Trading System — Live Monitor

Trading System ● Live Monitor

Active book: crypto 4h trend · PAPER only · as of 2026-11-30 · real-money gates 6/6
regime stable
factor: crypto_beta · macro driver: GOLD

Returns

BookValue $DayWeekMonth YTDTotalP&L $
THE ACCOUNT (everything, one book)$1,125.83-0.08%-0.7%-0.9%+12.58%+12.58%+125.83$
· crypto sleeve (hourly, 24/7)$623.580.0%0.0%0.0%+24.72%+24.72%+123.58$
· stock sleeve — signal-driven (market hours)$502.25-0.17%-1.57%-1.99%+0.45%+0.45%+2.25$
Crypto sleeve: equity $623.58 (+123.58$) · worst drawdown -9.35% · both sleeves trade on signal; account history starts when both sleeves went live

The Account — one book, two sleeves ALERTS: brain stuck in offline fallback (check Together credits/billing first, then key/endpoint)  equity $1125.83 (+125.83$) · crypto sleeve $623.58 · stock sleeve $502.25 · started $1,000 on Jul 23, 2026

Crypto sleeve — trades 24/7 · return = since entry
PositionSideSize $ % of sleeveHeldReturn
no open positions
Stock sleeve — signal-driven, market hours · return = since entry · 72.7% invested, rest in cash (names failing the trend filter)
PositionSideSize $ % of sleeveHeldReturn
HYGLONG$62.5312.5%12d 21h-0.62%
AGGLONG$54.9110.9%12d 21h-0.36%
TLTLONG$27.165.4%12d 20h-0.58%
EFALONG$19.083.8%2d 1h-2.11%
DIALONG$17.293.4%1h 26m-1.52%
XLULONG$16.863.4%9d 23h-1.63%
VWOLONG$14.362.9%9d 21h-5.66%
XLILONG$14.302.8%12d 19h-3.25%
IWMLONG$13.942.8%3d 1h-3.19%
HONLONG$12.702.5%9d 21h-3.98%
XLCLONG$12.452.5%12d 20h-2.37%
BACLONG$12.282.4%5d 23h-3.92%
MCDLONG$12.272.4%7d 1h-2.07%
GLDLONG$11.862.4%5d 23h-5.02%

Trades per day  1/11: 36 · 1/12: 20 · 1/13: 70 · 1/16: 7 · 1/17: 5 · 1/18: 11 · 1/19: 22  ·  view full trade log →

Thu Nov 19, 10:30 AM ET — Sold $3 of AGG at $95.46
Thu Nov 19, 10:30 AM ET — Sold $4 of HYG at $76.74
Thu Nov 19, 10:16 AM ET — Bought $6 of AGG at $95.39
Thu Nov 19, 10:16 AM ET — Bought $2 of EFA at $100.17
Thu Nov 19, 10:16 AM ET — Bought $2 of TLT at $79.92
Thu Nov 19, 10:16 AM ET — Bought $7 of HYG at $76.73
Thu Nov 19, 9:42 AM ET — Sold $2 of PEP at $137.48
Thu Nov 19, 9:36 AM ET — Sold $4 of VWO at $53.7

Signal fleet — speculative strategies hunting for alpha (both asset classes)

