افتتاح حساب تجاری 
Hi Traders , Members and Friends ..
The word optimization will mean the selection of the most robust set
of parameters for a trading strategy. Furthermore, optimization is the identification
and validation of those strategy parameters that are best capable
of generating peak trading performance in real time.
Notice that the emphasis is on peak performance in real-time trading.
This may seem to be the obvious goal of optimization. It is the goal and result of optimization done correctly. Unfortunately, many practitioners of
optimization are not performing optimization correctly and they are consequently
not able to achieve this goal. Sadly, many users of trading strategy
development software still operate under the delusion that the trading
strategy that shows the biggest profit under optimization is one and
the same with the trading model that will generate peak real-time trading
performance.
Before we proceed, consider the following mathematical definition of
parameter from the New Oxford American Dictionary: “A quantity whose
value is selected for the particular circumstances and in relation to which
other variable quantities may be expressed.”
A trading strategy comprises rules, formulae, indicators, and so on.
Many of these different strategy components will have variables or parameters.
For example, a period of three for a moving average is an instance
of a parameter for an indicator. The rules, formulae, and other
components provide the structure of the strategy. It might be said that
the correct parameters for these components breathes life into the
strategy.
In a practical sense, the optimization process is the calculation of the
historical performance of a number of different instances of the trading
strategy on a fixed historical price sample. For example, referring back to
the MA trading strategy, an optimization of this strategy will entail the calculation
of the historical performance of all of the different combinations
of moving average lengths under examination on a historical sample of yen
price data ranging from 16/2/2001 to 31/3/2016.
The results of each of these historical simulations will vary from one
another because each simulation uses a different set of the model parameter
values that are under optimization, that is, the subject of the test. For
example, the MA strategy will examine the trading results of all possible
combinations of MA1 from 2 to 12 at steps of 1 and for all possible combinations
of MA2 from 20 to 40 at steps of 2. This will consist of 121 individual
simulations, each with a unique pair of parameters for each of the moving
averages.
A top parameter set is then selected from this batch of 121 historical
simulation results, based on a set of evaluation criteria. Assume that the
objective function is the Sharpe Ratio. Then, the parameter pair of moving
average periods that produced the historical simulation with the highest
Sharpe Ratio is selected as the best parameter set.
If this optimization process and selection are both done with the
proper attention to all of the appropriate details and rules, the resulting
top models will be those that offer the greatest potential for real-time trading
profit.
to be continue .....
of parameters for a trading strategy. Furthermore, optimization is the identification
and validation of those strategy parameters that are best capable
of generating peak trading performance in real time.
Notice that the emphasis is on peak performance in real-time trading.
This may seem to be the obvious goal of optimization. It is the goal and result of optimization done correctly. Unfortunately, many practitioners of
optimization are not performing optimization correctly and they are consequently
not able to achieve this goal. Sadly, many users of trading strategy
development software still operate under the delusion that the trading
strategy that shows the biggest profit under optimization is one and
the same with the trading model that will generate peak real-time trading
performance.
Before we proceed, consider the following mathematical definition of
parameter from the New Oxford American Dictionary: “A quantity whose
value is selected for the particular circumstances and in relation to which
other variable quantities may be expressed.”
A trading strategy comprises rules, formulae, indicators, and so on.
Many of these different strategy components will have variables or parameters.
For example, a period of three for a moving average is an instance
of a parameter for an indicator. The rules, formulae, and other
components provide the structure of the strategy. It might be said that
the correct parameters for these components breathes life into the
strategy.
In a practical sense, the optimization process is the calculation of the
historical performance of a number of different instances of the trading
strategy on a fixed historical price sample. For example, referring back to
the MA trading strategy, an optimization of this strategy will entail the calculation
of the historical performance of all of the different combinations
of moving average lengths under examination on a historical sample of yen
price data ranging from 16/2/2001 to 31/3/2016.
The results of each of these historical simulations will vary from one
another because each simulation uses a different set of the model parameter
values that are under optimization, that is, the subject of the test. For
example, the MA strategy will examine the trading results of all possible
combinations of MA1 from 2 to 12 at steps of 1 and for all possible combinations
of MA2 from 20 to 40 at steps of 2. This will consist of 121 individual
simulations, each with a unique pair of parameters for each of the moving
averages.
A top parameter set is then selected from this batch of 121 historical
simulation results, based on a set of evaluation criteria. Assume that the
objective function is the Sharpe Ratio. Then, the parameter pair of moving
average periods that produced the historical simulation with the highest
Sharpe Ratio is selected as the best parameter set.
If this optimization process and selection are both done with the
proper attention to all of the appropriate details and rules, the resulting
top models will be those that offer the greatest potential for real-time trading
profit.
to be continue .....
