We have been trying to test the difference made to a simple mean-reversion strategy on the ES based on Keltner bands by different types of entry and exit methods.

The strategy is a standard mean-reversion one where we enter in a single go by selling at the top of the band or buying at the bottom. We exit in stages after we are either past the middle of the band, or when past barriers scaled out smoothly from centre of band to the opposite side. In the latter case if we had just entered long at the bottom of the band and had 3 exit levels these would be at the middle of the band, a quarter of the way from the top of the band to the bottom and at the top of the band.



We tested the base strategy - i.e. always in and just alternating between long / short with a single contract.



We then tried scaling out between the centre band and the exit (opposite band).



As you can see performance (Net profit) increased but it didn't take the shape ratio up much.

In my next post we will walk through some of the tests we have done with the same strategy and various entry / exit methods (e.g. Rectangle, Triangle, etc).

Related links:

Scaling in and scaling out...when the total is NOT the sum of its parts Have you ever abandoned a strategy because it had untolerable drawdown? When examining the potential profit additions and potential loss reductions with scaling, the possibilities are quite significant. The following comes with some assumptions. …

We tested three types of position sizing: Rectangular, Triangular and Inverted Triangular, where we entered/exited over a different number of periods.

To give a specific example, if we had 6 maximum positions and 3 rows then

[table] border="1" align="center"
|-
| class="tcat" colspan="4" align="center" | Entry Numbers and levels
|-
| style="font-style: italic" | Entry / Exit Barriers
| style="font-style: italic" | Rectangular
| style="font-style: italic" | Triangular
| style="font-style: italic" | Inverted Triangular
|- align="center"
| 1
| 2
| 1
| 3
|- align="center"
| 2
| 2
| 2
| 2
|- align="center"
| 3
| 2
| 3
| 1
[/table]

So in this case we would always try to enter/exit over in stages as long as the relevant signal was in place. Also in the case of a strategy with more than one exit barriers that had hit the top of the band so we sold fully short, then fell so we closed some (but not all) short positions, and then returned to the top we would fully go short again.