I backtested my dataset yesterday morning resulting in two long positions (e.g. "V", "W"). My strategy is programmed in a way that I have to sell the two positions today (since I am holding positions just one day) and engage in new positions today. But today's backtest result is that the two positions engaged yesterday are not shown in the Positions-tab but a different stock ("X") and that the two positions are also not shown in the Signals-list (where they should appear as to be sold today, since I only hold them for one day)! So the backtest results from yesterday are not linked/coherent to the results for today!
Rename
1. Maybe your strategy is programmed in a way that it "loses memory" of your positions.
2. Another possible culprit is using an unstable indicator together with an unsuitably small seed period (StartIndex).
Can't tell more without looking at the actual code. At any rate, make sure that "Retain NSF positions" is enabled.
2. Another possible culprit is using an unstable indicator together with an unsuitably small seed period (StartIndex).
Can't tell more without looking at the actual code. At any rate, make sure that "Retain NSF positions" is enabled.
Yesterday, we alerted to changes in the Wealth-Data S&P 500 list -
https://www.wealth-lab.com/Discussion/Testing-S-amp-P-500-with-Wealth-Data-6925
Our apologies if this affected you, but one of the changes was "UA" (Under Armour). Technically, UA (Class C shares) remain part of the S&P 500, but we discovered that the Class C didn't exist when it was added to the index in April 2014. The Class A shares, UAA, were added at that time, while the 'C' shares were added after the share split in 2016.
Since different share classes and tracking stocks for the same company tend to have a near 100% correlation, we only include 1 stock for the same company in our Wealth-Data lists. If we didn't you would almost certainly be buying the same company more than once, using different stocks.
If this wasn't one of the stocks that was affected, or you weren't using the Wealth-Data S&P 500, it would help to know more details, especially the symbols.
https://www.wealth-lab.com/Discussion/Testing-S-amp-P-500-with-Wealth-Data-6925
Our apologies if this affected you, but one of the changes was "UA" (Under Armour). Technically, UA (Class C shares) remain part of the S&P 500, but we discovered that the Class C didn't exist when it was added to the index in April 2014. The Class A shares, UAA, were added at that time, while the 'C' shares were added after the share split in 2016.
Since different share classes and tracking stocks for the same company tend to have a near 100% correlation, we only include 1 stock for the same company in our Wealth-Data lists. If we didn't you would almost certainly be buying the same company more than once, using different stocks.
If this wasn't one of the stocks that was affected, or you weren't using the Wealth-Data S&P 500, it would help to know more details, especially the symbols.
Retain NSF position is enabled!
How can an indicator be unstable and how can this affect the signals list? I am using an indicator for transaction weight out of the drop down-list (no self-programmed indicator!) and had no problems occuring with that for the last weeks.
How can an indicator be unstable and how can this affect the signals list? I am using an indicator for transaction weight out of the drop down-list (no self-programmed indicator!) and had no problems occuring with that for the last weeks.
QUOTE:
How can an indicator be unstable and how can this affect the signals list?
Let me quote own reply from a past discussion:
https://www.wealth-lab.com/Discussion/Why-did-Strategy-Monitor-not-detect-signals-6085
QUOTE:
Another potential pitfall that can cause the set of trades to change is the usage of progressively-calculated indicators which reduce lag at the price of having their initial values unstable. After they stabilize (usually after 3-4 times their lookback period) they're safe to use. Given the 140-period lookback and 1 year of data it's less likely (1800+ hours in a year, enough to stabilize such indicator). However, here I assume you now get the full year of intraday data from IQFeed and the stocks in your DataSet aren't IPOs.
Hm, ... I am not using intraday data just end-of-day quotes. And the backtesting period starts with 1.1.1987 (up to now).
If you're using that Stock Screener dynamic DataSet (https://www.wealth-lab.com/Discussion/By-Symbol-doesn-t-show-all-symbols-of-a-DataSet-6911), make sure to UNcheck "Keep up-to-date automatically" as this will change the DataSet's composition every day!
https://www.wealth-lab.com/extension/detail/DataExtensions#screenshots
https://www.wealth-lab.com/extension/detail/DataExtensions#screenshots
Where can I uncheck "Keep up-to-date automatically"? Pleas give me a hint!
In addition to being documented in the Help, you will find it on a screenshot if you follow the 2nd link in my post above.
Thanks for the help! However, the problem still remains.
Hi Braunstein,
We don't have enough information to help you. In my Strategies, this doesn't happen. So it makes me believe the problem lies somewhere in your Strategy. If you want, share the Strategy here or email it to us at support@wealth-lab.com. Thanks!
We don't have enough information to help you. In my Strategies, this doesn't happen. So it makes me believe the problem lies somewhere in your Strategy. If you want, share the Strategy here or email it to us at support@wealth-lab.com. Thanks!
Got your strategy's rules. There's in fact the use of an "unstable" indicator but the fact that the problem disappears after choosing another unstable indicator plus the big data loading range make me think it's a change in the data. For example, you could have reloaded it. What data provider are you using here?
Your Response
Post
Edit Post
Login is required