Crypto — 24/7, trades whenever its signal fires
StrategyValue $ReturnTrades
meanrev stretch$161.91+61.91%1866
oversold bounce$143.02+43.02%902
meta combo$135.58+35.58%4813
ml ridge$131.64+31.64%6656
live minhold 12h$118.18+18.18%1704
online sgd$112.06+12.06%5259
live minhold 2h$108.46+8.46%1644
live minhold 0h$108.27+8.27%1623
below ma revert$107.52+7.52%5132
outcome sized$99.05-0.95%451
ml gbm$94.35-5.65%5917
kelly trend$93.49-6.51%2065
live convnotch on$91.97-8.03%51
live convnotch off$90.32-9.68%32
dip buy uptrend$86.55-13.45%1998
trend slow$76.22-23.78%2444
funding tilt$71.91-28.09%393
momo accel$71.42-28.58%1975
breakout range$70.34-29.66%483
adaptive exit$70.11-29.89%3264
trend med$70.06-29.94%3347
momentum phase$70.06-29.94%5814
relative strength$69.92-30.08%1178
vol contraction$69.91-30.09%615
trend fast$69.21-30.79%3781
vol expansion$66.83-33.17%124
Stocks — market hours, trades whenever its signal fires
StrategyValue $ReturnTrades
kelly trend$98.21-1.79%3607
outcome sized$92.93-7.07%5136
overnight hold$92.89-7.11%6905
trend slow$90.68-9.32%1773
momentum phase$88.61-11.39%8582
relative strength$88.18-11.82%3056
adaptive exit$87.12-12.88%4016
below ma revert$86.34-13.66%4878
trend med$82.85-17.15%3707
meta combo$73.54-26.46%4979
ml gbm$73.47-26.53%13004
trend fast$72.10-27.9%8079
vol expansion$70.68-29.32%3427
dip buy uptrend$70.63-29.37%1865
ml ridge$70.55-29.45%2409
breakout range$69.92-30.08%2966
vol contraction$69.92-30.08%6159
momo accel$69.92-30.08%10988
online sgd$69.87-30.13%6427
oversold bounce$69.61-30.39%481
meanrev stretch$41.46-58.54%1290
Stock sleeve traded on signal (challenger to the monthly book): $502.25 · +0.45% · 20 positions · 72.6% invested · 2330 trades — acts whenever the trend actually flips, with a cost governor, instead of waiting for month-end.
What this is: each strategy runs its OWN separate virtual $100 book — none of this is inside The Account above. Returns are NET of estimated costs (10 bps per side charged on every fill). Trades happen only when that strategy's own signal changes, so trade counts show each one's natural cadence. Several are expected to fail — losers are information about which regimes a signal breaks in. Live forward results decide what earns promotion into the real book.

The story — what the system is doing, in plain English

What we're running — and why

One paper account ($1,000), two sleeves of $500 each, mirroring the eventual single brokerage account. The STOCK SLEEVE holds trend-qualified names from a 66-stock liquidity-screened universe and trades whenever a trend signal actually flips (cost-governed, market hours). The CRYPTO SLEEVE trades 10 liquid pairs 24/7, long-only (shorts aren't fillable on spot venues), acting on confirmed 4h trend flips; right now it's flat. Orders also route to a REAL broker (Alpaca paper) so fills and slippage are measured, not assumed. An AI reasoning layer reviews live conditions every 30 minutes and judges events in context; an autopilot promotes or benches whole strategies based only on live results.

What's working / what's not

Best recent performer: static 60 40 (risk-adjusted score 1.4); weakest: brain directed (-3.05). The consistent pattern: defensive trend-following and adaptive blends work; raw momentum without a regime filter doesn't. Important honesty: these rankings are still mostly HISTORY. The true out-of-sample scorecard started accumulating — it has 3 month(s) so far; live results, not history, will decide who keeps capital.

Thesis check & alpha-decay watch

Thesis check: the system profiles every market it watches — of 85 securities, 16 currently behave momentum-driven and 35 mean-reverting, and strategies are weighted to fit each one rather than forced onto all. The core bet — that no single strategy survives regime change, so the edge is in adapting — is so far SUPPORTED by the data: every static strategy we tested decays somewhere in history, while the adaptive blend holds up. Alpha-decay watch: raw momentum's recent score is far below its long-run average (crowding), which is why the book leans defensive trend instead.

Post-mortem vs the market

Last month the champion book returned -3.4% vs the S&P's -3.8% — it outperformed. Why: the book is deliberately defensive (big dollar-index position, no crypto longs, trend-qualified equities only), so it lags in sharp rallies and protects in selloffs — that's the design, not an accident. Its whole 25-month simulated run: +26.3% with a worst drawdown of -4.87%, i.e. it earns by not losing.

