TWO TIERS. NO TRICKS.
Most weeks, the right
trade is none.
Most tools need you excited: their fee depends on it. We tested the thing they sell you and it came out a coin flip, so that finding is free, with every case behind it. We don’t sell you an edge · we sell the data that sharpens the one you already have. Where the crowd sits today instead of in Tuesday’s stale report, what each market did the last times it looked like this, and how far it ran against you. One badly sized futures trade costs more than a decade of it.
€0
Forever · no card. Everything you need to judge the tool.
- The full ledger, all 44 markets, every week
- The top file of the week, complete
- The Friday Letter in your inbox
- The free sample backtest file
- The daily nowcast on the ledger, all 44 markets
- Methodology, baselines & corrections, always open
All of the above stays free. The graded odds, the backtests and every trade file are Pro.
Start free€49
/monthbilled monthly · cancel anytime
Free opens one file a week. Pro opens the file on every trade you’re about to take, all 44 markets, updated daily.
- The full confirmation layer, six checks on every market: conditions grade, what marked past turns and whether it is here yet, the stop that survived, how long the wait ran
- Which of the 44 markets are even worth your week: usually 3 to 5 qualify · the rest we tell you to ignore
- How wide the stop has to be to survive this market: its own median shakeout: how far past turns ran against the trade before they worked
- Never trade a 3-day-old report: the daily nowcast read inside every file
- Size from the worst case on file, not the average: the 13-week worst drawdown for this market, with the number of cases it rests on
- Every past trade numbered on the chart: the full backtest file per market, winners and losers
- Daily tape alert on the markets you star: plus the Friday watchlist e-mail
30 days. Not useful? Write to us and we refund it.
cancel anytime · unlocks on your account in seconds · incl. VAT where applicable
use the same e-mail at checkout as for sign-in. Pro unlocks on it automatically
OPEN NOW · NO CARD, NO SIGN‑UP
One complete file is open: Gold.
31 crowded extremes on record since 2006, 15 of them reversed, the other 16 still on the page. It is the deepest file we have, which is the point: judge us on the longest record, not on whichever market happens to be stretched today. That one is free as well.
WHAT IT ACTUALLY LOOKS LIKE · SCREENSHOTS, NOT MOCKUPS
Every picture here is a page you can open right now.
The Gold file is free and complete, losses included. Pro is this for all 44 markets, every Friday, with the week’s verdict on top.
Every extreme, numbered on the tape →31 crowded extremes in Gold since 2006, each marked where it happened. Tap an arrow and its trade file opens.
The table that keeps the losers →Entry, direction, what the crowd did over 4, 8, 13 and 26 weeks, whether it turned, and the peak move in hindsight.
Our own calls, running in public →Entered blind at the extreme, timestamped, graded at fixed check weeks. Winners and losers in one list.
And the page for what failed →Larry Williams’ commercials rule, one of the most repeated in this corner of the market, graded on 1,657 cases. It did not survive.OR BUY THE RECORD ONCE · NO SUBSCRIPTION
The research pack · €149 one-time
Every graded positioning extreme since 2006 as four CSVs: the outcome of each at fixed horizons, per-market turn timing, the fingerprint factors, and one row per market naming the price instrument its record was measured on, with a README stating the caveats. Single-buyer research licence, refreshed weekly. For people who want the data, not the weekly read.
WHY THE CONFIRMATION LAYER EXISTS
The naked COT read everyone else sells you is a coin flip. We proved it · so we built the layer that was missing.
And you can check how we grade, because we publish the tests that failed, including our own ideas. Each test was written down in advance. None gave an edge at entry. All stay on the record.
Mechanized, our own protocol failed its registered bar too: gated entries still averaged −0.04R in the years the rules had never seen. One thing did repeat with the same sign in both periods, and it is narrower than an edge. Extremes that matched the turn fingerprint came out +0.50R (training) and +0.22R (confirmation) better than the extremes that failed it, and the gap survives a within-market control. That confirmation gap is not statistically significant. So the fingerprint is a priority map: it says which extremes are worth reading first, never that a trade is worth taking. That map is what you pay for, and it is why we refuse to sell it as a “buy now” signal.The one pattern that did repeat, the turn fingerprint, is not statistically significant out of sample. It tells you which extremes to read first. It never tells you a trade is worth taking, and that is why we do not sell a signal.
1,206
extremes backtested, the numbers you’re buying are public-audited
1 free file
every week’s top file stays free, judge the product before paying
0 gurus
no coaching upsell, no Discord hype, an instrument, not an influencer
Don’t trust us, check the receipts first.
Everything Pro sells has a free surface you can inspect before paying a cent. If any of these disappoint you, don’t buy.
DATA & API · NOT A SUBSCRIPTION
Rather have the record as data?
The two tiers above cover the site. If you run your own research or trade systematically, the same graded record ships as files you can load directly: every extreme since 2006 with what happened next attached.
RESEARCH PACK · CSV · €149 ONE-TIME
Four clean CSVs: every extreme since 2006 with outcomes at every gate, per-market turn timing, the fingerprint factors, and the price instrument behind each market · plus a README that states the caveats the way the site does. Single-buyer research license, current release, refreshed weekly.
checkout via stripe · incl. VAT where applicable · the download starts right after payment
THE API · EARLY · CUSTOM
The ledger (/api/index.json), all 44 markets with state and percentile, is open and free. The per-market payloads · graded episodes, turn fingerprint, survival curves, the daily nowcast with its own hit rate · are keyed, and bulk access is the paid product. Schema and availability can still change while it is young.
early users shape the schema · no self-serve key yet
Full payload shapes, the free-ledger endpoint and the honest terms: Data & API →
Before you ask
Can I cancel anytime?+
Yes, cancel renewal anytime in one click via the Stripe receipt e-mail. You keep full access until the end of the paid period. And within the first 30 days, if Pro is not useful to you, write to support@cotreport.ai and we refund that period in full · no reason needed. After those 30 days a started period is not refunded (§18 FAGG: you agree to immediate access at checkout, waiving the statutory 14-day withdrawal right; our 30-day promise is voluntary and sits on top of your legal rights).
When does Pro unlock after I pay?+
Use the same e-mail at checkout as for sign-in. Pro unlocks on your account automatically within seconds of payment. If anything hangs, write to support@cotreport.ai and it gets fixed manually.
What stays free forever?+
The full 44-market ledger, this week’s top file, one complete backtest file, the Friday Letter, nowcast direction on every market, and the entire methodology with every baseline and correction.
Is this trading advice?+
No. Everything here is statistics about the past, for information and education. Most comparable setups did not turn, we print that on every file. Every decision and its full risk are yours.
Do I get an invoice?+
Yes. Stripe e-mails a receipt/invoice for every payment, with VAT itemized where it applies (EU); non-EU customers pay no EU VAT. Business details that need adjusting: support@cotreport.ai.
payments for a started period are non-refundable, you keep access to the period’s end and can cancel renewal anytime · at checkout you expressly agree to immediate access, waiving the 14-day withdrawal right (§18 FAGG)
cancel anytime · prices incl. VAT where applicable · no signals are recommendations, see the disclaimer in the footer · how the numbers are made