- ago
Strategy Monitor has become a very effective tool for auto-trade of intraday strategies.
However it appears SM picks up data from ext hours when collecting its initial backtest positions. This results in unexpected orders issued in the early minutes of the US regular hours - different than had only regular hours data been taken into account by the initial backtest.

SM does not have a box to specifically filter out ext. hours, however it does have a "Market:" selection for US Stocks - which has a defined set of trading hours. Hence the discrepancy and expectation for only regular US Stocks trading hours to be used.
0
466
Solved
8 Replies

Reply

Bookmark

Sort
Cone8
 ( 24.56% )
- ago
#1
The odd thing is that my experience is completely the opposite. I cannot pre/post data even when I try by disabling the pre/post filter, saving, and adding to the Strategy Monitor.

What verification do you have to show that the Strategy Monitor is picking up ext hours?
0
- ago
#2
verification in this case is two early minutes post-opening manual backtest run of the strategy stored in SM; with and without ext hours filter set.
specifically the strategy will:
1 - by design, avoid overnight positions by limiting last trade times before market close
2- without ext. hours filter set (ext hours enabled) during the manual backtest, positions show an opening during pre-market hours; then closed in the early minutes of regular trading hours. this is inconsistent with US Market time setting for the strategy in SM
3 -with ext. hours filter set (ext hours not enabled), there is no open position before market opening. the first position time of day will differ from (2) as a consequence

point (2) can have different outcomes depending on what trades happened in pre-market times.

Cone - I use TDA intraday pre-market data. IB in my case does not provide API pre/post-market intraday data. Since you seem to use IB in many posts, wonder if you could try to reproduce the issue with TDA
0
Cone8
 ( 24.56% )
- ago
#4
I tested this with TDA, but the provider doesn't make a difference.
"Tested" means:
1. Checking the number of bars loaded for 1 week of 30-minute bars in the S. Monitor Log (65 or less is the right number).
2. Verifying every bar loaded for the strategy by writing them to a file.

The only unknown to me is your strategy.

QUOTE:
IB in my case does not provide API pre/post-market intraday data.
You said "in my case", so I won't argue with this statement. However, IB certainly does provide pre/post market data -

0
Cone8
 ( 24.56% )
- ago
#5
Looked at this again with 30 and 1-minute bars today with a "no filter" setting. The 30-minute bars had no premarket data. The 1-minute bars also had no premarket data for all days prior to "today", but today's history had two extra bars: 0929 and 0930.

We'll keep investigating and fix that.
0
Cone8
 ( 24.56% )
- ago
#6
Working on this one today. It seems to be isolated to TDA 1-Minute Streaming Bars. Following Strategy Activation all of the (TDA) 1-minute Streaming Bars are being added - they're not filtered. So if the market opens at 09:30 and you activate your 1-minute strategy at 09:10, you'll get 20 (or 21) extra bars from 09:10 to 09:30 inclusive.
0
Cone8
 ( 24.56% )
- ago
#7
We'll have a round of build releases this weekend and this is fixed in TDA Build 9.
0
Best Answer
- ago
#8
Appreciate it Cone !
1

Reply

Bookmark

Sort