Keep three records separate
| Record | What it shows | What it does not prove |
|---|---|---|
| Current-rule backtest | Today's rules applied to past prices | That the signal was published at the time |
| Saved signal board | The candidates saved for that board date | That anyone placed those trades |
| Trade journal | Entries and outcomes under the journal's rules | Your broker's actual fills or account return |
How PatternLedger measures a result
History starts from 1 January 2010, or the ticker's later first available trading date. Some patterns need extra history before they can qualify. The current-rule study enters at the next market-session open after a signal.
The displayed exits use 1, 3, 5, or 10 completed periods. Daily periods are market sessions. Weekly periods are weeks. Positive returns mean profit for the selected Buy or Sell direction. These backtest returns exclude fees and slippage, which is the difference between an assumed price and an actual fill.
Backtests and AI Journal use different rules
Backtests measure individual patterns over fixed holding periods. AI Journal trades use AI selection and separate exit rules.
For example, a 10-session Daily backtest measures the return after 10 trading days. An AI Journal trade can close earlier or later.
Win rate is not the same as profit
Consider ten made-up trades: six gain 1% each and four lose 2% each. The win rate is 60%, but the simple average return is minus 0.2% before costs: (6 × 1% − 4 × 2%) ÷ 10.
This example is not a compounded portfolio result. It shows why you need both the frequency and size of wins and losses. An attractive win rate alone is not enough.
Check what is missing
- Check the number of completed outcomes. A few results give limited evidence.
- Pending outcomes are not completed trades. Missing values are not zero returns.
- Keep pattern, direction, quality, timeframe, holding period, and MA100 or MA200 study separate.
- The current ticker universe can omit failed or delisted securities. This can bias the results.
- Past prices and assumed fills cannot reproduce every real trading condition.
Costs are excluded
PatternLedger's backtest and Journal returns exclude all trading costs, including fees and slippage. This applies to every AI trade and personal tracked trade. No entry or exit cost is deducted from the displayed return.
Actual fees depend on the broker, instrument, and market. The SEC's investor guide explains why costs reduce investment returns.
Where Pro members can download evidence
Open Backtests, choose the market, timeframe, and family, then use the summary CSV button. A ticker's Backtest History has a trade CSV button. CSV is a table file that opens in a spreadsheet.
Keep the exported filters and holding period with your notes. A download is a research record, not proof of future profit.