New posts

Reply to thread

Message

Further update
[USER=82023]@Linus van Pelt[/USER] thank you for your ongoing interest. The issue that I raised was that Friday's backtest buy signals are accurate but the "additional information displayed" (shares & price) is irrelevant as it uses the previous open for its calculations. On a weekly system that's the previous Mondays open. (one week ago)

Amibroker uses the previous open for its calculations
After "Mondays open" Amibroker updates this additional information but not the buy signals. Why? because backtesting uses the most current open, the recent open is used to calculate the additional information (shares & price) being the number of shares to buy using the simple calculation: Investment $10k / share price (the opening price) equals the number of shares purchased. (after the fact)

Recalculations of the additional information of a buy signal
Amibroker uses one formula for this additional (number of shares & the buy price) before Mondays open & another after Mondays open. With time-shifting of time-related data, the information displayed becomes irrelevant & unusable information before the open. After Mondays open & with the positions already in the markets the additional information is also inaccurate as those calculations can only be completed when the opening price is known. All I'm saying is it's easy to come to a conclusion of the number of shares to buy after the opening price has been established.

Non-trading days
Amibrokers backtest report is a stand-alone portfolio manager. The "Portfolio Backtest Manager" buys the required number of position (in this case 10 positions) & stops generating new buy signals until one of those positions are sold. If the portfolio manager reports a buy that for one reason or another was not physically taken there is no way to delete that position. (that's my dilemma). Using the Exploration Analysis it displays all raw signals. (something that can be manipulated)

Using padding or "Buy on Monday array" isn't a fix
Whether you use Norgate's padding (that buys & recalculates the additional data using on the most recent bar with data, being Wednesday the 27th opening price) or if you stay with using the padding feature of Amibrokers it simply uses the previous Mondays data. Both of which become irrelevant when the position is not purchased). My buy condition is good only for Monday, meaning if my conditional order is not taken - the buy is purged at the end of Monday's trading.

In summary
Amibroker includes the position as a buy (using either padding system) whereas in reality the position was never purchased in the first place. Amibroker doesn't allow adjustments to be made to the Portfolio Manager to accurately display only the physical positions in the portfolio. Using my custom Exploration Analysis solves this dilemma (for me) as those signals can be manipulated.

Skate.

<blockquote data-quote="Skate" data-source="post: 1144442" data-attributes="member: 57364">Further update[USER=82023]@Linus van Pelt[/USER] thank you for your ongoing interest. The issue that I raised was that Friday&#039;s backtest buy signals are accurate but the &quot;additional information displayed&quot; (shares &amp; price) is irrelevant as it uses the previous open for its calculations. On a weekly system that&#039;s the previous Mondays open. (one week ago)Amibroker uses the previous open for its calculationsAfter &quot;Mondays open&quot; Amibroker updates this additional information but not the buy signals. Why? because backtesting uses the most current open, the recent open is used to calculate the additional information (shares &amp; price) being the number of shares to buy using the simple calculation: Investment $10k / share price (the opening price) equals the number of shares purchased. (after the fact)Recalculations of the additional information of a buy signalAmibroker uses one formula for this additional (number of shares &amp; the buy price) before Mondays open &amp; another after Mondays open. With time-shifting of time-related data, the information displayed becomes irrelevant &amp; unusable information before the open. After Mondays open &amp; with the positions already in the markets the additional information is also inaccurate as those calculations can only be completed when the opening price is known. All I&#039;m saying is it&#039;s easy to come to a conclusion of the number of shares to buy after the opening price has been established.Non-trading daysAmibrokers backtest report is a stand-alone portfolio manager. The &quot;Portfolio Backtest Manager&quot; buys the required number of position (in this case 10 positions) &amp; stops generating new buy signals until one of those positions are sold. If the portfolio manager reports a buy that for one reason or another was not physically taken there is no way to delete that position. (that&#039;s my dilemma). Using the Exploration Analysis it displays all raw signals. (something that can be manipulated)Using padding or &quot;Buy on Monday array&quot; isn&#039;t a fixWhether you use Norgate&#039;s padding (that buys &amp; recalculates the additional data using on the most recent bar with data, being Wednesday the 27th opening price) or if you stay with using the padding feature of Amibrokers it simply uses the previous Mondays data. Both of which become irrelevant when the position is not purchased). My buy condition is good only for Monday, meaning if my conditional order is not taken - the buy is purged at the end of Monday&#039;s trading.In summaryAmibroker includes the position as a buy (using either padding system) whereas in reality the position was never purchased in the first place. Amibroker doesn&#039;t allow adjustments to be made to the Portfolio Manager to accurately display only the physical positions in the portfolio. Using my custom Exploration Analysis solves this dilemma (for me) as those signals can be manipulated.Skate.</blockquote>

[QUOTE="Skate, post: 1144442, member: 57364"] [B]Further update[/B] [USER=82023]@Linus van Pelt[/USER] thank you for your ongoing interest. The issue that I raised was that Friday's backtest buy signals are accurate but the "additional information displayed" (shares & price) is irrelevant as it uses the previous open for its calculations. On a weekly system that's the previous Mondays open. (one week ago) [B]Amibroker uses the previous open for its calculations[/B] After "Mondays open" Amibroker updates this additional information but not the buy signals. Why? because backtesting uses the most current open, the recent open is used to calculate the additional information (shares & price) being the number of shares to buy using the simple calculation: Investment $10k / share price (the opening price) equals the number of shares purchased. (after the fact) [B]Recalculations of the additional information of a buy signal[/B] Amibroker uses one formula for this additional (number of shares & the buy price) before Mondays open & another after Mondays open. With time-shifting of time-related data, the information displayed becomes irrelevant & unusable information before the open. After Mondays open & with the positions already in the markets the additional information is also inaccurate as those calculations can only be completed when the opening price is known. All I'm saying is it's easy to come to a conclusion of the number of shares to buy after the opening price has been established. [B]Non-trading days[/B] Amibrokers backtest report is a stand-alone portfolio manager. The "Portfolio Backtest Manager" buys the required number of position (in this case 10 positions) & stops generating new buy signals until one of those positions are sold. If the portfolio manager reports a buy that for one reason or another was not physically taken there is no way to delete that position. (that's my dilemma). Using the Exploration Analysis it displays all raw signals. (something that can be manipulated) [B]Using padding or "Buy on Monday array" isn't a fix[/B] Whether you use Norgate's padding (that buys & recalculates the additional data using on the most recent bar with data, being Wednesday the 27th opening price) or if you stay with using the padding feature of Amibrokers it simply uses the previous Mondays data. Both of which become irrelevant when the position is not purchased). My buy condition is good only for Monday, meaning if my conditional order is not taken - the buy is purged at the end of Monday's trading. [B]In summary[/B] Amibroker includes the position as a buy (using either padding system) whereas in reality the position was never purchased in the first place. Amibroker doesn't allow adjustments to be made to the Portfolio Manager to accurately display only the physical positions in the portfolio. Using my custom Exploration Analysis solves this dilemma (for me) as those signals can be manipulated. Skate. [/QUOTE]

Verification