Scoreboard
AI news forecasts: the full record
Before each session an AI model reads the news and forecasts whether EUR/USD, EUR/CHF and USD/CHF will go up or down. Every forecast is listed here once its window has closed, hits and misses alike, with the reasoning behind it and what actually moved the price. The method has been revised several times; the numbers below combine all versions, and each version is also scored on its own. Background: Can AI predict EUR/USD from the news? 120 forecasts later
Hit rate as forecasts add up
shaded = 95% range · dashed = coin flipEarly on a few forecasts swing the rate wildly; the range narrows as more come in. A range that still includes 50% means the result can't be told apart from guessing.
By pair
bar = hit rate · whiskers = 95% range · dashed = coin flip
By session
EU 08–12 UTC · US 12–21 UTC · Day = 08–21 UTC, used only 5–8 Oct
Misses: how sure the AI is of the reason
after each window the AI explains the move and rates how sure it is of that explanation
- Clear reason 7
- Plausible reason 35
- Unclear 43
Day by day
one cell per forecast · EU first, then US · hover a cell for details hit missVersions and weekly audits
now on v6The forecasting method has a version number, shown on every call below. Every Saturday an AI agent audits the past week: it looks at what went wrong, tests new ideas on years of price history, and has a separate skeptic agent try to break each result. A new version ships only when an idea passes both; most weeks nothing does. Versions are scored separately and never mixed, so a change can't hide in an average. We audit the method; we don't promise it gets better. How the weekly audit works.
| Version | In use | Calls | Right | What changed |
|---|---|---|---|---|
| v6 | 2 Nov – 16 Nov | 63 | 31 (49%) |
Tested 4 ideas; one passed: on US federal holidays the US session is much quieter, so its expected size is scaled down. Direction calls are unchanged. Rejected: a direction-aware edge flag, prices reversing after central-bank decisions, and high currency volatility predicting continuation. |
| v5 | 26 Oct – 30 Oct | 30 | 16 (53%) too few calls to compare |
The first weekly audit, together with six studies run the same week. Version 5 adds an expected size of each session, moves in bonds, stocks, oil and gold as context, and a lower confidence cap on days the Fed chair speaks. One tested idea became a rule: statistical edge flags no longer tilt a call, because the skeptic showed they only echoed each pair's long-run drift. Three ideas were rejected. |
| v4 | 15 Oct – 23 Oct | 42 | 17 (40%) too few calls to compare |
The forecaster reads the day's economic calendar and has to search properly before calling. On two days it had made every call after one stale search, including on a US inflation release day. |
| v3 | 12 Oct – 14 Oct | 18 | 14 (78%) too few calls to compare |
A driver such as bond yields can back a call only through how it moved since the previous close, never through its level alone. |
| v2 | 8 Oct – 9 Oct | 9 | 6 (67%) too few calls to compare |
Each call also gets statistics from 15 years of hourly prices, such as how often the pair rose in this window after a similar move. |
| v1 | 5 Oct – 7 Oct | 9 | 2 (22%) too few calls to compare |
The first version: before each session the AI reads the morning news and calls each pair up or down, with no price statistics. |
Below about 50 calls a version's hit rate is mostly luck: the 95% range is wider than ±14 points, so two versions ten points apart can't be told apart. The numbers are shown so nothing is hidden, not as proof that one version is better.
