or range-bound read for one week, one month, three months and six months.
Each claim has a ticker, direction, time horizon, stated condition and serial number. Once it is published, the wording stays put.
I was tired of market commentary that could never really be checked. The good calls were remembered. The bad ones disappeared. Predictions stayed vague enough to be declared right later.
So I built a system that leaves receipts. It makes its calls in advance, keeps the original wording, and scores the result against rules that were set before the record existed.
I do not know yet whether the system has a real edge. That is the point of publishing the experiment. If it turns out to be useful, the record will show it. If it turns out to be average, the record will show that too.
Every claim is timestamped before its result window begins. No retroactive entries and no rewriting after the fact.
A stock going up during a broad rally is not automatically a good call. Directional claims are judged on market-adjusted returns.
A conditional claim only counts when its stated condition happens. A branch that never fires is marked void, not quietly counted as a win.
The ledger is the whole record: hits, misses, partials, voids and pending claims. Nothing resolved gets removed.
Accuracy is reported with confidence intervals and stated probabilities. A neat-looking percentage without context is not enough.
Read the call. Check when it was published. Come back when the window closes. The record does the talking.
One email each Monday with:
No daily feed and no urgency tricks.
Read it, inspect the evidence, and decide whether it is useful.