baseline_momentum_live_v1 Live Review Package

Generated 2026-06-05T02:08:43.661937+00:00 · read-only review package · no orders submitted

Metric Glossary

Transaction Costs (TC) are estimated fees, spread, slippage, and market-impact assumptions deducted from gross returns. Gross Sharpe Ratio is the annualized return/risk ratio before estimated Transaction Costs. Net Sharpe Ratio is the annualized return/risk ratio after estimated Transaction Costs. Compound Annual Growth Rate (CAGR) is annualized compounded return. Maximum Drawdown (MaxDD) is the worst peak-to-trough loss. Out-of-Sample (OOS) is the period not used for fitting or tuning. Algorithm Trading (AT) is the offline algorithm/backtest path, Paper Trading (PT) is broker paper-account execution, and Real Trading (RT) is live-money execution.

Review Verdict
needs_fix_before_live_expand
Net Sharpe Ratio
0.833
Compound Annual Growth Rate (CAGR)
35.49%
Maximum Drawdown (MaxDD)
-57.92%
Current Long
$4,591.23
Latest Signal
2026-06-04

Manual Mode And Risk Reduction - 2026-06-05

Status: unattended automation disabled. The strategy is now manual Wednesday review/trigger only; live orders require explicit approval.

ItemCurrent Setting / Result
AutomationDisabled in crontab, StrategyRegistry, schedule registry, runtime config, and strategy state.
Manual Wednesday review pathpython /workspace/steve/alpha_vault/tools/ops_baseline_deploy.py --environment live --budget 10000 --metrics-only
Approved cap reductionSell-only cap reducer submitted 6 orders through ExecutionGateway + AlpacaBroker at 2026-06-05T19:24Z.
Filled sellsAMAT, AMD, GS, MU, QCOM, SOXX all filled.
Verified exposure afterAbout $10,020 managed open market value; broker open orders = 0.
Future live order pathCONFIRM_LIVE_CAP_REDUCE=true python tools/ops_baseline_cap_reduce.py --cap <target_usd> --submit-live --wait-seconds 60, only after explicit approval.

Resolved Ops Findings

Schedule mismatch resolved operationally: the unattended baseline cron command was disabled on 2026-06-05 and the strategy is now manual-only.

Budget mismatch resolved operationally: runtime config and schedule registry now use manual review mode with target cap about $10k. Do not restore a $20k unattended entrypoint without a new approval.

Strategy Eval Metrics

MetricValue
Window2022-12-13 to 2026-06-02 · 869 trading days
Gross Sharpe Ratio / Net Sharpe Ratio0.877 / 0.833
Total Return Gross / Net209.99% / 185.03%
Net Compound Annual Growth Rate (CAGR) / Volatility / Maximum Drawdown (MaxDD)35.49% / 54.99% / -57.92%
Out-of-Sample (OOS) Sharpe Ratio3.158 since 2026-01-01
Rolling Sharpe Ratio126 trading-day slow/full-context latest 2.846; 30 trading-day fast/recent latest 7.110
Correlation Matrixmax absolute correlation 0.458; pass<0.8=True
Cost Assumptionannual Transaction Costs (TC) drag 2.43% Net Asset Value per year
Biaslookahead shift test pass=True
Needs InputPaper Trading (PT) / Real Trading (RT) return stream; fill latency, queue position, partial fill assumptions

Eval dashboard · Bias report · Internal report · Summary markdown with glossary

Rolling Sharpe Ratio Chart

Visible chart generated from gross_returns.csv. It shows Algorithm Trading (AT) gross return rolling annualized Sharpe Ratio, not a live broker profit-and-loss stream. Default standard windows are 126 trading days as the slow/full-context line and 30 trading days as the fast/recent line.

-8-7-6-5-4-3-2-10123456789 20222023202420252026 Rolling Sharpe Ratio from gross return stream 126 trading days (slow/full context)30 trading days (fast/recent) Annualized Sharpe Ratio X-axis is rolling-window start date; latest window ends 2026-06-02.

Current Live Status

AccountAitist Alpaca Real Trading (RT) live 865201662
Cash$38,917.20
Portfolio value$43,508.43
Open orders0
Configured scheduled budget$20,000.00
Last scheduled rebalance2026-05-29T15:30:24.215181Z
Last manual cap reduce2026-06-03T19:52:54.557080Z

Positions

SymbolQtyMarket ValuePriceUnrealized Profit/Loss
AMD2.124510908$1,074.09$505.57$50.54
LRCX1.744626956$570.49$327.00$13.72
MU1.630618998$1,552.33$951.99$273.10
QCOM3.464310587$824.51$238.00$21.66
SOXX0.976535738$569.81$583.50$17.05
UNH0.000000861$0.00$396.86$0.00

Latest Signal Snapshot

Signal date 2026-06-04 from latest closed daily bar. Public legacy dashboard signal is stale; this page is current for the review package.

