$25 Account
Every action, every backtest, every number. Generated 2026-11-08 02:59 UTC from the engine's own logs. Nothing on this page is typed by hand.
Current state
Mode observe · Cash $0.00 · Positions 0 · Trades today 0
Broker-side trade approval required: False · Equity session open: False · Last tick 2026-11-08T02:58:53+00:00
Trades
No orders placed. The engine is running in observe mode and has not been funded yet.
Backtest — 8 rule families, 10 instruments, 6 years of daily bars
Signal computed on bar t executes at the open of bar t+1. Costs charged at 5 basis points per side. Buy-and-hold over the same window is the comparison, never a zero line.
| Rule | Mean CAGR | Beat B&H | Mean max DD | Trades |
|---|---|---|---|---|
| buy & hold | 22.3% | — | — | 80 |
| SMA 50/200 | 9.5% | 0/10 | 35.0% | 45 |
| Mom120 | 8.8% | 1/10 | 46.1% | 266 |
| Mom60 | 7.9% | 1/10 | 47.2% | 364 |
| SMA 20/50 | 4.2% | 1/10 | 51.3% | 159 |
| RSI2<10 | 2.4% | 1/10 | 53.3% | 2007 |
| Donchian20 | 2.3% | 1/10 | 35.0% | 1065 |
| RSI2<5 | 0.7% | 1/10 | 53.9% | 1900 |
| GapDown6% | 0.1% | 0/10 | 40.0% | 886 |
Mean CAGR is the average across instruments, not a portfolio return. “Beat B&H” counts the instruments where the rule's CAGR exceeded buy-and-hold on the same symbol over the same dates.
Post-earnings drift — the rule was tested and rejected
Real earnings dates and real EPS versus consensus from the broker, real daily bars, 30-trading-day holding period after the report, 5 basis points per side.
| n | Mean | Median | Win rate | |
|---|---|---|---|---|
| after an earnings beat | 34 | +3.10% | -2.56% | 47.1% |
| after an earnings miss | 14 | +9.96% | +7.84% | 85.7% |
The textbook prediction is the opposite of what happened. Buying after a beat produced a negative median and a 47 percent win rate. Buying after a miss did better on this sample. A rule built on post-earnings drift is therefore falsified before it was funded, which is the entire point of testing first.
Sample limits, stated plainly: only the trailing eight quarters are exposed by the API, giving 48 observations across 18 months of a strong bull market in high-beta names. This is enough to reject a rule. It is not enough to establish one.
Candidates considered and not placed
None yet.
Method and limits
- Backtests use the broker's own historical bars, cached to disk, re-runnable at will.
- Costs are charged explicitly. A zero-cost backtest is a fictional backtest.
- No lookahead: a signal formed on one bar is executed on the next bar's open.
- Risk caps are enforced in code, not in instructions: one position, full size, no leverage, no short options, notional never exceeds settled cash.
- Past results on 6 years of a historic bull market are not a forecast. The sample is one regime.
Account state read 2026-11-07. This page is generated by build_site.py from the artifacts beside it.