I want to backtest a couple of different futures markets in the PORTFOLIO TRADER and encode for each market a separate commission and slippage assumption.
However in Portfolio Trader I can only encode that on the startegy level rather than the Instrument level. How can I encode on an Instrument level ? It makes no sense to assume the same commission and slippage for all markets .
Slippage settings in Portfolio Trader
Re: Slippage settings in Portfolio Trader
I don't think there is the kind of setting you are seeking. However, if symbol specific commissions and slippage are important, why not simply view them as a hurdle that your strategy has to meet?
Re: Slippage settings in Portfolio Trader
Adding on to Orion's comments, I think slippage is often misunderstood. It was by me until I went from backtesting to live testing. Slippage is thought of by many as a fast moving market and I only fill maybe half of my position where I want, the rest is filled adversely against me, that's the theoretical definition of slippage I suppose.
The reality is, and not a knock on MC, there's no way around it, if you plan on using limit orders, MC will fill you in backtesting, but in the real world, you'll likely not going to get filled much at all UNLESS price trades through your limit. So to truly match backtests results in terms of trade frequency (after all if you don't trade as frequent, your real and back tests won't match), you need to trade live with market orders. Which then to add to Orion's comment would say, assume that to make money live your backtests should be able to absorb one or two ticks (in the case of futures) of "expense." Expense being the cost of filling at market versus limit.
When I first started with automated systems, I was so excited about how great my backtests were. I saw trade frequency, win rates, etc and mistakenly thought I would trade just as much, but might have a little "slippage" but that's all. The reality is, when I went live, I failed to enter trades that back tests / or replay testing said I should have filled, and that's because price never traded through the limit whereas with MC a touch of price fills.
Now you can I believe, change backtest criteria so limits fill when price trades through, but again, that's not always accurate 100% either.
The reality is, and not a knock on MC, there's no way around it, if you plan on using limit orders, MC will fill you in backtesting, but in the real world, you'll likely not going to get filled much at all UNLESS price trades through your limit. So to truly match backtests results in terms of trade frequency (after all if you don't trade as frequent, your real and back tests won't match), you need to trade live with market orders. Which then to add to Orion's comment would say, assume that to make money live your backtests should be able to absorb one or two ticks (in the case of futures) of "expense." Expense being the cost of filling at market versus limit.
When I first started with automated systems, I was so excited about how great my backtests were. I saw trade frequency, win rates, etc and mistakenly thought I would trade just as much, but might have a little "slippage" but that's all. The reality is, when I went live, I failed to enter trades that back tests / or replay testing said I should have filled, and that's because price never traded through the limit whereas with MC a touch of price fills.
Now you can I believe, change backtest criteria so limits fill when price trades through, but again, that's not always accurate 100% either.
Re: Slippage settings in Portfolio Trader
You are right , when MC gets filled on a limit order you might incur slippage in real trading. In order to factor that in, I always use very high commission (includes my average slippage) rates in the historical backtests. However each market has different slippage amounts . This is due to tick value in futures. Each market has a different one. So in Portfolio Trader of Multicharts you cannot encode different slippages/commissions per market but only per system. That to me makes absolutely no sense . Why would anybody backtest a portfolio of several futures markets with the same slippage amount for each market ?
Would be nice if we could have market specific slippage/commission settings in Portfolio Trader.
Would be nice if we could have market specific slippage/commission settings in Portfolio Trader.
Re: Slippage settings in Portfolio Trader
I believe slippage should be configurable in the QuoteManager per instrument and/or modifiable/editable from the code.
This clearly is a FLAW which leads to unrealistic calculation results, especially for Futures...
Please VOTE so this feature be included in the next versions of MultiCharts.
http://www.multicharts.com/pm/viewissue ... no=MC-1708
http://www.multicharts.com/pm/viewissue ... no=MC-1657
Thanks
This clearly is a FLAW which leads to unrealistic calculation results, especially for Futures...
Please VOTE so this feature be included in the next versions of MultiCharts.
http://www.multicharts.com/pm/viewissue ... no=MC-1708
http://www.multicharts.com/pm/viewissue ... no=MC-1657
Thanks
Re: Slippage settings in Portfolio Trader
This is a feature rather than a flaw. Slippage needs to be a function of liquidity and not instrument. There are many models of liquidity. MC places no restrictions on how you choose to model liquidity and consequently the slippage in your backtest.I believe slippage should be configurable in the QuoteManager per instrument and/or modifiable/editable from the code.
This clearly is a FLAW which leads to unrealistic calculation results, especially for Futures...
Re: Slippage settings in Portfolio Trader
Slippage not only depends on liquidity!!! There are huge differences of "slippage" between brokers for instance...This is a feature rather than a flaw. Slippage needs to be a function of liquidity and not instrument. There are many models of liquidity. MC places no restrictions on how you choose to model liquidity and consequently the slippage in your backtest.
I still believe that slippage shouldn't be configured in the strategy parameters, this is ridiculous, but per instrument (or group of instruments), especially for futures.
Re: Slippage settings in Portfolio Trader
Slippage being a function of liquidity applies at the exchange. That is how institutions, who have direct access to the exchange and are only limited by the speed of light, model slippage. Clearly, for a retail trader, slippage will also depend on the broker.
The point I was making is that slippage is dynamic and time varying and static configuration in MC based on instrument or groups of instruments will not be of much help if your desire is accuracy. On the other hand, MC gives you all the freedom to model slippage to as much accuracy as you desire in your backtest.
The point I was making is that slippage is dynamic and time varying and static configuration in MC based on instrument or groups of instruments will not be of much help if your desire is accuracy. On the other hand, MC gives you all the freedom to model slippage to as much accuracy as you desire in your backtest.
Re: Slippage settings in Portfolio Trader
"to model" = to write additional code in EVERY strategy... in addition to the slippage parameter in the strategy settings!?MC gives you all the freedom to model slippage to as much accuracy as you desire in your backtest.
This isn't an easy/convenient task for most users.
This is why I said that slippage should be configurable in the QuoteManager per instrument and/or modifiable/editable from the code...
-
- Posts: 83
- Joined: 18 Mar 2013
- Has thanked: 11 times
Re: Slippage settings in Portfolio Trader
In many ways, slippage is COMPARABLE TO MARGIN which is defined in QuoteManager.
Slippage should be removed from the strategy hard-coded parameters.
It makes NO SENSE to assume the same slippage for all markets!
Concerning COMMISSIONS, tell me what you see:
https://www.interactivebrokers.com/en/index.php?f=1590
https://www.TS.com/products/p ... ns-pricing
YES! RIGHT! Commissions are defined per product, once again, there's no reason that they should be strategy hard-coded parameters!!
They also should be configurable in the QuoteManager per instrument (or category of instruments) and/or modifiable/editable from the code...
Slippage should be removed from the strategy hard-coded parameters.
It makes NO SENSE to assume the same slippage for all markets!
Concerning COMMISSIONS, tell me what you see:
https://www.interactivebrokers.com/en/index.php?f=1590
https://www.TS.com/products/p ... ns-pricing
YES! RIGHT! Commissions are defined per product, once again, there's no reason that they should be strategy hard-coded parameters!!
They also should be configurable in the QuoteManager per instrument (or category of instruments) and/or modifiable/editable from the code...