More Decimal Places

Here are the key bugs and issues found in your code regarding instrument splits and ATR stop-loss logic:

1. Data Availability and Garbage Historical Prices

• You set your backtest start date to 1997, but TQQQ did not exist before February 2010. Before its IPO, QuantConnect fills older history with tiny “dummy” prices (like $0.01), causing your ATR and stop calculations to generate meaningless values that will always trigger your stops.
• Recommendation: Start your backtest at 2010 or later to avoid garbage pre-inception data.

2. Data Normalization Mode

• By default, QuantConnect uses ADJUSTED data with splits/dividends applied, which makes historical prices appear tiny after many splits, further magnifying your ATR and stop issues.
• Recommendation: Explicitly set the data normalization mode to RAW:

  self._tqqq = self.add_equity("TQQQ", Resolution.MINUTE, data_normalization_mode=DataNormalizationMode.RAW).symbol
  

  This ensures actual traded prices are used (no split/back-adjustment).

3. ATR and Stop Calculation Magnitude

• Your ATR (and thus stop-loss distance) is based on highly split-adjusted (tiny) prices in the past, so ATR * 0.05 yields extremely small numbers, causing near-instant stop-outs.
• Even after fixing the timeline and normalization, using a percent of ATR as stop may not be robust for leveraged/splitting ETFs unless you inspect the scale in modern prices.

4. Symbol Availability & Consolidator

• The consolidator and symbol object logic are fine as long as your timeline & data source exist. However, with a pre-IPO start date, it continuously works with empty or nonsense data, never trading correctly.

5. ATR/Price Logging for Debugging

• You should log ATR and price values at each entry, especially around split dates to ensure the true scale matches the traded price.

  self.debug(f"{self.time} Price: {price} ATR: {self._atr.current.value} StopDist: {stop_distance}")
  

6. Watch for Price/ATR Data Dropouts

• If price or ATR is exceptionally small or zero (e.g., price < 1 or atr < 0.01), skip trading and log a warning. This signals data corruption, missing history, or a big recent split.

Summary of Required Fixes

• Use a start date after 2010.
• Set data_normalization_mode=DataNormalizationMode.RAW when adding TQQQ.
• Add ATR/stop value logging before trading decisions.
• Warn and skip days when price or ATR is implausibly small.
• Consider hard-coded date or minimum price threshold to detect/prevent bogus pre-IPO trades.

After these changes, your ATR/price logic will be far more robust and stop-loss logic should perform as expected.

Sources: Trailing Stop Orders - QuantConnect.com, Sparse Optimization - QuantConnect.com, Data Preparation - QuantConnect.com