Liquidity LlamaLiquidity LlamaMarkets · Macro · Crypto
ResearchLatest report
← Back to homeTrack Record
Transparency

Track Record

Every idea the framework generates, with the date it was opened and where it stands now — marked honestly, whether it’s working or not. Ideas are logged when the monthly report publishes, not backfilled afterwards.

Track record chart
The GS / JPM spread — the ratio against its 45-day mean, with the entry marked.
4Open ideas
3Working
1Watch
29 Sep 2026Last updated
MonthlyMarking cadence
Open ideasmarks as at 6 Jul 2026 · not re-marked since
IdeaOpenedThesisMark (6 Jul)Status
GS / JPM spread Jul 2026 (Wk 1) Ratio below its 45-day mean of 3.22 — mean-reversion in relative bank valuation +2.4% as of 6 Jul Working
Korea — EWY Jun 2026 Won firming alongside improving regional export data +8.3% on the month, as of 6 Jul Working
India — contrarian May 2026 Contrarian entry; upgraded to PASS on a second consecutive positive month Upgraded as of 6 Jul Working
Cash-heavy positioning Nov 2025 ~95% cash, trading only — housing/permits tail risk not compensated n/a Watch
Update, 29 September 2026. The marks above have not been refreshed since 6 July. This page does not publish entry prices for these ideas, so a September re-mark cannot be reconstructed from it and I have not substituted unverified prices. The status labels are also as of 6 July. The next full re-mark comes with the September report, after the ISM release on 1 October.
Framework signal changes — the calls behind the ideas
DateSignalChangeTrigger
6 Jul 2026LabourNeutral → BearishJune NFP +57k vs ~115k consensus; fired the pre-set <75k rule
6 Jul 2026RatesBearish → Neutral2026 hike odds pared to ~42%; WTI −24% on the month
6 Jul 2026Volatility (ATR)Bullish → NeutralRealised vol ~1.5%/day, up ~50% — wider stops, smaller size
6 Jul 2026Composite score+7 → +6Net of the three flips above; headline stays bullish
4 Aug 2026LabourBearish → NeutralFactory employment 52.8 from 49.7 — first expansion in 33 months; payrolls on 7 Aug were the test
4 Aug 2026VolatilityNeutral → BullishVIX 16.49 from 20.65 — the late-July stress lasted days, not weeks
4 Aug 2026RatesNeutral → BearishLong bonds still pinned near 52-week lows; the July upgrade did not hold. Not in the original log, added 29 Sep 2026
4 Aug 2026Composite score+6 → +10ISM 55.6 four-year high, employment expanding, vol collapsed; rates the lone bearish holdout. Corrected from +9 to match the report’s own 11 bull / 3 neutral / 1 bear tally. The three flips logged here net to +1; the other +3 came from signals whose changes were not logged
7 Aug 2026LabourNot confirmedJuly payrolls first printed at −23k, not the 125k+ the upgrade rule needed (later revised to +21k). The 4 Aug report treated payrolls as the confirmation, and they did not confirm it. BLS
Sep 2026ModelReplacedThe old 15-signal equal-weight score was replaced by the 12-row weighted v3 model (dissertation part plus a labelled overlay). Scores before and after are not comparable
Sep 2026Composite score−2 NeutralFirst score on the v3 model. Not a like-for-like move from +10. The Fed raised rates on 16 September. The full report follows the ISM release on 1 October

How to read this page

These are research ideas generated by a framework, not a trading account. They carry no position sizing, no leverage assumptions and no fee or slippage drag, so they should not be read as a return stream. What they do show is whether the framework’s calls were directionally right, and how quickly it admits when they aren’t.

Marks are updated once a month when the regime report publishes. An idea is closed when its thesis is invalidated or the signal that generated it flips — and closed ideas stay on this page, including losers. A track record that only shows winners is a marketing document, not a record.

Sample size caveat. This log is young. A handful of ideas over a few months tells you very little about skill either way — the honest read is that it takes years of marks before a record means anything. It’s published from the start so the history is real rather than reconstructed.

For educational purposes only. These are framework-generated research ideas, not recommendations, not advice, and not a representation of actual trading returns. Past performance does not indicate future results.