Developing An Edge
by John Sweeney
Once you've developed something with an edge on the market, it's only natural to want to exploit it
mercilessly, making hay while the sun shineth, so to speak. The simplest way I know to do this is to add
more positions to your basic direction, buying dips when long, selling rallies when short. In my last three
Settlements, I've battered one harebrained idea into reasonably profitable shape so that it gives us good
trend selections and gets me out in a timely manner. However, there's more.
Once we're in a good trade (which we know via the maximum adverse excursion [MAE] statistics), we
may as well add stocks or commodities, but in some rational formulation. How much, we ask, of a dip is
enough to trigger additional entries?
Two things to think of immediately here: (1) We define the prospects for an additional good trade not by
pyramiding (which is using the profits on early contracts for margin on additional contract) but by having
the cash to buy more on a completely independent basis; and (2) we define a good trade in this mode just
as we define a good trade in the original mode — exit and entry rules together with measurements of the
worst price moves against the position.
Along this line, Figure 1 shows the trades from the basic trading system developed to date, together with
a 10-day simple moving average. Take it from past work (Settlement, April and May 1991) that averages,
because of their lag behavior, efficiently skim along the tops and bottoms of price ranges. You can
eyeball this "efficiency" or you could test for it quantitatively. To speed things up and avoid the
arguments of optimization, I'm simply specifying a value range I've used for years: eight to 10 days.