I have go live several strategies in strategy monitor, but there are some details where I'm not sure if my understanding is correct.
As I understand that the signals given by the strategy monitor may be different from my actual account situation. For example, on the first day the strategy went live, my account had no positions, but the software's calculation based on the set data range may indicate that I already held some positions.
So I need to enable NSF, so that the software can provide all the signals. Then I need to manually select the signals I need to trade based on the transaction weight and the actual holdings in my account.
I don't know if my understanding is correct. Is this the right way to manage positions?
The second question is that since data range is a relative time, suppose I set the data range as 6 months.
Would it be possible that when backtesting today, a stock meets the conditions and therefore generates a buy signal. So I bought it in my account.
But after some time has passed, on that day, using the data from the past 6 months would not generate any buy signal, so the software would think I did not buy that stock, and therefore would never generate a sell signal?
As I understand that the signals given by the strategy monitor may be different from my actual account situation. For example, on the first day the strategy went live, my account had no positions, but the software's calculation based on the set data range may indicate that I already held some positions.
So I need to enable NSF, so that the software can provide all the signals. Then I need to manually select the signals I need to trade based on the transaction weight and the actual holdings in my account.
I don't know if my understanding is correct. Is this the right way to manage positions?
The second question is that since data range is a relative time, suppose I set the data range as 6 months.
Would it be possible that when backtesting today, a stock meets the conditions and therefore generates a buy signal. So I bought it in my account.
But after some time has passed, on that day, using the data from the past 6 months would not generate any buy signal, so the software would think I did not buy that stock, and therefore would never generate a sell signal?
Rename
QUOTE:No. Disabling NSF doesn't have anything to do with a live account. The backtest still can hold positions that you don't own.
So I need to enable NSF... Is this the right way to manage positions?
I don't fully understand your second scenario, but I don't think you're statement is correct, and the result would depend on the strategy and Portfolio Sync option you select.
If you want the strategy to trade positions you already own, you can enable "Use Live Positions" in the Trading Prefs (F12). Most single-position strategies are compatible with this option.
It will also have the opposite effect to make the strategy look for new entries if you don't own the position an "offline" backtest might already own. See the User Guide for all the details for "Use Live Positions".
It's also fine not to "Use Live Positions". The strategy might try to exit a position held in the backtest, but if you don't own the position, the trade will just error out and not be placed. Use the appropriate Portfolio Sync options in Trading Prefs (F12).
QUOTE:
I don't fully understand your second scenario
Hi Cone, I may not have expressed the scenario I thought of very accurately. I tried to demonstrate it with an example. Suppose my strategy is to buy when the price reaches 5 and sell when it reaches 3. Setting different data ranges will produce different results.
Of course, I can set the Data Range as long as possible.But it still seems that this will happen in some extreme cases.
But it seems can be solvable by using the "Live Position" as you mentioned. I will study it.
You could use a date range with an end date some time far off into the future. That way the start day is always the same.
QUOTE:
... I need to manually select the signals I need to trade based on the transaction weight and the actual holdings in my account.
Well, how are you actually setting the Weights on the Transaction?
By default, WL sets these weights at random, so I'm not so sure it matters if you follow the Transaction weights in this case. Now I do things differently. I have a custom C# class that scans through the last 5 positions held for that stock and computes a Sharpe Ratio for the off-the-Chart signal purchase, which is assigned to the Transaction weight for that off-the-Chart bar. As a result, the weights reflect the recent Sharpe Ratio for that stock on the Signal Pane.
So I do buy signals with the highest weights or recent Sharpe Ratio for individual stocks. But this live trading technique is not necessarily going to follow the backtest simulation. However, if you selected the trading options to have the Backtest sync with the actual stocks owned, WL will attempt to sell those rogue purchases when it's time.
Your Response
Post
Edit Post
Login is required