When backtesting Forex Strategies is there a particular minimum sample size of trades?

Backtesting Forex Strategies when building an Expert Advisor or Algorithmic Trading System is essential.  Backtests with real ticks give the most accurate account of what would have happened in the past should the trades been actually made.  The best part about the Backtesting process is that the only resource used is time, but you can find ways to reduce that time.  However, this article is about the question posed by a Quora User concerning the sample size.  This will be a recurring Thursday article, it will be a highlighted question straight from Quora with an answer delivered directly to the user who has likely forgot about the query in the first place.  Let’s dig into the topic at hand by this Anonymous Quora User.

This is actually a two part question, even though it is intended to be just one question.  The answer comes in two parts and the response may not be what the person who asked the question necessarily had in mind.

The question boils down to the concept of a minimum amount of trades and profitability on a constant-basis (or maybe the word the questioner is looking for is ‘consistently’?).

Backtesting Forex Strategies Requires a Minimum Amount of Trades

It is actually far more complex than just a mere minimum amount of trades when backtesting any strategy.  This is going to be disappointing to read, but it is important to understand that a sheer minimum amount of trades is not an adequate sample size, for there is no such thing as an adequate sample size over a time horizon.  This is not taking a sample of the population in a snapshot period of time for a scientific purpose or using a sample for the purposes of polling popular opinion on a particular subject at a given time.  This is a sample size over the course of days, weeks, months or years.  Conditions change over time, which means that a cut-off on the amount of trades may reflect a recency bias.

If a sample size of 100 trades were placed, the backtesting process would be woefully shallow in terms of reflecting market environments.  Backtesting Forex Strategies and understanding the market environments goes hand-in-hand.

Is a currency pair in a bullish or bearish market, where a clear direction exists for the overall sentiment?  Is a currency pair in a ranging market?  What about volatility or major news breaks (Brexit, Greek Debt Crisis, Election of Donald Trump)?

How can this be addressed?  It’s not so simple, but the objective should be to backtest FX strategies in multiple environments and backtest within given time frames where the market is in a particular mode.  This way, you can figure out what markets the strategy responds best.  This enables you to change the code to place trades when it is advantageous.

Beware of model-fitting.

The issue here is that when Backtesting Forex Strategies, you can end up with a cherry-picked approach that places trades based on historical anomalies.  Your criteria cannot be such that it places trades only when a narrow set of events takes place at once.  Often with model-fitting, the number of trades going forward or even looking back in a backtest frequently are low.

It’s one thing to observe that a strategy is better for a particular market environment, it’s quite another to be far too specific.  Backtesting requires understanding the market environments of the past so that the interpretation of the trading strategy’s results can be understood.  Otherwise, it just seems like the backtested strategy may have been inconsistent for reasons that are unexplained.  Sometimes, the inconsistencies can be explained in a backtest, which is why the coding process requires frequent backtesting.

It’s not just for the analysis, it’s for the execution as well.  Understanding that the criteria for entry and exit are met and executed in a backtest are important.

Backtesting Forex Strategies is a process that involves three things:

• Debugging and ensuring that the code functions as it should.
• How a strategy stands the test of time.
• How a strategy handles multiple trading environments and understanding the environments.

It is ideal to backtest a trading strategy across a long period of time and generate a large number of trades during this course.  There’s no exact number, but if the sample size seems too low or if it is too high, something went wrong.

Backtesting Forex Strategies is not necessarily as easy as looking at the results of the backtest and making a conclusion.  However, going through this is far better than trading manually, which is haphazard.

Consistent Profitability is Not a Given

There are few sure things in life outside of death and taxes, but then again even these two certainties can be uncertain.  After all, you could create a situation for yourself where you pay no taxes at all and death is considered to be something that is going to be outdated.

Consistent profitability is very difficult to define.  It’s also difficult to attain even based on the loosest definitions.  Not every trade is going to turn out well and going immediately from backtesting to implementation is skipping some important steps.  One important step skipped is the forward test, which ensures that the actual execution of trades takes places beyond the black box.

There is no one specific strategy that is going to be profitable every month.  It is why despite Backtesting Forex Strategies with success and even going through all of the steps to go-live, it is never a sure-thing.  However, implementation of multiple strategies, even those that are not of your own creation is the best approach.  Instead of treating Forex Trading like it is a casino (playing one game at a time), create a portfolio like an investor.

Freevestor can help you create that portfolio for FREE.  So get started on your process to reducing risk and generating strong returns now.

The post Backtesting Forex Strategies Requires a Substantial Sample Size? appeared first on freevestor.