Track record

Every signal is logged with a timestamp before its outcome exists; +1 / +5 / +20 trading-day returns are filled in afterwards. Losers stay. BACKTEST and LIVE are labeled apart and never mixed. One git commit a night is the notary.

All signals

Kind
Mode
kindtickertime (UTC) +1d+5d+20d modesummary
Loading…

Hit rate by source

Hit = the return's sign matches the signal's direction. Every percentage carries its N=; rows with N below 20 are early readings, not conclusions.

kindmodeN +5d hit+20d hit+20d avg
Loading…

K-index backtest

BACKTEST

Loading…

USD liquidity backtest BACKTEST

Reconstructed backtest · N=1 event · bought at the signal-day close, adjusted closes · data as of 2026-08-28

asset2026-04-08 close → last +1w+1m +3mto date
QQQ 605 → 716 +5.2% +14.7% +19.5% +18.3%
SPY 674 → 769 +3.5% +8.2% +11.5% +14.1%
SMH 423 → 553 +7.1% +27.7% +43.7% +30.8%

Reconstructed backtest (BACKTEST) · N=1 event · index = funding spread + bank reserves + TGA flow + net-liquidity momentum, reviewed every trading day, pushed on regime change · a record, not a promise

Hypothetical, equal-weight, excludes fees, slippage and taxes. Past performance is not indicative of future results. Not investment advice.

Hypothetical equity curve LIVE · pushed

LIVE signals, equal-weight, held 20 trading days, base 100, no fees / slippage / taxes. Another drawing of the ledger — not any real account.

Notes

  1. +1 / +5 / +20 = close on the 1st / 5th / 20th trading day after the signal vs. the signal-day close (%). Blank means the day hasn't arrived.
  2. BACKTEST rows are historical replays; LIVE rows were actually pushed at the time. They are tallied separately and never mixed.
  3. The ledger file is committed to GitHub nightly; the history shows every change.

Backtests are hypothetical and exclude fees, slippage and taxes. Latency figures are measured, not a service commitment. Not affiliated with Telegram, the SEC, or any issuer named.

回测为假设性结果,不含手续费、滑点与税费。延迟数字为实测值,不是服务承诺。 与 Telegram、美国证监会(SEC)或文中提到的任何发行人均无关联。