[ PRACTICE ACCOUNT · ALPACA ]
LIVE · 20S POLLPractice trading account.
Total value, stocks held, live P&L, kill-switch runway, signal freshness — all live from the practice broker (Alpaca). The page renders instantly from the last committed snapshot, then rehydrates every 20s. Fake money. Real prices. No real dollars at risk.
Alpha vs S&P 500 · live
loading…
How the bot has done compared to buying the index ETF and doing nothing. Real money would have been better off in the boring choice if this number is red. Refreshes every 20 seconds from the broker.
[ EQUITY HISTORY ]
Need at least two history points to plot. Cron writes one row per run; chart hydrates automatically after next scheduled cron.Drawdown from peak · live
-4.82%
Peak $100,389
Below peak -$4,843
How far below the highest balance the account has ever hit. Real traders call this the worst pain a strategy puts you through. Red means below peak; green means at a new high.
Rolling 60-day stats
Sharpe (annualised)
-3.19
Return per unit of bumpiness. Above 1.0 is professional grade. Negative means losing for the risk taken.
Up-day rate
25.0%
How often the day-end balance was higher than the day before. Above 55% is solid; 50% is coin-flip.
Return (last 60 d)
-4.45%
Compounded return over the window. Compared against the headline alpha vs S&P 500 above this is the absolute view.
Quick read · the page in 60 seconds
A model picks 20 large US stocks every market day and trades them with practice money on a real broker (Alpaca). Real prices, real fills, no real dollars at risk. The numbers on this page reflect the current state of that practice account, live.
+What every word here means
- Equity
- Total worth of the account right now — cash plus the live market value of every stock owned.
- Cash
- Money sitting in the account that isn't invested in any stock.
- Buying power
- The maximum that could be spent on new stocks right now. Usually higher than cash because the broker is willing to lend a little.
- Position
- Any stock the account currently owns. "20 positions" means twenty different stocks are held.
- Profit / Loss
- How much each stock is up or down compared to what it was bought for. Not real until the stock is sold.
- Practice account
- Real broker, real prices — but fake money. Nothing on this page can lose or earn real dollars.
LIVE · practice accountupdated just now · refreshes every 20s
- Equity
- $95,548.08
- Cash
- $86,854.76
- Buying power
- $371,117.04
- Status
- ACTIVE
▲ +$0.00 (+0.00%) today
Total value of the account = cash + everything the stocks you own are worth right now.
Money sitting unused — not invested in any stock right now.
Most the broker lets you spend on new stocks right now (cash + a margin loan).
Account is healthy and allowed to trade. Paper = fake money, no real risk.
Building intraday curve — first live points arrive within ~20s. Each point is one snapshot of your equity. Together they form today’s live curve.
Stocks currently held (18)
Live value of every stock owned.
| Ticker | Type | Shares | Bought at | Now | Value now | Profit / Loss | % change |
|---|---|---|---|---|---|---|---|
| AMAT | long | 1 | $575.50 | $572.94 | $572.94 | -$2.56 | -0.45% |
| AMD | long | 1 | $511.88 | $537.00 | $537.00 | $25.12 | 4.91% |
| CIEN | long | 1 | $410.28 | $441.35 | $441.35 | $31.07 | 7.57% |
| COHR | long | 1 | $324.50 | $307.00 | $307.00 | -$17.50 | -5.39% |
| DDOG | long | 2 | $260.15 | $259.45 | $518.90 | -$1.40 | -0.27% |
| DELL | long | 1 | $400.41 | $422.38 | $422.38 | $21.97 | 5.49% |
| DVA | long | 2 | $233.39 | $235.58 | $471.16 | $4.39 | 0.94% |
| FLEX | long | 4 | $139.02 | $128.28 | $513.11 | -$42.97 | -7.73% |
| FTNT | long | 3 | $159.96 | $158.11 | $474.32 | -$5.56 | -1.16% |
| GLW | long | 3 | $190.68 | $181.84 | $545.52 | -$26.53 | -4.64% |
| GNRC | long | 1 | $250.26 | $226.95 | $226.95 | -$23.31 | -9.31% |
| HPE | long | 12 | $43.19 | $46.96 | $563.52 | $45.24 | 8.73% |
| INTC | long | 5 | $122.54 | $102.90 | $514.50 | -$98.20 | -16.03% |
| MRNA | long | 8 | $79.94 | $67.01 | $536.08 | -$103.40 | -16.17% |
| MRVL | long | 4 | $242.51 | $216.00 | $864.00 | -$106.03 | -10.93% |
| PANW | long | 1 | $348.20 | $329.10 | $329.10 | -$19.10 | -5.48% |
| VRT | long | 1 | $305.00 | $304.50 | $304.50 | -$0.50 | -0.16% |
| WDC | long | 1 | $519.40 | $551.00 | $551.00 | $31.60 | 6.08% |
Things we killed
10 retractionsStrategies that looked promising in a single window, then failed the paired multi-seed gate or Hansen-SPA test. Each one is committed to git under the SHA below — verifiable, not a marketing list.
