I think the timezone is whatever your local browser is. Perhaps you are in Moscow?

Hi Simon

Yeah that seems to make sense (currently based in Athens GMT+3).

However does that mean that if you want to schedule in your strategy the execution for a specific time, the time has to be in relation to the local browser time as opposed to the time of the Exchange the asset trades on?
And more importantly if the same code, with scheduled functions to run at specific times, is run by 2 different people that have a different local browser time will it yield a different backtest?

Thanks

Those times are exchange-local times.

Thanks Simon

Does that mean that the backtesting engine is resolving the time-inconsistency internally?
Because if you want to backtest a strategy where the signal is being produced for example at 10am exchange-time, but the timestamp of the data used in the backtest is based on your browser time then you are obviously not backtesting the right thing, unless the data is actually in the local exchange-time but only displays in the logs with a timezone offset.

Yeah, I'm a little confused by your question, but basically, the logs are being printed by your browser. The backtest itself is running on Quantopian's back end, and those times are all EST.

Just wanted to be absolutely sure as i am not very familiar yet with the Quantopian interface.

Having said that it doesn't look like the data is timestamped in EST but rather in UTC format. (see screenshot)

Hence the question remains: If you want to backtest a strategy that is scheduled to perform operations at 10am NY time, do you put in your code 10:00 or do you put 10+4 = 14:00 UTC when on Daylight saving (Summer time) & 10+5=15:00 UTC during Winter?

Here are the results I get when I run your code:

2016-10-07 09:31  PRINT Equity(24 [AAPL])     222496.5  
Equity(8554 [SPY])    469472.8  
dtype: float64  
2016-10-07 09:32  PRINT Equity(24 [AAPL])     236477.0  
Equity(8554 [SPY])    567329.4  
dtype: float64  
2016-10-07 09:33  PRINT Equity(24 [AAPL])     244683.3  
Equity(8554 [SPY])    638179.6  
dtype: float64  
2016-10-07 09:34  PRINT Equity(24 [AAPL])     248601.4  
Equity(8554 [SPY])    630507.5  
dtype: float64  
2016-10-07 09:35  PRINT Equity(24 [AAPL])     260805.9  
Equity(8554 [SPY])    624699.4  

1. that is 10. For some reason it's reformatting "10." as "1.". It might be quicker if you just test these things.
   

Hi Ted,

Can you rerun the attached and see what timestamp you get for the SPY data in relation to the time of your browser?

I am in ET. Here are the results:

2016-10-07 09:31  PRINT                            Equity(24 [AAPL])  Equity(8554 [SPY])  
2016-10-07 13:31:00+00:00           959229.0            934435.0  
2016-10-07 09:32  PRINT                            Equity(24 [AAPL])  Equity(8554 [SPY])  
2016-10-07 13:32:00+00:00           217153.0           1147551.0  
2016-10-07 09:33  PRINT                            Equity(24 [AAPL])  Equity(8554 [SPY])  
2016-10-07 13:33:00+00:00           159739.0            861136.0  
2016-10-07 09:34  PRINT                            Equity(24 [AAPL])  Equity(8554 [SPY])

... skip most of the day, until the close

2016-10-07 15:57  PRINT                            Equity(24 [AAPL])  Equity(8554 [SPY])  
2016-10-07 19:57:00+00:00            89051.0            371664.0  
2016-10-07 15:58  PRINT                            Equity(24 [AAPL])  Equity(8554 [SPY])  
2016-10-07 19:58:00+00:00           126454.0            613276.0  
2016-10-07 15:59  PRINT                            Equity(24 [AAPL])  Equity(8554 [SPY])  
2016-10-07 19:59:00+00:00            92842.0            502498.0  
2016-10-07 16:00  PRINT                            Equity(24 [AAPL])  Equity(8554 [SPY])  
2016-10-07 20:00:00+00:00           499050.0           2508848.0  
2016-10-10 09:31  PRINT                            Equity(24 [AAPL])  Equity(8554 [SPY])  
2016-10-10 13:31:00+00:00          1206017.0            949635.0  
2016-10-10 09:32  PRINT                            Equity(24 [AAPL])  Equity(8554 [SPY])  
2016-10-10 13:32:00+00:00           376144.0            577482.0  
2016-10-10 09:33  PRINT                            Equity(24 [AAPL])  Equity(8554 [SPY])  
2016-10-10 13:33:00+00:00           282627.0            280505.0  
2016-10-10 09:34  PRINT                            Equity(24 [AAPL])  Equity(8554 [SPY])  
2016-10-10 13:34:00+00:00           247877.0            418072.0  
2016-10-10 09:35  PRINT                            Equity(24 [AAPL])  Equity(8554 [SPY])

etc ...  

Got it, thanks a lot Ted. Just wanted to verify that the data is indeed in UTC

Best

Thanks Simon, appreciate your responses

Photi,

Hence the question remains: If you want to backtest a strategy that is scheduled to perform operations at 10am NY time, do you put in your code 10:00 or do you put 10+4 = 14:00 UTC when on Daylight saving (Summer time) & 10+5=15:00 UTC during Winter?

A good way to make sure you are working with proper market timing relative to your time zone is to make use of the datetime library. Copy and paste the code sample below into an algo, run it, and view the print logs. You will be able to tell how market time relates to the time in your browser.

import datetime

def initialize(context):  
    context.security = symbol('AAPL')  
def handle_data(context, data):  
    price = data.current(context.security, 'price')  
    Market_Time = get_datetime('US/Eastern')  
    print Market_Time  
    print price

Now if you want to execute some code at 10:00 AM EST, you can do something like this:

Market_Time = get_datetime('US/Eastern')  
Market_Hour = Market_Time.hour  
Market_Minute = Market_Time.minute

if Market_Hour == 10 and Market_Minute == 0:  
        # Conduct logic for 10AM EST

Hi Frank

Appreciate this.
Thanks

No prob Photi. Happy trading.

You're grappling with some of the challenges we do while building the platform. We work very hard to make the API easy and intuitive, but sometimes it isn't easily resolved.

One of of the nice things about Quantopian's scheduled functions is that they describe events in relation to "market open" and "market close". You don't need to know the time zone, or daylight savings, or hours-offset-from-your-current-local-timezone. You just need to say, "have this happen 10 minutes after market open" or "have this happen 15 minutes before close". It just doesn't matter what the timezone is when you use the API framework.

@Dann
Yeah it became obvious how people here deal with these issues after Frank's comments.

Many thx