Hello,
Recently I recognized a phenomenon, about which I would like to ask. Example:
Monday Evening: I update the data and run my strategy. Result: No signals. So, I don't create any orders.
Tuesday Evening: I update the data and run my strategy. Result: Beside "buying" / "shorting" signals, I also get a "cover" for a 1 day strategy (I should have shorted tuesday morning and now get the signal to cover wednesday morning).
But: There was no "Short" signal showing up on monday evening after running the strategy!
So, I check again: I run the strategy again for Tuesday morning. And this time the "Short"-Signal shows up. However, this signal was not showing up on Monday evening.
My interpretation is, that the data for monday (or the trading days before) were changed after monday evening. I don't know if that is true.
Since I had this already several times, I want to ask, If I am right with my assumption and if there is a way to avoid this problem? In the end, it can change the calculated return of my strategy significantly.
Kind regards
Werner
Recently I recognized a phenomenon, about which I would like to ask. Example:
Monday Evening: I update the data and run my strategy. Result: No signals. So, I don't create any orders.
Tuesday Evening: I update the data and run my strategy. Result: Beside "buying" / "shorting" signals, I also get a "cover" for a 1 day strategy (I should have shorted tuesday morning and now get the signal to cover wednesday morning).
But: There was no "Short" signal showing up on monday evening after running the strategy!
So, I check again: I run the strategy again for Tuesday morning. And this time the "Short"-Signal shows up. However, this signal was not showing up on Monday evening.
My interpretation is, that the data for monday (or the trading days before) were changed after monday evening. I don't know if that is true.
Since I had this already several times, I want to ask, If I am right with my assumption and if there is a way to avoid this problem? In the end, it can change the calculated return of my strategy significantly.
Kind regards
Werner
Rename
It's likely the Transaction.Weight (a.k.a. Position Priority) issue.
(We're going to have to post a detailed article on this so we can point to just one explanation.)
So, let's start somewhere.
1. What kind of orders are you placing?
2. Are you assigning Transaction.Weight?
(We're going to have to post a detailed article on this so we can point to just one explanation.)
So, let's start somewhere.
1. What kind of orders are you placing?
2. Are you assigning Transaction.Weight?
Thanks for your help!
With this strategy I am only placing market orders.
Yes, I use Transaction Weight.
Werner
With this strategy I am only placing market orders.
Yes, I use Transaction Weight.
Werner
If you look at the Metrics Report > NSF Positions, how many missed trades does your system have?
T.Weight should be deterministic (and unique) so that it doesn't change from run to run. If the T.Weights aren't unique, you risk a random result, and that's what we need to get to the bottom of.
For example, an RSI calculation at double precision for 2 series that are not identical guarantees a unique result, for all practical purposes.
How are you assigning T.Weight?
For example, an RSI calculation at double precision for 2 series that are not identical guarantees a unique result, for all practical purposes.
How are you assigning T.Weight?
That used to happen to me in WL6. I'm waiting for build 8 before I go live so don't know if it's happening with WL7.
I always just wrote it off to "corrected" data.
I always just wrote it off to "corrected" data.
For now, you have to reload data in WL7 to correct it, but, you can also get data from different providers, and the provider can change based on the order you stack them in Data Manager > Historical Providers.
Generally the closing price for all EOD Providers is the same; the official settled close. If you're only using the close in your rules for a Market order system with T.Weights, I wouldn't expect variability in the results even between providers - assuming no one is missing bars!
Generally the closing price for all EOD Providers is the same; the official settled close. If you're only using the close in your rules for a Market order system with T.Weights, I wouldn't expect variability in the results even between providers - assuming no one is missing bars!
Your Response
Post
Edit Post
Login is required