FEATURE POLL. Charting, scanning, backtesting, & ATE (1)
- Andrew Kirillov
- Posts: 1589
- Joined: 28 Jul 2005
- Has thanked: 2 times
- Been thanked: 31 times
- Contact:
FEATURE POLL. Charting, scanning, backtesting, & ATE (1)
Charting, scanning, backtesting, & ATE (1)
Last edited by Andrew Kirillov on 19 Jun 2008, edited 1 time in total.
Multiple strategies portfolio backtesting for one symbol
When trading Futures, multiple portfolio strategies backtesing is quite important feature. I guess current portfolio backtesting is not suited in this respect. If I am wrong correct me to show how to do it.
-
- Posts: 11
- Joined: 14 Dec 2007
Hi Andrew,
Could you add a poll for the feature "Accurate Volume and price bars"?
Multicharts users may not be aware of this as they may not be using a separate charting program to compare the vol+price bars.
This is related to the acknowledged issue in a separate thread that volume bars are not accurate on startup/refresh for the connection bar, and the price bars may also display incorrectly (screenshots in separate thread).
This would be important to traders who use Multicharts to trade off the charts, as indicators and signals are built off price+vol, so price+vol is critical to the accuracy of the signals.
e.g. if a signal is based on a 10/20 moving avg crossover and vol must be high during the crossover, then incorrect price bars may affect the accuracy of the next 20 bars for the signal due to the bad bar taken into account during 20MA calculation.
A suggestion would be to use minute data to verify the previous bar after the connection bar was fully built off tick data, as tick data may have been missed during refresh/first load of Multicharts.
Could you add a poll for the feature "Accurate Volume and price bars"?
Multicharts users may not be aware of this as they may not be using a separate charting program to compare the vol+price bars.
This is related to the acknowledged issue in a separate thread that volume bars are not accurate on startup/refresh for the connection bar, and the price bars may also display incorrectly (screenshots in separate thread).
This would be important to traders who use Multicharts to trade off the charts, as indicators and signals are built off price+vol, so price+vol is critical to the accuracy of the signals.
e.g. if a signal is based on a 10/20 moving avg crossover and vol must be high during the crossover, then incorrect price bars may affect the accuracy of the next 20 bars for the signal due to the bad bar taken into account during 20MA calculation.
A suggestion would be to use minute data to verify the previous bar after the connection bar was fully built off tick data, as tick data may have been missed during refresh/first load of Multicharts.
- Marina Pashkova
- Posts: 2758
- Joined: 27 Jul 2007
Re: Multiple strategies portfolio backtesting for one symbol
Could you please be more specific and provide more details regarding what you thinks is not working in portfolio and what you would like to see implemented?When trading Futures, multiple portfolio strategies backtesing is quite important feature. I guess current portfolio backtesting is not suited in this respect. If I am wrong correct me to show how to do it.
Thank you.
Portfolio Trading one symbol with multiple strategies
When trading only one Futures symbol, traders creates multiple workspaces and each workspace has multiple charts. All the charts naturally plots the same symbol but having different resolutions (maybe some could have same resolutions). Then each chart has different strategy (I mean here strategy as in terms of TS2ki's strategy, that is a named set of signals) applied. What they would like to see in portfolio backtesting is how each strategy perfoms and how all the strategies work together to show the overall combined performance.
I am not sure how this can be done in the current Portfolio backtesting program.
I am not sure how this can be done in the current Portfolio backtesting program.
I lack one thing in Portfolio Backtester in relation to futures contracts. Would it be possible in ''Format Settings'' - ''Poperties'' to make a box called ''Contract Leverage'' or similar name.
The box would say ''If the price of the contract goes up / down by $0.1 then your income / loss is multipled by by ... (10 / 25 / 50 / 100 / 250) - each user would type the factor applied to a given contract.'' Different futures contracts have different leverage.
So, the return of the account would be calculated based on the real income / loss generated on a given futures contract. It is my guess (I may be very wrong here because I've been playing with Portfolio Backtester for only 3 days) that the return on account treats all futures contract the same way.
