In the first part of this piece I looked at the impact of goals across various leagues to see how they affected Total Goals lines and the Asian Handicap market. The aim was to use StrataData to see if there is any edge to be gained from the patterns that we identify.
To expand the investigation further I now want to take a look at how chance creation affects these lines, if at all.
For starters, let me quickly explain the concept of the data we collect to refresh your memory…
At Stratagem we put chances into six buckets, labelled “Superb”, “Great”, “Very Good”, “Good”, “Fairly Good” and “Poor”. Each of these buckets has its own conversion rate, which is linear across all 22 competitions that we cover and is proven to hold significant edge over purely shot-based expected goals models. For reference, a fuller explanation can be found here.
The above table shows the average conversion rates for each of our six chance types. Using these figures we can estimate an expected goals total for each team in every game they play. This is the very top level of what can be done and utilises a simplistic method, as things such as game state are not taken into account for ease of use in the rest of this post.
An example of how this can be used is shown below:
When summed, these chances equal 1.96 to Manchester City and 0.69 to Arsenal. As the game finished 2-1 this can be considered a good example of chances mirroring the outcome of the match. However, it should be noted that these are average conversion rates and that there will often be games where a team scores three Poor Chances and others where they create a host of Greats and miss them.
Going further, the combined total of chances created in this game equals 2.65, which makes it reasonable to assume that the Total Goals line for this game could have been set at either 2.5 or 2.75. It was actually 2.75 in this case, which proved to be extremely accurate.
Looking in broader strokes we can see how often the chance data is close to the Total Goals line and if it can be used as a predictive measure.
To do this I looked at how often the Total Goals lines were within a fixed margin of both the actual goals and the expected goals. This allows us to see how accurate StrataData is at determining how many goals there would be in a game if average conversion rates held true for all teams in all games.
Over the course of a number of games it can be useful for seeing which teams are over or underperforming consistently, against both the goal line and their expected goals scored and conceded. In theory this would allow smart traders to get ahead of the curve if the markets are slow to react.
Beginning with a 0.25 goal margin either side of the Total Goals Line I looked at how often the actual number of goals and the StrataData expected goals fell within this range, before increasing this margin to 0.5, 0.75 and 1:
The above graph shows that StrataData is better at being within the specified range of the Total Goals Line than actual goals is. This may seem counter-intuitive, but as both are reactive measures known after the game has been played it is key to see how teams could perform on a repeated basis rather than what they actually did.
To give a clearer example I will take two games and break them down completely:
The first is West Ham United vs. Watford from the 10th September 2016 in the Premier League. This game finished 4-2 to the away side and comfortably beat the pre-match Total Goals Line of 2.5.
So does this mean the line was set wrongly and it should have been higher to start with?
Not necessarily, because StrataData tells us a different story. The game was actually much tighter than the total goals scored suggests, with poor quality chances being converted at a much better than expected level to give an actual xG over the whole game of 1.47, which was well under the natural goals line.
So, then, is it safe to assume that Watford could consistently outscore and beat the Total Goals Line?
No, not at all… Take their home game against Hull City on the 22nd October in the Premier League, which had a Total Goals Line of 2.25 pre-match and was expected to be low scoring.
This time it was, as it finished 1-0 to Watford thanks to a late own goal. However, once again the score did not tell the whole story of the game. Indeed, StrataData suggests that had average conversion rates applied then the total xG would be 3.82, which comes out well above the Total Goals Line.
Of course it is easy to pick out individual games and fit a narrative to them, but crucially what StrataData can do is better estimate teams that are consistently over- and under-performing against the Total Goals Line. This in turn can be used in a variety of models as a better predictor of future performance than simply using actual goals, or just use as an indicator for which side of a Total Goals trade to take for bettors who rely more on feel.
Ultimately, any edge that traders can get over the market can be significant in terms of long-term gains and that is what StrataData aims to provide. So if you are a StrataBet customer and want more information on how to use the data collected in our Reviews to improve your trading, please feel free to get in touch at firstname.lastname@example.org for some tips.
Dave Willoughby (@donceno)