Probability, Difference And The EPL Draw (2)

"You can scream from the roof top what the secret sauce is and it doesn't matter because people won't do it"

Although it isn't actually my "All-Time" most popular post, blogger's algorithm for such things tells me that my January piece on the draw has certainly attracted some interest.

In that post, I promised to update the English Premier League numbers in the summer, and with the NHL and NBA seasons now over, and serious sporting investment opportunities down to just the MLB for now, I was able to spend some time updating the EPL Draw numbers over the weekend.

We now have six full seasons with Pinnacle's closing price data, a total of 2280 matches.

If you're one of the few people who have not yet read the January post, it's worth reading, detailing my interest in the Draw as a bet going back to 1999, before playing with Elo ratings and accidentally stumbling across an edge which became the hugely successful XX Draws. 

Admittedly 201 isn't a huge sample, but the ROIs were decent enough.

Ultimately the key to their success was finding matches where, after accounting for home advantage and recent form, the teams were fairly evenly matched, an effort that problematically took several hours to calculate each week, hours that I no longer had given that my new position required regular travel. 

One adapts however, and I mentioned in the January post the idea of calculating the difference between the teams by simply using the odds (Pinnacle's Closing Prices), and using that as your selection criteria.

For example, in matches where the teams win probabilities are close, i.e. the difference is less than 25%, backing the Draw is +78.91 points from 393 matches which is an ROI of 20.1%. Not far short of the XX Draws except that it's a return made with a fraction of the effort. Note that I remove Pinncle's over-round to come up with the 'true' probabilities for each outcome.

If we use the raw difference as our guide, i.e. subtract one probability from the other, we can pivot around the zero for a 126.74 profit from 907 selections, an ROI of 14%. 

Let's take a look at when NOT to back the Draw. Backing the Draw in every EPL game over the last six seasons would have lost you 61.17 points. It's the second worst 'blind' strategy behind backing the Away team, and probably no surprise that after six seasons, the only 'blind' strategy in profit is Laying the Away team, for a profit of 0.6 points and an ROI of 0.026%.  

Backing the Draw when the Home team has an implied probability greater then 0.5 (2.0) would have lost you 81.57 points, while backing the Draw when the implied probability is greater than 0.27 would have lost you 121.80 points.

Eliminating just one of these categories and you have a profitable strategy which takes just a few minutes a week to apply. Set up a spreadsheet with some filters and see the ROIs in the high teens / low twenties: 

These numbers are for illustrative purposes only

The bottom line here is that you really need to make some poor choices to come up with a losing strategy for backing the draw in the EPL. The problem for most people is that it's not the most exciting way to make money, and as soon as an inevitable losing run comes along, they quit.  

I've also previously mentioned Big 6 and Little 14 matches. Overall backing the Draw in Big 6 matches for the past six seasons would have made 14.74 points (ROI 8.2%) while in Little 14 matches the profit is 17.01 points (ROI 1.6%). Exclude Manchester teams from the Big 6 and the profit is 25.41 points, ROI 35%! 

Worth noting here that the Big 6 matches are currently on a record 10 game losing streak dating back to 4 February, beating the previous record of 8 matches from 31 January to 26 April 2015, while Little 14 selections have lost eight straight. 

In Big 6 matches where the percentage difference is less than 25%, the ROI is 60%, while in Little 14 matches it is 14.7%.  

“Don't worry about people stealing your ideas. If your ideas are any good, you'll have to ram them down people's throats.” - Howard Aiken