Russia Foreign Exchange Reserves

The tests show no consistent, material advantage over the no-change baseline.

Model: Agentic Investor TSFM (foundation model)

Accuracy vs naive

+7%Matches the no-change baseline over 12 tests (12 tests, 7 won)

Direction

58%No clear edge over a coin flip (83 of 144 calls right in 12 tests)

Turning points

2 of 45Anticipated some turns (45 turns in 12 tests, 0 without a turn)

Error vs baseline

7% lowerError ratio 0.93; 1.00 is a tie.

Tests won

7 / 12Complete rolling backtests; ties count as no win.

Test period

2014-11-01 to 2026-10-0112 observations ahead per test.

80% band coverage

79%Observed outcomes inside the band across all stored test windows.

• Band coverage uses all stored test windows, including model-selection windows when present; it does not determine the confidence label.
• Evaluated 2026-11-15 against the cached history through 2026-10-01.
• This is an evaluation of the serving model; it does not establish a winner across four model classes.