Symbol20d MomentumFull Strategy Target WeightCurrent Live State
MU54.35%33.0%held
AMD28.14%17.1%held
SOXX22.39%13.6%held
AMAT22.22%13.5%not held
QCOM19.70%11.9%held
GS18.10%11.0%not held

Rebalance Timing Sensitivity Study

Net Sharpe Ratio means annualized return/risk after estimated transaction costs. Gross Sharpe Ratio is before those costs. Compound Annual Growth Rate (CAGR) is annualized compounded return, and Maximum Drawdown (MaxDD) is worst peak-to-trough loss. This study is read-only and uses daily OHLC proxies from Alpaca; intraday open+30m, mid-day, and 15:30 ET bars were not used, so the chart does not claim intraday precision.

Best Combo
Wednesday close/MOC

Net Sharpe Ratio 1.05; MaxDD -39.5%.

Current Friday Proxy
0.55

Prior-close signal, Friday close/MOC proxy; MaxDD -45.9%.

Cost Assumption
10 bps

One-way cost per turnover; average trade cost shown as turnover x 10 bps.

Policy Read
Semi-auto

Use a human/intraday veto until 15:30 ET intraday timing is validated.

Weekday x execution Net Sharpe Ratio heatmapPrior close -> next openPrior close -> close/MOCSame-day close -> next openMonday0.720.700.76Tuesday0.650.420.78Wednesday0.821.050.83Thursday0.770.780.66Friday0.690.550.51Best: 1.05; daily OHLC proxy, not intraday precision.
Execution-time average Net Sharpe Ratio0.73Prior close -> next openbest combo 0.820.70Prior close -> close/MOCbest combo 1.050.71Same-day close -> next openbest combo 0.83
Risk chart: Maximum Drawdown (MaxDD) by timing optionPrior close -> next openPrior close -> close/MOCSame-day close -> next openMonday-52.7%-54.3%-53.7%Tuesday-53.7%-62.3%-53.1%Wednesday-53.1%-39.5%-41.3%Thursday-42.1%-47.6%-45.4%Friday-43.7%-45.9%-56.1%Best: -39.5%; daily OHLC proxy, not intraday precision.
WeekdayExecution proxyNet Sharpe RatioGross Sharpe RatioCAGRMaxDDTurnoverAvg cost bps
WednesdayPrior close -> close/MOC1.051.1555.3%-39.5%0.939.3
WednesdaySame-day close -> next open0.830.9337.6%-41.3%0.919.1
WednesdayPrior close -> next open0.820.9138.9%-53.1%0.939.3
TuesdaySame-day close -> next open0.780.8836.4%-53.1%0.929.2
ThursdayPrior close -> close/MOC0.780.8934.6%-47.6%0.939.3
ThursdayPrior close -> next open0.770.8734.2%-42.1%0.939.3
MondaySame-day close -> next open0.760.8536.5%-53.7%0.979.7
MondayPrior close -> next open0.720.8134.3%-52.7%0.959.5
Stress subsetWednesday close/MOCFriday close/MOC
Recent semiconductor stress proxy2026-03-04 to 2026-04-29; return 50.3%, net Sharpe 4.36, worst weekly loss -5.1%2026-03-06 to 2026-04-24; return 9.2%, net Sharpe 1.33, worst weekly loss -14.3%

Recommendation panel

Research conclusion: daily proxy favors Wednesday close/MOC over Friday. Operational conclusion: do not make this fully automatic yet. Today exposed a schedule/cap mismatch, and the requested 15:30 ET close-30m timing still needs true intraday-bar validation.

Discussion default: semi-automatic weekly rebalance, candidate timing Wednesday close/MOC or Friday true 15:30 ET only after intraday validation, with an intraday risk veto when SPY/QQQ/SOXX are weak or when live cap differs from scheduled budget.

Artifacts: timing_sweep.json · timing_sweep_summary.md · generator tools/baseline_timing_sweep.py.

Real Trading vs Paper Trading Cash Results

Apples-to-apples: NO. No true Paper Trading (PT) mirror return/cash ledger exists for this strategy in the available review artifacts. The closest comparable source is the Algorithm Trading (AT) Strategy Eval simulation, so this section is a diagnostic bridge, not an execution-quality verdict.

RT cash/P&L versus closest AT simulation Not apples-to-apples: RT is live cash snapshot after manual overlays; AT is full-window simulated net result. $0k$10k$20k$30k$40k $22,305.16 RT start cash 2026-06-03 $16,612.04 RT cash raised sell cashflow $38,917.20 RT end cash review snapshot $376.07 RT unrealized P&L open sleeve $20,000.00 AT start value simulation $37,006.80 AT net P&L 2022-2026 Gap: -$36,630.73 (-99.0% vs AT P&L) RT cash bridge: start cash + sell cashflow = current cash RT open-position unrealized Profit and Loss (P&L) Closest comparable: Algorithm Trading (AT) simulated net Profit and Loss (P&L), not Paper Trading (PT)
RT Cash Window
$22.3k -> $38.9k

2026-06-03 risk-off through review snapshot; sell proceeds about $16.6k.