Regards
The box would say ''If the price of the contract goes up / down by $0.1 then your income / loss is multipled by by ... (10 / 25 / 50 / 100 / 250) - each user would type the factor applied to a given contract.'' Different futures contracts have different leverage.
So, the return of the account would be calculated based on the real income / loss generated on a given futures contract. It is my guess (I may be very wrong here because I've been playing with Portfolio Backtester for only 3 days) that the return on account treats all futures contract the same way.
Regards
I found a good name for this box in investopedia.com www.investopedia.com/ask/answers/146.asp
It might be called 'multiplier' after the article:
''Index futures trade with a multiplier that inflates the value of the contract to add leverage to the trade. The multiplier for the Dow is 10, for the Nasdaq it is 100 and it is 250 for the S&P.
For example, if a Dow Jones Index future is trading at 10,000, this means that if an investor purchased one futures contract, it would be worth $100,000. What this really means for the investor is that every one-point change in the Dow will cause a $10 change in real terms for the investor. If the Dow falls 100 points, the holder of the contract on the long side will lose $1,000.''
Regards
If this feature is already in Portfolio Backtester, let me now. I am a rookie in MC, so I might have missed it
It might be called 'multiplier' after the article:
''Index futures trade with a multiplier that inflates the value of the contract to add leverage to the trade. The multiplier for the Dow is 10, for the Nasdaq it is 100 and it is 250 for the S&P.
For example, if a Dow Jones Index future is trading at 10,000, this means that if an investor purchased one futures contract, it would be worth $100,000. What this really means for the investor is that every one-point change in the Dow will cause a $10 change in real terms for the investor. If the Dow falls 100 points, the holder of the contract on the long side will lose $1,000.''
Regards
If this feature is already in Portfolio Backtester, let me now. I am a rookie in MC, so I might have missed it
- Marina Pashkova
- Posts: 2758
- Joined: 27 Jul 2007
Hi Tresor,
What you need to do is change bigpointvalue.
However, we have to admit that we do not yet support a very importand contract calculation mode based on margin trading. Currently, MC can't factor in the possibility that a position can be entered with the number of contracts considerably exceeding the amount of money available - because MC uses narrow stop losses. In the current ersion, it is implied that no leverage is used. But we are going to improve this behavior in future.
What you need to do is change bigpointvalue.
However, we have to admit that we do not yet support a very importand contract calculation mode based on margin trading. Currently, MC can't factor in the possibility that a position can be entered with the number of contracts considerably exceeding the amount of money available - because MC uses narrow stop losses. In the current ersion, it is implied that no leverage is used. But we are going to improve this behavior in future.
Marina,
Thanks for the reply. There is one more issue related to backtesting that scratches my mind. Namely, 3D optimization charts.
Kill me or something worst, but I have problems, due to the complexity of the 3D pictures, to understand the wisdom that is stored in these pictures.
Please look at the chart that I made with 3D optimization charts. I cant't just read it. Would you consider changing this type of 3D into the other one that I also attach? Or adding a new 3D charting type (MS Excel like)?
Regards
Thanks for the reply. There is one more issue related to backtesting that scratches my mind. Namely, 3D optimization charts.
Kill me or something worst, but I have problems, due to the complexity of the 3D pictures, to understand the wisdom that is stored in these pictures.
Please look at the chart that I made with 3D optimization charts. I cant't just read it. Would you consider changing this type of 3D into the other one that I also attach? Or adding a new 3D charting type (MS Excel like)?
Regards
- Marina Pashkova
- Posts: 2758
- Joined: 27 Jul 2007
Hi Tresor,
The 3D optimization chart should not be black. There is a bug that sometimes makes the chart look black when viewed from certain angles.
Normally, it is coloured from green through blue etc to yellow and red - depending on how robust a zone is to the strategy parameters. The green zones are robust, the red zones are highly sensitive to the parameter changes. This colouring pattern helps avoid curve-fitting.
The 3D optimization chart should not be black. There is a bug that sometimes makes the chart look black when viewed from certain angles.
Normally, it is coloured from green through blue etc to yellow and red - depending on how robust a zone is to the strategy parameters. The green zones are robust, the red zones are highly sensitive to the parameter changes. This colouring pattern helps avoid curve-fitting.