Change log
| Date | What | Outcome | Details |
|---|---|---|---|
| 15 Nov 2026 | Weekly audit 2026-W46 3 ideas tested · 0 passed |
no new version | Tested 3 ideas; none passed, so the method stays at version 6. Rejected: the reaction to a central-bank statement reversing later in the session, bond-market uncertainty making decision days bigger, and the first hour after a Swiss rate decision setting the afternoon direction. |
| 7 Nov 2026 | Weekly audit 2026-W45 4 ideas tested · 0 passed |
no new version | Tested 4 ideas; none passed, so the method stays at version 6. Rejected: a run of US data surprises setting the dollar's direction, a late move before the European open fading, and Swiss bank deposit data predicting the franc. A volatility idea looked promising but is now checked on new data before it can change any call. |
| 31 Oct 2026 | Weekly audit 2026-W44 4 ideas tested · 1 passed |
v6 shipped | Tested 4 ideas; one passed: on US federal holidays the US session is much quieter, so its expected size is scaled down. Direction calls are unchanged. Rejected: a direction-aware edge flag, prices reversing after central-bank decisions, and high currency volatility predicting continuation. |
| 26 Oct 2026 | Lessons v2 approved by a person |
lessons v2 on | The first lesson was reworded so it no longer depends on statistical edge flags, which stopped tilting calls in version 5. |
| 24 Oct 2026 | Weekly audit 2026-W43 4 ideas tested · 1 passed |
v5 shipped | The first weekly audit, together with six studies run the same week. Version 5 adds an expected size of each session, moves in bonds, stocks, oil and gold as context, and a lower confidence cap on days the Fed chair speaks. One tested idea became a rule: statistical edge flags no longer tilt a call, because the skeptic showed they only echoed each pair's long-run drift. Three ideas were rejected. |
| 22 Oct 2026 | Lessons v1 approved by a person |
lessons v1 on | First lesson switched on: when there is no dated news for the window, the call stays at 50% and says it is a coin flip. |
| 15 Oct 2026 | Method v4 by hand, before the weekly audits |
v4 started | The forecaster reads the day's economic calendar and has to search properly before calling. On two days it had made every call after one stale search, including on a US inflation release day. |
| 12 Oct 2026 | Method v3 by hand, before the weekly audits |
v3 started | A driver such as bond yields can back a call only through how it moved since the previous close, never through its level alone. |
| 8 Oct 2026 | Method v2 by hand, before the weekly audits |
v2 started | Each call also gets statistics from 15 years of hourly prices, such as how often the pair rose in this window after a similar move. |
| 5 Oct 2026 | Method v1 by hand, before the weekly audits |
v1 started | The first version: before each session the AI reads the morning news and calls each pair up or down, with no price statistics. |
Locked in — window not closed yet
6 forecastsThese forecasts are already made, but you can't see them yet - only each one's hash: a 64-character code computed from the forecast and a random secret. Change one letter of the forecast and the hash comes out completely different, so we can't swap a forecast after the price moves. Copy a hash now and compare it after the result. How to check.
a2ed6ee54f7670f3615b80a6dc2b184a7c53b02933834c69848da16f7a61144a
.ots proof
sealed 11:17 UTC · OpenTimestamps: submitted 17 Nov 11:17 UTCdc71df323eb6e5ebb77c81841fe647c610475a643bfc912542358751f748efa1
.ots proof
sealed 11:17 UTC · OpenTimestamps: submitted 17 Nov 11:17 UTCc7e2e94762ef5591274dfcaa776ce5ac9616ae7c99823593d137aab821d21338
.ots proof
sealed 11:17 UTC · OpenTimestamps: submitted 17 Nov 11:17 UTCebe94c226f20fadf8bf8f80c1f2b89900a430c27243eb2284ea3fdc27968c7bb
.ots proof
sealed 07:37 UTC · OpenTimestamps: submitted 17 Nov 07:37 UTCe8f568706266f88464e22942aae49acf7a73c1a8f471c8506c2d7ae4ab181d04
.ots proof
sealed 07:37 UTC · OpenTimestamps: submitted 17 Nov 07:37 UTC1b8b6fcf7d748f52c0cd7d96b6df310eb9218dd580c0a4a796d76b96779e4156
.ots proof
sealed 07:37 UTC · OpenTimestamps: submitted 17 Nov 07:37 UTCEvery forecast
171 forecasts
"Why it moved" is written by the AI after the window closed, from the day's news. It can be wrong, and it says so: clear, plausible or unclear is its own confidence in the explanation. Moves are in pips from the window's open.
