Filter Bursting

Jameson wasn't the first to ask:

hi. what is the second filter you applied for draws? 1st is ip 25+, and second?

Unfortunately, as I told Stewboss, since this filter has added value in each of the five EPL seasons for which we have Pinnacle Closing price data available, the filter will remain proprietary. A great man once told me:

Secrets are secrets. An edge shared is still an edge halved. If I give or sell you a secret then it's no longer a secret.

How valuable is such a 'secret' though? 

Some of you undoubtedly noticed that the number of Draws in last season's Serie A was the lowest since 1995-96. Of the 50 top flight seasons in England, France, Germany, Italy and Spain during the last ten years, only four have been lower, and two of those were in the traditionally higher scoring Bundesliga where you'd expect fewer draws. 

Unsurprisingly backing the Draw in every Serie A match last season would have been costly as usual, but the benchmark (matches where the Draw probability is 0.25 or greater) would have again been improved by applying the filter, reducing a loss of 35.01 points to a loss of just 8.99 points. 

It was the same story across Europe last season, with the filter adding a total of 133.54 points to the benchmark:

An ROI% of -1.61% isn't usually something that sets the pulse racing, but it's a whole lot better than the -12.37% on the benchmark.

One swallow doesn't make a season though, and one season doesn't make a golden goose, so I looked back a couple more seasons to see if the pattern held. 

As you can see, it did, turning a benchmark loss of nearly 150 points into a profit of over 100 points. 

For the five seasons for which we have Pinnacle's Closing prices, the results are here sorted by league / season Draw percentage:

While the correlation between the Draw and returns is obvious, the (usually) added value of the filter isn't apparently influenced by the Draw percentage.