The Accountability Ledger — 2026-09-03
Every signal and bot prediction Stock Smith publishes is frozen the moment it is made — the entry price, profit target, protective stop, and the stated confidence are written once and never edited afterward. We then grade each prediction against what the market actually did and publish the score here. No cherry-picking, no quiet deletions, no hindsight edits.
Signal Scorecard — as of 2026-09-03
Day- and swing-trade signals are frozen at their published price and graded against real market bars: target reached first = hit, stop reached first = miss, neither by the horizon = expired.
Overall hit rate: 16% across 4,001 resolved predictions (832 still open).
| Signal Type | Total | Resolved | Targets Hit | Stops Hit | Expired | Hit Rate |
|---|---|---|---|---|---|---|
| Day Signals | 1462 | 1442 | 122 | 289 | 1031 | 8% |
| Swing Signals | 1546 | 988 | 95 | 380 | 513 | 10% |
| Crypto Day Signals | 1070 | 1053 | 185 | 240 | 628 | 18% |
| Crypto Swing Signals | 755 | 518 | 224 | 260 | 34 | 43% |
Confidence Calibration
A prediction marked 80% confident should hit roughly 80% of the time. This compares what Stock Smith claimed against what actually happened, grouped by confidence band.
| Confidence Band | Predictions | Stated Confidence | Actual Hit Rate |
|---|---|---|---|
| 60–69% | 1538 predictions | 63% | 14% |
| 70–79% | 697 predictions | 74% | 12% |
| 80–89% | 1382 predictions | 82% | 19% |
| 90–100% | 384 predictions | 95% | 15% |
Bot Performance vs the S&P 500
Each simulated trading bot started with $150,000 and trades the live market in real time alongside SPY normalized to the same starting capital.
| Bot | Total Return | Stats |
|---|---|---|
| Quant AI Trader | +3.3% vs SPY +3.6% | 8 closed trades |
| Swing Trader | +5.8% vs SPY +3.6% | Win rate: 69% · Profit factor: 2.11 · 26 closed trades |
| Day Trader | -5.7% vs SPY +3.6% | Win rate: 44% · Profit factor: 0.83 · 342 closed trades |
The real-money Auto Trader is excluded — it trades on each user's private brokerage account and its results are personal, never aggregated or published.
How We Score Every Prediction
The day- and swing-trade signals are frozen at their published price. We then watch the real market bars that follow: if the price reaches the profit target first, it counts as a hit; if it hits the protective stop first, it counts as a miss; if neither happens by the prediction's horizon, it expires. Because the entry is locked at publication and always sits strictly between the stop and the target, a prediction can never be retroactively scored as an instant win.
Confidence Calibration
A prediction we mark 80% confident should reach its target roughly 80% of the time. Our calibration view groups every resolved signal by the confidence band we claimed and compares that claim against the rate it actually hit — an honest check on whether our stated confidence is meaningful or just decoration. Rates only appear once a confidence band has enough resolved predictions to be statistically meaningful; until then we show the raw counts and say so plainly.
Bot Performance vs the S&P 500
Each of the three simulated trading bots — the Quant AI Trader, the Swing Trader, and the Day Trader — started with a $150,000 book and trades the live market in real time. For each bot we publish its equity curve alongside SPY normalized to the same start date and starting capital, so you can see whether the strategy actually beat simply buying and holding the index. Win rate, profit factor, and average hold time appear once enough trades have closed to be meaningful.
The real-money Auto Trader is intentionally excluded from this public ledger. It trades on each user's own private brokerage account, so its results are personal and are never aggregated or published.
Full Transparency for Subscribers
Subscribers can open the complete prediction ledger — every individual prediction, frozen at publication and graded against the market, filterable by type and outcome. The aggregate scorecard, calibration, and bot equity curves on this page are public to anyone.
Past and simulated performance does not guarantee future results. Stock Smith is a research and educational tool, not investment advice. Markets carry risk, including loss of principal.
Create your free account — 100% free to use, cancel the optional Ad-Free membership anytime