Model's technical note
driver: US 10y -1.8bp since 21:00 (with up; r20 -0.08 weak; S&P futures +0.5% r20 -0.52 slightly against); dollar softer in early trade with EURUSD ~1.1196 +0.3%; German Sept retail sales slump limits euro; US ISM 15:00 is after window; L1 not applicable; stats: pre-window +0.48 ATR; no edge; EDGE after prev day down not used; snippets only - fetches partialModel's technical note
coin flip; no driver: yields -1.8bp favour down but S&P +0.5% (r20 +0.40) favours up so the drivers conflict; L1 applies and coin: up; stats: pre-window -0.32 ATR; no edge; EDGE continue not usedModel's technical note
driver: risk-off (S&P futures -0.5%; VIX +1.2) favours CHF; EZ inflation heating up limits ECB dovish case; EURCHF regime untrusted; L1 n/a; stats: pre-window -0.48 ATR; EDGE continue - not usedModel's technical note
driver: US 10Y -2.9bp since 21:00 (with up; r20 -0.25); US ECI 12:30 and Chicago PMI 13:45 in window; EZ Q3 GDP beat but sticky EZ inflation; gold -1.1% against; L1 n/a; stats: pre-window +0.31 ATR; no edgeModel's technical note
driver: US 10Y -2.9bp since 21:00 (with down; r20 +0.55) and risk-off (S&P -0.5%; VIX +1.2) favours CHF; ECI 12:30 in window; already -0.56 ATR pre-window; L1 n/a; stats: no edgeModel's technical note
driver: risk-off with S&P futures -0.3% and VIX +1.2 favours CHF; France Q3 contraction and EZ GDP 10:00 cons 0.1% in window; no cross-asset edge; L1 n/a (dated news); stats: pre-window -0.24 ATR; no edgeModel's technical note
US 10Y yield -1bp since 21:00 and lower overnight on AI-selloff flight to safety (with up; r20 only -0.25); EZ flash GDP 10:00 cons 0.1% in window; France Q3 contraction a drag; L1 n/a (dated news); stats: pre-window +0.01 ATR; EDGE reverse - not usedModel's technical note
driver: US 10Y -1bp since 21:00 after one-month-low yields overnight (with down; r20 +0.55); risk-off bid for CHF; L1 n/a (dated news); stats: pre-window -0.13 ATR; EDGE reverse - not usedModel's technical note
ECB 13:15 UTC decision in window; L1 applies (no dated pair driver; snippets only - fetches blocked); coin flip; driver: US 10y -1.4bp since 21:00 (flat; EURCHF regime don't trust); stats: pre-window -0.12 ATR; EDGE continue - not usedModel's technical note
ECB 13:15 UTC hold expected and US Q3 GDP/claims 12:30 in window; L1 applies (no dated pair driver; snippets only - fetches blocked; 6 searches); coin flip; driver: US 2y -0.8bp since 21:00 (flat - no support); stats: pre-window -0.31 ATR; no edgeModel's technical note
US GDP/claims 12:30 in window; L1 applies (no dated pair driver; snippets only - fetches blocked); coin flip; driver: US 10y -1.4bp since 21:00 (against up so 50); stats: pre-window +0.18 ATR; no edgeModel's technical note
Coin flip with no driver; EURUSD up and USDCHF down roughly cancel; ECB 13:15 after window; L1 applies - 50; driver: none (EURCHF regimes untrusted; US 10y -1.1bp r20 +0.50); stats: pre-window -0.09 ATR; no edgeModel's technical note
Fed held on Warsh with Dec hike bets barely moved (investinglive snippets only; tradingeconomics fetch blocked); ECB 13:15 and US GDP/claims 12:30 fall after the 12:00 close so no in-window release; L1 applies (no pair-specific driver) so kept near coin flip; driver: US 2y -1.4bp since 21:00 (with up; beta -2.1 pips/bp) - 51; stats: pre-window -0.24 ATR; EDGE continue not usedModel's technical note
Fed hold left USD little changed (snippets only); no in-window release; L1 applies (no pair-specific driver); driver: US 10y -1.1bp since 21:00 (with down; r20 +0.56 beta 3.3 pips/bp) and gold -1% - 51; stats: pre-window +0.15 ATR; EDGE not usedModel's technical note
