Entry Filters

Entry and exit signals can employ filters. In the most general sense, a filter is additional information intended to improve the quality of an entry or exit. The only purpose of a trading filter is as additional confirmation or validation of the entry.

Definition: An entry filter rule adds additional information to produce a more accurate and reliable trading strategy entry or exit.

In other words, entry or exit filters add additional indicators, analysis, market facts, or trading rules to the primary entry rules. A filter is really part of a more complex entry or exit rule.

A filter can be very simple, and this single, simple filter can be applied in conjunction with the entry rule. For example, if today’s close is higher than yesterday’s close, then take the current buy entry signal.

A single, complex filter can be used. For example, if today’s close, high, and low are all higher than yesterday’s, then take the current buy entry signal.

Multiple simple filters and multiple complex filters can be used. The main purpose of the entry filter is to increase the overall accuracy, reliability, and quality of the trading strategy’s buy-and-sell entries.

For the purpose of illustration, let us consider the following short list of filter examples. These are examples of filters on buy signals (sell signal filters are the opposite). The trading strategy will accept a buy entry as confirmed by its respective filter if:

1. The Relative Strength Index is below 30

2. The Relative Strength Index was below 30 on the previous bar and is

now above 30 and rising higher

3. Today’s high is higher than yesterday’s

4. The last buy signal was profitable

5. The close today is higher than the close 20 days ago plus 10.00 points

6. The close yesterday is in the top third of yesterday’s daily range

Therereallyisnolimittothelevelofcomplexitythatcanbeintroduced into a trading strategy with the use of filters. It is important to note that as the number and complexity of the trading filters increases, so can the difficulty of coding and testing a trading strategy. It should also be noted that the likelihood of overfitting also rises with the number of filters. To take this to an absurd extreme, an extremely—and absurd—overfit model would feature a different filter for every bar in the simulation data. Such

an overfit model would exhibit exceptional profit in simulation and unprofitable performance in real-time trading.

Entry Filters