RT Open P&L
$376.07

Current unrealized Profit and Loss in remaining baseline symbols; realized attributed P&L unavailable.

Closest AT Net P&L
$37,006.80

$20k simulated sleeve, 2022-12-13 to 2026-06-02 Strategy Eval window.

Gap
-$36,630.73

RT unrealized P&L versus AT full-window P&L; scale/window diagnostic only.

FieldReal Trading (RT)Paper Trading (PT) / Comparable Evidence
Account / environmentAitist Alpaca live alpaca_live_main, account 865201662No true PT mirror. Comparable source: AT Strategy Eval using paper/data historical bars.
Comparison window2026-06-03T19:36:34Z light risk-off through 2026-06-05 review snapshot2022-12-13 through 2026-06-02 Strategy Eval window
Starting cash / capital base$22,305.16 before approved 2026-06-03 risk-off/cap-reduce overlays$20,000 assumed sleeve start for cash bridge
Ending cash / ending value$38,917.20 cash in review snapshot; remaining long value about $4,591.23$57,006.80 simulated ending value from net total return
Cashflow / P&LSell-order proceeds about $16,612.41; observed cash change about $16,612.04. Current RT unrealized Profit and Loss (P&L) in remaining baseline symbols about $376.07. Strategy-level realized P&L unavailable without a complete attributed tax-lot/cash ledger.Net simulated P&L about $37,006.80 over the full AT window; realized/unrealized split not applicable to this daily return stream.
Absolute / percentage differenceRT observed unrealized P&L versus AT full-window net simulated P&L: -$36,630.73, or -99.0% relative to AT. This is not a valid PT-vs-RT performance gap because the windows, scale, and overlays differ.
Known causesNo PT mirror; live/paper fill assumptions; slippage; partial fills; timing; account constraints; shared-account attribution; manual light risk-off and cap-reduction overlays on 2026-06-03.

Sources: RT state from /workspace/steve/alpha_vault_ops/data/runtime/memory/strategies/baseline_momentum_live_v1/state.json; public RT snapshot embedded in this page; AT inputs reports/review_packages/baseline_momentum_live_v1_20260603/inputs/; eval summary summary.json; review docs docs/baseline_momentum_live_review_2026-06-05.md.

Strategy Algorithm

Visible end-to-end path from data input to live monitoring. The diagram reflects the canonical scheduled path plus the manual sell-only risk-off override used on 2026-06-03.

1. Data Inputscurrent.yaml universeAlpaca daily bars + SPY 2. Timestamp Cutoffdrop current daily baruse latest closed candle 3. Feature Signal20 trading-day momentumprice / lag price - 1 4. Rank / Filterpositive momentumtop 6, exclude SPY 5. Target Weightsselected momentumdivided by sum 6. Sleeve Sizingtarget dollars = budgetminus ignored baseline qty 7. Risk Controlsmin trade thresholdqty precision + dust rules 8. Schedule GuardFriday close windowflock + duplicate date 9. Order IntentExecutionGatewayclient_order_id prefix 10. Broker SubmitAlpacaBroker onlyno direct submit_order 11. Runtime Logssignals, metrics, ordersStrategyStateStore 12. Monitoringlive status + dashboardreview feedback loop Discretionary Override PathApproved sell-only helpers: light risk-off and pro-rata cap reduction.They refuse buys, new symbols, open-order conflicts, and closed-market submission.

Code Walkthrough

Algorithm Trading / Paper Trading / Real Trading Separation

LayerStatus
Algorithm Trading (AT)Historical Strategy Eval complete through 2026-06-02; P0 pass, P1/P2 contain needs_input items.
Paper Trading (PT)Paper/data account used for historical data; paper return stream is still needs_input.
Real Trading (RT)Live Aitist account active, current exposure near $5k after approved sell-only cap reduction; no open orders in read-only check.

Export Provenance

ItemValue
Metrics producerPAW strategy-eval skill using tools/eval_strategy.py.
Strategy Eval command patternpython tools/eval_strategy.py --positions positions.csv --asset-returns asset_returns.csv --gross-returns gross_returns.csv --metadata-json metadata.json --factor-returns factor_returns.csv --existing-returns existing_returns.csv --strategy-label baseline_momentum_live_v1 --out-dir eval-report
Evaluation inputsreports/review_packages/baseline_momentum_live_v1_20260603/inputs/
Evaluation outputsreports/review_packages/baseline_momentum_live_v1_20260603/eval-report/
Data sourceAlpaca daily bars through tools.ops_baseline_deploy.latest_targets(...); live account read via AlphaVault Ops create_trading_client("live").
Timestamp cutoffCurrent in-progress daily bar is dropped by closed_daily_prices(...); latest signal date shown here is 2026-06-04.
Report producerCustom static HTML generated from Strategy Eval artifacts, gross_returns.csv, and a read-only Real Trading (RT) status snapshot.
Publisher/deploy pathCloudflare Pages project paw-baseline-live-review, deployed from reports/review_packages/baseline_momentum_live_v1_20260605/public/.