Public Self-Scorecard
The app grades its own calls — winners, losers, and WAIT-resolutions, never hidden.
How results are counted just got more accurate
Previously, if price reached the target, the analysis was counted as "correct" — even when price never reached the suggested entry. Since July 18, 2026, only analyses whose entry price actually traded are counted. This is why the record number is lower. Past analyses have not changed — only the counting is now more honest.
This is analysis, not financial advice — the final decision is yours.
For a buy or sell call, “correct” means price reaches the first target before the stop-loss — the price that breaks our plan. Every call is graded against the levels it was published with — moving them later never re-grades it. If it reaches neither inside the window, the call still counts, just not as correct.
Resolved so far: 0 of the 10 calls needed. We show the accuracy figure once the count reaches 10.
A WAIT call is scored only on whether the price reached the level it named within its stated horizon — never on what would have happened after.
We give analysis, not buy or sell orders.
The decision is always yours.
Last computed: just now
This is not a percentage and it does not replace the accuracy figure above; it measures how far price moved past each call's own published entry, target, and stop — not what someone who followed the call, including any exit, would have made.
This count is not the same as the resolved-calls count above and must not be read against it: episodes that reached neither target nor stop — timed out, still open, or never entered — are left out here, because their realized return cannot be read off the original plan alone.
When the market gapped through the stop, a stop-out's R is booked at the breaching bar's open — the fill a holder could actually have got, which can be worse than -1R. Where no bar is on record it falls back to the analysis's own stop price.
The unit is R: one R is the distance from a call's entry to its stop — the amount that call puts at risk. A clean stop-out reads as −1R, and a negative number here is entirely normal.
This section is about the app's own behaviour: how often it steps away from a call, how long a call lives, and whether stepping away helped.
None of the numbers here is a score, and a high one is not good news; do not read them alongside the percentage at the top of the page.
This scorecard updates weekly, and no call is ever removed — it is always right here.
Forecast record
Each forecast is recorded with its record code before the deadline, and its numbers are published only after it is judged. Each question counts once, however many forecasts were made for it, and misses are shown at the same size as hits.
What will the US central bank decide on its policy rate on Wed 28 Oct?
Open Deadline:
What happened: Not yet — the deadline has not passed
Its numbers are published once it is judged; until then only its record code is shown.
Forecasts recorded for this question: 1
Record code of the latest forecast: b1abb459cfc7fdb13262b0fa509d2c7b4ed4b91efdca12ec32b4eb121366f58f
Will the US 10-year real yield on Tue 20 Oct be higher than on Tue 13 Oct?
Open Deadline:
What happened: Not yet — the deadline has not passed
Its numbers are published once it is judged; until then only its record code is shown.
Forecasts recorded for this question: 1
Record code of the latest forecast: 995f697e8f5569bb4a998ccb5761f70e3f8ae9a17927d6ac5b904dc3b7c011d5
Download the raw record (JSON)
The raw file also carries the numbers of judged forecasts.
Latest link in the chain: b1abb459cfc7fdb13262b0fa509d2c7b4ed4b91efdca12ec32b4eb121366f58f