Update rolling window for splits and dividends?

I'm interested in this as well as we primarily trade small cap stocks which go through reverse splits quite often.

Doing a bit more testing, I think I answered the question if prices in rolling windows are adjusted for splits and dividends.  At the moment I am not live trading so the observation applies only to backtest.  The answer is that prices are adjusted for splits but not for dividends. 

I used AAPL as a test case and I see prices in my rolling window are adjusted for the 4:1 split that occurred on Aug 31, 2020.  Prices are not adjusted however, for the dividend that occurred on Nov 6, 2020.  I tried all different Data Normalization settings for AAPL including TotalReturn but none of these provide dividend adjusted prices. 

That said the prices in my rolling window are the same as prices using a Historical Data request. Both provide split adjusted prices but neither adjusts for dividends.

import pandas as pd
import numpy as np

class RollingWindowAdjustedPrices(QCAlgorithm):

def Initialize(self):
self.SetStartDate(2020, 8, 28)
self.SetEndDate(2020, 11, 9)
self.SetCash(100000)
EquityObj = self.AddEquity("AAPL", Resolution.Daily)
self.SymbolObj = EquityObj.Symbol
self.Securities[self.SymbolObj].SetDataNormalizationMode(DataNormalizationMode.TotalReturn)

self.rolling_window = pd.DataFrame()
self.window_size = 51
self.SetWarmUp(self.window_size)

def OnData(self, data):
if self.SymbolObj not in data.Bars:
return

close = data[self.SymbolObj].Close

if self.IsWarmingUp:
# Add latest close to rolling window
row = pd.DataFrame({"close": [close]}, index=[data.Time])
self.rolling_window = self.rolling_window.append(row).iloc[-self.window_size:]

return

closes = np.append(self.rolling_window['close'].values, close)[-self.window_size:]
row = pd.DataFrame({"close": [close]}, index=[data.Time])
self.rolling_window = self.rolling_window.append(row).iloc[-self.window_size:]


def OnEndOfAlgorithm(self):
self.Log(f"\nRolling Window:\n{self.rolling_window.to_string()}\n")
self.Log(" ")

histCloses = self.History(self.SymbolObj,self.window_size,Resolution.Daily)['close'].unstack(level=0).dropna()
self.Log(f"\nHistory Window:\n{histCloses.to_string()}\n")

1) Prices in backtesting by default in LEAN are adjusted for splits and dividends;  live mode is always the live/raw prices.

2) However, is it correct that prices in RollingWindows will not automatically be back adjusted for splits and dividends?
-> When you get a split or dividend event you should repopulate the rolling window with a history request.

If you believe there is a missing dividend please post a data issue to the data explorer with a backtest to reproduce the issue. This is the relevant page that shows the latest missing dividend.

Generally please try and avoid making categorical statements that conflict with the documentation. They are generally incorrect, but worse confuse community members who will just scan the post and not understand the nuance.

Wanted to note here that in my case, the issue was actually missing split data for one specific name. As Jared mentioned, it was easy enough to re-populate rolling windows with in response to a spilt event. After notifying support, the issue was quickly fixed.