In today's post I'll review the performance of all the teams that have been assessed as favourites by the TAB in games played during the period 2006 to the end of Round 17 in 2015, excluding only those games where the TAB bookmaker installed equal-favourites.
That leaves 1,862 games to analyse, which I'll split into six roughly equally-sized groups on the basis of the favourite's price, giving:
- 324 games where the favourite was priced under $1.12 (implying a victory probability of about 85% or more assuming a flat 5% overround)
- 362 games where the favourite was priced between $1.13 and $1.25 (implying a victory probability of between about 75% and 85%)
- 293 games where the favourite was priced between $1.26 and $1.36 (implying a victory probability of between about 70% and 75%)
- 308 games where the favourite was priced between $1.37 and $1.50 (implying a victory probability of between about 63% and 70%)
- 280 games where the favourite was priced between $1.51 and $1.65 (implying a victory probability of between about 58% and 63%)
- 295 games where the favourite was priced at over $1.65 (implying a victory probability of about 58% or less)
What I'm focussing on today is the scoring performance of the favourite relative to the underdog in each game to see whether the favourite registered more, fewer or the same number of scoring shots, and whether it converted those scoring shots at a higher, lower or the same rate.
The table below gives raw counts of the number of games based, firstly, on the outcome from the favourite's viewpoint and, secondly, on the favourite's relative scoring performance as just described. Note that some combinations of relative scoring shot and conversion performance are not possible for some results - a favourite can't, for example, lose yet register more scoring shots and have the same or a higher conversion rate.
From this table you can see, for example, that favourites priced under $1.12 have won 7 games during the period we're analysing where they registered fewer scoring shots than the underdogs but converted those scoring shots at a higher rate.
Summing this data by the favourite's result reveals that favourites win at about the rate implied by their pre-game bookmaker prices, which is just another way of saying that the TAB bookmaker is generally well-calibrated.
Favourites priced in the $1.26 to $1.36, and the $1.51 to $1.65 ranges do, however, seem to win a little less often than their TAB prices would imply, suggesting that there might be some slight miscalibration in those ranges or, instead, that the overround embedded in prices might increase as the favourite's price increases, and not be the flat 5% that I assumed earlier. (The topic of bookmaker overround and how to calculate it on a team-by-team basis has been discussed many times here on MoS - see, for example, this post where I sought to define a general framework. A site search on the term will return other posts.)
We can also sum the original data to reveal how often favourites record more scoring shots than the underdogs or convert at a higher rate than the underdogs, regardless of whether the favourite wins, draws or loses.
That summary shows that the favourites superiority manifests in scoring shot production far more than in scoring shot conversion. Favourites priced under $1.12, for example, produce more scoring shots than their opponents 91% of the time, but convert them at a higher rate only 61% of the time.
More strikingly, favourites priced in the $1.26 to $1.36, the $1.51 to $1.65, and the over $1.65 ranges all tend to produce more scoring shots than their opponents but also tend to convert them at a lower rate than their opponents.
Overall, favourites produce more scoring shots 70% of the time, but convert at a higher rate only 52% of the time, which is more evidence, I'd suggest, that conversion rates have a large random component.
Narrowing our focus to look only at games lost by favourites, we find that such losses are, overall, marginally more often attributable to poorer conversion alone than to lesser scoring shot production alone: 29% of games lost by favourites had them with a lower conversion rate (and the same or a larger number of scoring shots), while only 27% of games lost by favourites had them with fewer scoring shots (and the same or a higher conversion rate). In the remaining 44% of games the favourite had both fewer scoring shots and a lower conversion rate.
This overall finding is not true of each subgroup of losing favourites, however. Those priced in the $1.37 to $1.50 range, and in the over $1.65 range have tended to lose more often through producing fewer scoring shots than from having a lower conversion rate.
Finally, looking only at games won by favourites the analysis reveals that superior scoring shot production is the source of victory far more often than a higher conversion rate. Overall, 54% of winning favourites had registered more scoring shots yet had an equivalent or poorer conversion rate.
Conversely, only 10% of victorious favourites had a higher conversion rate yet the same number of or fewer scoring shots than their opponents.
SUMMARY AND CONCLUSION
Over the period 2006 to the end of Round 17 in 2015, favourites have won at rates roughly consistent with the probabilities implied by their pre-game prices. There's nothing remarkable about that.
Losing favourites have been about as likely to lose by generating fewer scoring shots than their opponents as by converting scoring shots at a lower rate than them, though some sub-groups of favourites, defined by their pre-game TAB prices, have tended to lose more often from one of these two sources than the other. The differences are never huge though.
Winning favourites, however, have rarely relied on their conversion rate being sufficiently superior that it has overcome a scoring shot equivalence or deficit. Over one half of all winning favourites have registered more scoring shots than their opponents yet converted at the same or at a lower rate.
In short, favourites lose about as often from inaccuracy as the do from an inability to register more scoring shots, but they win far more often from an ability to register more scoring shots than their opponents and not from merely converting them at a higher rate.