ArbMakerPlus extras: ranking scan results

Posted by on May 9, 2012 in ArbMaker News! | No Comments

“I can’t imagine trying to trade cointegrated pairs again without your application, the ranking system (1 – 4) really makes a tremendous difference identifying appropriate pairs.”

–  Marcus L.

ArbMakerPlus contains a number of features designed to find the most tradable opportunities that can be difficult to otherwise easily identify. One of these is the ArbMakerPlus ranking system about which a user sent us the feedback above.

Scan results can run to the hundreds just for a single sector depending on the criteria used. Going through each of these results one by one to assess the Z-score, normality and other graphs/data is time intensive.

Most begin the process by considering the Z-score chart. This shows the cointegrated relationship between a pair of instruments in terms of standard deviations from the mean quilibrium. From this information trade entry/exit points can be determined.

An ideal Z-score chart might oscillate obediently like this (taken from an intra-day scan):

However, because most Z-score charts do not look so attractive it can take a lot of digging to find such examples. The ranking filters, shown below, can help:

These work in 2 ways:

  1. They find evenly distributed axis crossings
  2. They identify the pairs with the most normally distributed data

Normal data has many useful properties one of which is its very helpful probability distribution. The Bell Curve, often used for marking student scores, is a practical example.

Thus, finding these kinds of distributions it tends to help trading – and that is what the ranking filters do.


(i) The Z-score chart above for VALE/FOE was turned up using ranks of  ‘3’ across the board

(ii) For explanatory detail behind the ranking settings take a look at page 23 of our documentation

Leave a Reply