What's actually working — from our own trades

Across 529 closed round-trips: 47.1% win rate, +23 bps average, 81.2h average hold. Strongest patterns: symbol: 'CRM' averages +111 bps (n=8) vs 'BA' -260 bps (n=8); entry_hour_et: '12PM' averages +245 bps (n=27) vs '2PM' -159 bps (n=18); conviction: 'high >=0.6' averages +63 bps (n=124) vs 'mid 0.35-0.6' -66 bps (n=101). Cuts with fewer than 8 trades are marked provisional — the system will not act on a pattern it cannot yet distinguish from noise.

What drives our returns — adaptive factor set

We test a registered universe of factors and keep only what explains our own returns. Ranked now: CRYPTO_BREADTH (crypto), BTC (crypto), ALT_DISPERSION (crypto). Active set: CRYPTO_BREADTH, BTC — everything else was dropped for carrying no marginal explanatory power. Together they explain 3.9% of variance, so 96% remains unexplained: the strategies are largely idiosyncratic rather than factor-driven, which is worth knowing honestly. Notably all five macro factors were dropped — crypto breadth, not macro, is what moves this book. The set is re-derived every cycle, so it follows the market rather than a choice made once.

Fine-tuning: status & recommendation

The AI brain is NOT being fine-tuned yet — deliberately. A training set builds itself daily from the system's own decisions and their real outcomes (342 examples so far), but an anti-overfitting gate blocks training until the data is diverse and balanced (current blocker: too few examples (342 < 5000)). Recommendation: keep accruing through varied market conditions; once the gate opens, run a light LoRA fine-tune of the open-weight model on outcome-labeled decisions (teaching it when to veto and downsize, not to predict prices), then A/B it against the untuned brain on the forward leaderboard before it touches allocations. Warranted right now: no.

Risk & exit triggers

Exit triggers in force: the crypto book exits within hours when the 4h trend or its regime envelope breaks, and a kill-switch flattens everything at -25% account drawdown (not triggered). The monthly book exits any holding whose long-term trend breaks — but only at month-end rebalance; single positions are small (~3-8% each) and a market-stress scaler halves the whole book after a sharp broad selloff, which bounds mid-month damage from a single blowup (e.g. earnings). Position watch: no held name has broken trend this month.

Path to real money

Path to real money: 6 of 6 gates passed. The missing gates all need LIVE time — months of real forward record with adequate risk-adjusted return and controlled drawdown, plus a month of clean execution on the $100 account. No shortcuts: when all gates go green the dashboard flags it, and going live is still a deliberate human decision starting tiny.

Active book history (25 months)

MonthReturn
2026-11-3.41%
2026-10+0.76%
2026-09-1.64%
2026-08+1.39%
2026-07+0.19%
2026-06-0.79%
2026-05+1.86%
2026-04+4.39%
2026-03-3.46%
2026-02+1.53%
2026-01+2.26%
2025-12+0.56%

Strategy leaderboard — forward test locked 2026-08-01

StrategyFwd moFwd ret BacktestBt Sh
static 60 403+0.3%+13.4%1.4
champion learning brain3+0.5%+23.3%1.38
brain plus crypto4h3+0.5%+23.3%1.38
TSM standalone3+0.5%+24.3%1.34
brain fixed share3-0.4%+19.6%1.29
universal adaptive3-1.0%+4.2%0.63
crypto 4h trend3+15.9%+11.7%0.34
HRP experts3-1.0%+0.4%0.12
momentum realized gate3+0%-0.3%-0.14
momentum ungated control3-9.5%-6.0%-0.22
equal weight experts3-4.9%-5.4%-0.47
momentum VIX gate3-9.5%-10.7%-0.56
momentum DELTA gate3-9.5%-13.4%-0.74
rl bandit3-2.1%-23.6%-2.43
brain directed3-1.3%-1.8%-3.05
Fwd = live out-of-sample (the truth, fills from inception). Backtest = history, context only.

Automation footprint (live crypto trades)

SymbolBurstinessSize entropyFlow imbalance
BTC/USD1.870.2910.321
ETH/USD2.480.3830.206
High burstiness + low entropy = heavy algo footprint — one input to the "how botted is this market" thesis.
Everything above is generated from live system state · deterministic core + open-weight AI brain + autonomous improvement agent · PAPER, no real orders · daily audit snapshots in git (record/)