EPL Draw - Over Consideration

The NFL play-offs get underway today with the Wild Card games where historically the value is on Unders. Only twice in the last fifteen seasons has Overs 'won' more than two of the four games, the last time being the 2011 season. Of the other thirteen, three were split for an overall 40-27-1 record.

I hope a few of you were able to make some money on the Wild Card games this weekend, with all four matches going Under, the bet recommended above. 

This is the second sweep on Unders since 2012, and after the league reörganised in 2002, the Under is now 44-27-1. Tuck that thought away for this time next year. I also pointed out that:

Road 'Dogs are generally under-rated at this stage also, with a 28-22-2 record ATS, and in the last two season are 7-0-1.

Not perfect on this one, (Buffalo Bills missed out on making it so by 1/2 a point) but I'll take two out of three winners any day, and overall this bet is now 30-23-2 over the same period. The Seattle Seahawks were road favourites, but the record on such teams was an inconclusive 7-6-2 and as an Eagles fan, it would have been tough to back against them anyway.

Onto the Divisional Round next week where the value play tends to be the Overs, with Road teams continuing to do well ATS. Regarding the value on Road teams, I suspect that the bias at play here is that the Home team is rested in these games after a bye week, and expectations for them to do well are set unrealistically high.

Home teams do win the game 69% of the time, but ATS they are just 30-37-1. Road teams from the Eastern Time Zone have a 23-11 record, but for the first time ever (well, since 2001), there is not an Eastern Time Zone team on the road.

As for the apparent value for Overs, I'd be wary. We're dealing with a small sample again, and while Overs has a 37-31 record overall, most of that advantage is in games where the line is 12 or higher, and none of this weekend's four games are close to that.

I took a look at the data A Lucky A Day linked me too for the EPL Draw, and it has Money Line data going back to the 2007-8 season, 4695 matches and if you had backed the Draw in each one, you'd have an ROI of -3.9%.  

Ignoring the matches where there is no Pinnacle price for the O/U 2.5 market (and remember, these prices are all opening prices, so this exercise is out of curiosity rather than to 'prove' anything) we have 3311 matches, and if you had backed the Draw in all of these, your ROI would have been -1.39%. 

You really have to work quite hard to lose money backing the Draw, because by simply ignoring matches where the Draw is priced at 4.0 or higher, you are into profit, albeit by just 0.04%, but of course, readers will know there are other ways to eliminate losers, namely ignore games where one side is odds-on.

Now the ROI is 5.7%, with profits of 85.16 units from 1497 matches which might get your attention, although not that of everyone if this comment from the Betfair Forum is to be taken seriously (hint: it isn't):

i devised a system returning about 20% roi which is decent but not worth the time and effort that's involved

Good grief. Some people are clueless, this individual in more ways than one, but that's a good thing when using an Exchange!

Now it gets interesting. There are 229 matches where the Under and Over prices were both odds-on. Avoiding backing the Draw in these matches and you'd have boosted your return to 9.4%.

Which leaves 935 matches where Unders opened as the clear favourite, and 333 where Overs was clear favourite.

Now it gets very interesting. One might expect the "Under Clear Fav" row to generate more profit, but as you can see from the table below, this is not the case. 70% of the profits are coming from 26% of the matches. 

Pareto anyone?

Not only do these matches generate a higher percentage of Draws, but the average Draw price is also higher. 

While the percentages of matches resulting in 0:0 and 1:1 are close whether Under of Over is favoured, (and both above the long-term averages for these scores - top row below), the number of Draws of 2:2 and higher is greatly increased when goals are expected. 

It's important to remember that this data set uses the opening Pinnacle prices on the Under / Over market, prices which may well change significantly in the days prior to kick-off, and that the sub-set of matches already excludes games where one team is odds-on or the Draw is 4.0 or greater.

In these games, which in the EPL are already known to be profitable when backing the Draw, using the Under price isn't helpful. While still profitable, they are actually bringing down the ROI.

These are interesting numbers to keep an eye on as the data for Pinnacle's Closing prices from Joseph's site accumulate.

With only half a season of closing U/O data from Pinnacle, the loss is currently 11.32 units from 37 matches.

I mentioned the cost of the closing price data in my last post, which should make you appreciate the access to Joseph's free data all the more. If anyone has this data and wants to check the observations above, I'm sure many Draw lovers would be very interested in the findings.