Request: Simple example of how to get the current day's (daily resolution) OHLC and Volume values prior to Market close?

I thought maybe a consolidator would allow me to get the current days OHLC and Volume values prior to close, however when I run a backtest I get very unexpected results...

I set StartDate and EndDate to 4/15/21 yet when I run the backtest, my consolidator is being hit on 4/13/21 and 4/14/21  and not on the one day I would expect it to be called (4/15/21)? Further the PreMarketCloseHandler() is reporting a SPY Close of 415.87 on 4/14/21 even though SPY's High didn't reach $414 that day?

Log

PreMarketCloseHandler() called at 2021-04-13 22:46:00, current Close price is 411.45
EveryMarketClose() called.
Spy closed at 415.87 on 4/15/21 at 4:00 PM, current SPY price at 3:43PM is: 411.45
PreMarketCloseHandler() called at 2021-04-14 23:38:00, current Close price is 415.87

Code

Hi Brian,

We have an open GitHub Issue to add this functionality. Subscribe to our progress here.

For the time being, to retrieve a bar that represents the daily data from the open to the current time, we first add a daily consolidator.

We can then access the consolidated bar that's currently being built for the day.

See the attached backtest for reference.

Best,
Derek Melchin

Hi Derek Melchin ,

Thanks for helping me out - I'm excited to get my strategy live.

I have been playing with your code example and while it appears to work on historic days, it is failing when enddate is today's date, for example when running 4-20 to 4-21 earlier this evening:

class DeterminedTanPigeon(QCAlgorithm):

def Initialize(self):
    self.SetStartDate(2021, 4, 20) # Set Start Date
    self.SetEndDate(2021, 4, 21)
    self.SetCash(100000) # Set Strategy Cash
    self.symbol = self.AddEquity("SPY", Resolution.Minute).Symbol
    self.consolidator = TradeBarConsolidator(timedelta(1))
    self.SubscriptionManager.AddConsolidator(self.symbol, self.consolidator)

def OnData(self, data):

    if not data.Bars.ContainsKey("SPY"):
        return

    if data.Time.hour == 15 and data.Time.minute == 59:
        bar = self.consolidator.WorkingBar
        #self.Debug(str(bar))
        self.Log("Bar close {} high {}".format(bar.Close,bar.High))

Output:

2021-04-20 15:59:00 :Bar close 412.455 high 415.21

2021-04-21 15:59:00 :Bar close 411.8 high 411.81

The 4/20 values are correct, but note that it reports a Close of $411.80 on 4/21, whereas SPY actually closed at $416.07.

Derek Melchin I forward tested your method by deploying it live.

This is what it reported for SPY today at 3:59PM according to the logs:

2021-04-22 19:59:18 Bar close 412.33 high 413.43

Note the Close is accurate but the High is off, the SPY's High today was $416.78.

Hi Brian,

We just ran the algorithm again and the logs showed

The attached notebook shows that on 4/21/21, SPY closed at 416.07. The logs printed 415.98 because that's the latest price we received on SPY at 4/21/21 3:59pm. Lean does not allow look-ahead bias, so it's not possible to know the price of SPY at 4/21/21 4pm when it's only 4/21/21 3:59pm.

See the attached backtest and notebook for reference. Also consider reviewing our documentation on Understanding Time.

In regards to the live deployment of this algorithm, the high price can be off if we start the algorithm after the market open. When was the algorithm deployed? For further assistance, please attach the algorithm that was deployed.

Best,
Derek Melchin

 Derek Melchin -

"... so it's not possible to know the price of SPY at 4/21/21 4pm when it's only 4/21/21 3:59pm."

Yes - agreed, not confused on that, apologize if I was unclear.

What I was trying to say is that there may be a potential bug when the backtest EndDate = the date that you are running the script.

I ran (StartDate 4/20/21 EndDate: 4/21/21) on 4/21/21.

You ran (StartDate 4/20/21 EndDate: 4/21/21) on 4/22/21.

I wonder if you re-ran yours today (4/22/21) with StartDate 4/21/21 and EndDate 4/22/21 if the results would still look correct? 

"In regards to the live deployment of this algorithm, the high price can be off if we start the algorithm after the market open. When was the algorithm deployed?"

You are right - I deployed it midday this must have been the issue. 

Hi Brian,

Thanks for clarifying. Yes, the results would still look correct. However, the data for 4/22/21 wouldn't be available for backtesting until after the market close on 4/22/21.

Best,
Derek Melchin