Handling Splits During Warmup

After a little more testing. . .

• My algo uses options, so the data normalization automatically gets set to raw for the equity.  The warm up data will be raw, as well as data slice afterward will be raw.
• During initialization, if I set the datanormalization to raw explicitly, then History will return raw data.  If I don't set it explicitly, then History call during initialization will return adjusted data.   Any history call afterward will return raw data due to normalization automatically being set above.

Seems like I have the following options:

1. Don't use warmup.  Use History explicitly, with DataNormalization to adjusted prices. 
   • For those cases where the warm up contains the latest split, the indicators should be correct.  The OnData prices will be in raw ‘today’ prices, so this should work for live deployment and recent backtests.  For those backtests of previous dates where both a split occurs during the warm up, and a separate split occurrs anytime after historically, the indicators will not be correct since the relative OnData prices will not be ‘today’ prices.   
2. Set DataNormalization.Raw explicitly during initialization, and use either warm up or History with raw prices.
   • For those cases where the warm up contains a split, we'll need to adjust the indicators for the split. 
   • For those cases where the warm up does not contain a split, the indicators are correct from the beginning.
   • The problem is that the Split data doesn't seem to be exposed either in the slice data during warm up, nor during the History call, so don't know how to adjust the indicators for the split.   Adjusting for the split via OnData after warm up works well - I can reset the indicators and re-warm them up with raw price * factor.

I think I would prefer the second option to keep things consistent.   Is there an api that will allow me to fetch the historical splits?  

Even without this, I am not completely blocked though.  I am left with a work around to parameteratize the initialization: 

• For older backtest scenarios that have at least one split during or after the backtest start and end, don't pick a start date that has a split in the warmup. Just use raw explicitly across the board.
• For newer backtest scenarios, and live deployments, use the adjusted normalization history.  I shouldn't have to worry about a split being in the warmup or history period here. 

Hi Derrick Tan,
We're regretted that the Split data has no historical data call available. Meanwhile, we could explicitly call DataNormalizationMode.SplitAdjusted to warm up the indicator. After forcefully switched into DataNormalizationMode.Raw after warm up, you could access the Split data and adjust accordingly inside the OnData method. Refer to the attached backtest.
Best,
Varad Kabade