FOMC day with press conf in window (cap 55); euro soft on French fiscal stress but CHF safe-haven bid is fading so only a lean; consistent with EURUSD down; driver: US 10y +0.9bp since 21:00 (weak link for EURCHF); stats: prev day -0.47 ATR; EDGE continue (LB 53.1 vs 51.8) - not usedModel's technical note
FOMC 18:00 UTC hold expected (hawkish; Warsh press conf in window so confidence capped at 55) while euro is weak on French bond stress and fiscal worries; driver: US 2y +1bp since 21:00 (with down; r20 -0.29); snippets and investinglive only - tradingeconomics fetch blocked; stats: pre-window +0.04 ATR; no edgeModel's technical note
FOMC 18:00 UTC hawkish hold priced (Warsh press conf in window so cap 55) with SNB on hold at 0% and USD firm; driver: US 10y +0.9bp since 21:00 (with up; r20 +0.59); investinglive says initial USD rise may fade - kept modest; stats: pre-window +0.27 ATR; no edgeModel's technical note
snippets only - tradingeconomics blocked; coin flip; no pair-specific driver; L1 applies; kept down at 50 for consistency with EURUSD x USDCHF; driver: none (US 10y +2.3bp r20 +0.50 not trusted for EURCHF); stats: pre-window +0.29 ATR; no edgeModel's technical note
snippets only - tradingeconomics blocked; GfK fell to -30.8 and markets sit pre-FOMC (18:00 outside window); driver: US 2y +1.8bp since 21:00 (with down; r20 -0.29 weak); L1 n/a; stats: pre-window -0.15 ATR; EDGE continue not usedModel's technical note
snippets only - tradingeconomics blocked; pre-FOMC dollar bid with oil rally; driver: US 10y +2.3bp since 21:00 (with up; r20 +0.59); stats: pre-window +0.39 ATR; no edgeModel's technical note
French fiscal and consumer weakness keep franc bid but pair rebounded from 0.9200 area; driver: euro-specific weakness (no overnight move measured; US 10y -1.8bp r20 +0.51 unreliable for EURCHF) - 51; L1 applies; consistent with EURUSD down x USDCHF up; stats: pre-window -0.35 ATR; EDGE continue - not usedModel's technical note
French consumer confidence at lowest since 2013 and ECB loan-demand worries weigh on euro; driver: US 10y -1.8bp since 21:00 (against down; r20 -0.12 weak) - 50; L1 applies (no confirmed supporting driver); US durable goods 12:30 cons 1.5% and CB confidence 14:00 cons 80.5 in window; FOMC Oct 28 outside window; snippets plus investinglive only - tradingeconomics fetch blocked; stats: pre-window -0.50 ATR; no statistical edgeModel's technical note
Coin flip no driver: US 10y -1.8bp since 21:00 moved against up (r20 +0.59 with 10y); L1 applies; durable goods 12:30 and CB confidence 14:00 in window; pre-FOMC caution; stats: prev day +0.77 ATR; EDGE reverse - not usedModel's technical note
French bond spread and fiscal worries keep franc bid but pair bounced from 0.92 area this week; driver: French spread (no overnight move found); US 10y -0.9bp (r20 +0.51; EURCHF regime unreliable); L1 applies; kept down at 51 consistent with EURUSD down x USDCHF up; stats: pre-window -0.07 ATR; no edgeModel's technical note
French fiscal risk and Treasury yields near 5% with Fed-hike debate weigh on euro but quiet EU calendar and no 12:30 data before window; driver: US 10y -0.9bp since 21:00 (flat; no support) - 51; L1 applies (no clear window driver); kept down at 51; snippets only - tradingeconomics fetch blocked; US durable goods 12:30 cons 1.5% and CB confidence 14:00 are after the EU window; stats: pre-window -0.2 ATR; EDGE continue - not usedModel's technical note
Coin flip no driver: US 10y -0.9bp since 21:00 moved against up (r20 +0.59 with 10y) so no support; quiet EU calendar; L1 applies; set 50; stats: pre-window +0.16 ATR; EDGE continue - not usedModel's technical note