- SEC 8-K LLM nowcasting on SP5002026-07-07
TBDRetraction-first: gate scheduled 2026-07-14, verdict pendingGatepaired multi-seed on yfinance deep panel (test_days=21, top_k=20, LLM classifies 8-K filings via OpenRouter DeepSeek V4 Flash, min_confidence=0.7, lookback=5 trading days, hold=20 trading days)ResultWhy - Google Trends timing (pytrends interest spikes)2026-07-07
TBDRetraction-first: gate scheduled 2026-07-14, verdict pendingGatepaired multi-seed on yfinance deep panel (test_days=21, top_k=20, z_threshold=2.0 over 30-day rolling window)ResultWhy - Chronos-Bolt-Base zero-shot as full replacement for momentum-1262026-07-07
TBDRetraction-first: gate scheduled 2026-07-14, verdict pendingGatepaired multi-seed on yfinance deep panel (test_days=21, top_k=20, Chronos-Bolt-Base zero-shot forecast, 128d context / 20d horizon)ResultWhy - FinRL portfolio PPO (multi-seed)2026-07-07
TBDGate pending — Horikawa-Nakagawa 2024 predicts failure at ~85-90% probabilityGatepaired multi-seed via scripts/challenger/finrl_significance.sh (env + PPO + 4-sub-test honest gate)ResultNot run yet. Sidecar shipped per honesty contract; result auto-appended when the gate runs.WhyHorikawa-Nakagawa (2024) shows portfolio-RL 'edge' is stat-arb overlay not risk reduction. Expected gate probability ~10-15%. Ship anyway; publish the retraction if it fails. - Sector momentum on 11 SPDR ETFs2026-06-30
TBDFailed paired multi-seed gate vs SP500 momentum-126 2026-06-30Gatepaired multi-seed on yfinance 18.5-yr panel (test_days=21, top_k=5, 11-ETF universe)Result0/3 seeds win, pooled p(1-sided) = 1.0000 (extremely WRONG direction); sector Sharpe 0.717 (fixed — 11 ETFs no seed variation) vs SP500 momentum 0.94/1.29/0.90 — sector WORSE by 0.2-0.6 Sharpe per seedWhyAudit suggested sector momentum could lift +2-5pp annualized by reducing multiple-testing burden (11 ETFs vs 500 stocks → smaller DSR penalty). Reality: the 11-ETF universe is too narrow — top-5 picks have too little dispersion vs SP500's 500-name selection. The fewer-multiple-tests advantage doesn't compensate for the lost cross-sectional dispersion. Signal class kept for future ensemble experiments at higher top_k or with mean-reversion overlays. - Post-Earnings Announcement Drift (Bernard-Thomas)2026-06-25
TBDFailed paired multi-seed gate vs momentum-126 baseline 2026-06-25Gatepaired multi-seed on yfinance deep panel (test_days=21, top_k=20, 12.5 yrs data, 23k earnings events)Result0/3 seeds win, pooled p(1-sided) = 0.9999 (WRONG direction, extremely significant); PEAD Sharpe 0.58/0.65/0.51 vs momentum 0.94/1.29/0.90 — PEAD WORSE per seed by 0.36-0.64WhyBall-Brown 1968 → Bernard-Thomas 1989 found PEAD lasts 30-60 days. The literature documented decay since 2010 and our 2014-2026 yfinance result confirms: at top-k=20 long-only with surprise+day0-confirmed filter, PEAD picks too few names too rarely vs an always-on momentum-126. The signal class (`PEADSignal`) stays in code so future variants (top-k=5, multi-horizon, short-side) can re-test. - Insider cluster-buying (SEC Form 4 P-codes)2026-06-25
TBDFailed paired multi-seed gate vs momentum-126 baseline 2026-06-25Gatepaired multi-seed on yfinance deep panel (test_days=21, top_k=20, 8.5 yrs data, 586k insider tx)Result0/3 seeds win, pooled p(1-sided) = 0.9395 (WRONG direction); insider Sharpe 0.83/0.41/0.47 vs momentum 1.40/1.25/1.13 — insider WORSE per seed by 0.4-0.8WhyCohen-Malloy-Pomorski (2012) found real edge in cluster-buy signals in small/mid-cap names; in our SP500-only large-cap window the signal fires too rarely to compete with the always-on momentum signal. Real edge may exist with a wider universe or different threshold — wire stays in `signals.py` so future tuning can re-test. - Storm shield (SMA200 trend filter)2026-06-24
1284ee0Disabled on the live path 2026-06-24Gatepaired multi-seed on yfinance deep panelResult0/3 seeds win, pooled p(1-sided) = 0.9935 (WRONG DIRECTION); shield reduced Sharpe by ~0.33 each seedWhyA trend-following overlay should defend against bear-regime drawdown. On the corrected dataset it neither lifts Sharpe nor reduces drawdown materially. Wire stays in code (env-gated) so a future tuning can re-evaluate without re-implementing it. - LightGBM V2 (tuned Optuna 30-trial)2026-05-22
e3d85fcRolled back to V1 after failing vs V1 on multi-seedGatepaired multi-seed: V2 vs V1 baselineResultV2 lost 0/3 seeds; Optuna's in-sample tuning didn't generalise to held-out foldsWhyHonest gate caught the selection-bias from a 30-trial hyperparameter search before live capital touched it. - FRED macro features2026-05-19
1137696Retracted after failing paired multi-seed robustnessGatepaired multi-seed (macro vs no-macro across seeds 42 / 7 / 99)ResultPooled p = 0.81, macro wins 1/3 seeds — single-window +43% Sharpe was selection biasWhyA single backtest window's improvement looked huge. The multi-seed gate caught that the +43% was a happy seed, not a real edge.
Important: practice only
This is a practice account. The dollar amounts shown are fake money — real prices, fake balance, no real dollars at risk. The brain picking these stocks is a math program trained on past market patterns. It is not magic. Past performance does not promise anything about the future.