coin flip; no driver; no dated pair news: 5 searches + investinglive fetch; tradingeconomics blocked; snippets only; L1 applies (no dated driver) - 50; FOMC Oct 28 and PCE/GDP Oct 29 outside window; no major US release in window; driver: none confirmed since 21:00; stats: pre-window +0.46 ATR; no statistical edgeModel's technical note
coin flip; no driver; no dated pair news: 5 searches + investinglive fetch; tradingeconomics blocked; snippets only; L1 applies (no dated driver) - 50; FOMC Oct 28 and PCE/GDP Oct 29 outside window; no major US release in window; driver: US 10y +0.9bp since 21:00 (neutral); stats: pre-window +0.05 ATR; no statistical edgeModel's technical note
coin flip; no driver; no dated pair news: 5 searches + investinglive fetch; tradingeconomics blocked; snippets only; L1 applies (no dated driver) - 50; FOMC Oct 28 and PCE/GDP Oct 29 outside window; no major US release in window; driver: US 10y +0.9bp since 21:00 (with up; r20 +0.6) too small - 50; stats: prev day +0.89 ATR; EDGE reverse - not usedCheck a forecast yourself — 3 steps
A hash is a fingerprint of a text: the same text always gives the same 64 characters, and changing a single letter gives a completely different one. So if the hash you saw before the window matches the forecast shown after it, the forecast was not changed.
7b147f2ee27226a91593d4cf033e0524…fx2owl-fx-v1|2026-11-09|eu|EURUSD|up|55|secret✓ MatchesForecasts made before 9 Nov 2026 have no hash. The secret at the end of the line is random and shown only after the result - without it nobody could check the hash, and with it anyone can.
When was the hash made? Check it on opentimestamps.org
Each hash is also sent to OpenTimestamps, a free public service that writes it into the Bitcoin blockchain within a few hours. Nobody - us included - can change a Bitcoin block afterwards, so this proves the forecast was not rewritten after that block.
- Open the forecast's "Sealed before the window" and download two files: Download as .txt (the line) and Download the timestamp proof (.ots).
- Go to opentimestamps.org. In "Stamp & Verify", drop the
.otsfile. - When it asks for the stamped file, drop the
.txtfile. It is hashed in your browser, not uploaded. - It shows the Bitcoin block and its date - the hash existed by then. In the first few hours after sealing it shows the proof as pending: try again later.
The "sealed" time next to each forecast is when our server received the hash; the block's date is independent of us but comes later, usually one to a few hours after sealing.
For developers: the command line
The hashed line is fx2owl-fx-v1|date|session|PAIR|call|confidence|salt, UTF-8, no newline
at the end. The salt is 32 random hex characters, kept secret until the window is scored, so the hash
can't be guessed from the few possible forecasts. To hash a line yourself:
printf '%s' 'fx2owl-fx-v1|2026-11-06|eu|EURUSD|up|55|00112233445566778899aabbccddeeff' | sha256sum
(on macOS use shasum -a 256). On Windows PowerShell:
$t = 'fx2owl-fx-v1|2026-11-06|eu|EURUSD|up|55|00112233445566778899aabbccddeeff'
-join ([Security.Cryptography.SHA256]::Create().ComputeHash([Text.Encoding]::UTF8.GetBytes($t)) | ForEach-Object { $_.ToString('x2') })
The output must equal the published hash. With the
OpenTimestamps client,
put the downloaded .txt and .ots in one folder (their names already match) and run
ots verify fx2owl-<hash>.txt.ots; ots upgrade on the same file saves the
completed proof. The client verifies against your own Bitcoin Core node (a pruned one is fine).
Updated 17 Nov 2026 11:34 UTC. Source links go only to independent news sites and official sources (central banks, statistics offices). Educational content, not financial advice. Past results don't guarantee future ones.