I have reason to be suspicious about seasons where MoS has been startling in Round 1, so I feel much more satisfied that the results for Round 1 of 2015 are pleasingly positive without being yell-from-the-rooftops amazing.

In the Round just finished, Investors enjoyed a small (1%) profit on the back of 4 from 7 successful Head-to-Head and 4 from 7 successful Line wagers. Subdued but profitable is just the way we like it.

Pleasing too was the performance of the ChiPS System of Tipsters and Predictors, which landed 6 from 9 correct Head-to-Head tips, an MAE of 22.48 points per game, and a Log Probability Score of +0.0071 bits per game. None of those results are table-topping, but they hint at a season that might be better than last, and a season that better memorialises the dog after which the system was named and fashioned.

Again, in the interests of those reviewing MoS results for the first time, I'll explain the tables below in greater detail. (New visitors might find this blog helpful if they've not already read it.)

On the left is the table for the Head-to-Head Tipsters, which this week shows Combo_NN2 as the best Tipster having selected the winner in 7 of the 9 games, missing only the Dees and the Dogs. Twenty other Tipsters recorded next-best 6 from 9 results, and even the worst of the Tipsters managed 5 from 9. The all-Tipster average came in at a very high 6.1 from 9.

In the middle of the table are the results for the Margin Predictors, which show that Combo_NN1 produced the round's best Mean Absolute Error (MAE) at 21.84 points per game, just narrowly ahead of C_Marg at 22.48. The column headed "Gap" records, in points, how far a particular Predictor is from the leader. So, for example, C_Marg would need to be, in aggregate, about 6 points closer to the final margins in Round 2 than Combo_NN1 in order to move past it. The all-Predictor average MAE for Round 1 was 24.4 points per game, which is, in historical terms, very low. If any single Predictor finishes with a season MAE this low it will be, by far, a MoS best-ever result.

The column with the Flag at the top records how often a Predictor has been closer to the final game margin than all other Predictors, and the column headed with a Sand Trap records how often a Predictor has been furthest away. The first set of bar charts shows the proportion of games in which a Predictor has been within a given range of the final margin. So, for example, Combo_NN1 was within 6 points of the final margin in 44% of games in Round 1.

Next, the block headed "Line Betting" records how a punter would have fared had he or she used a Predictor's margin predictions as the basis on which to wager in Line markets. Using C-Marg, for example, would have produced correct wagers in 89% of contests and generated a healthy positive ROI.

Lastly, the rightmost columns provide, firstly, the Log Probability Scores for the various Head-to-Head Probability Predictors and show that, unusually, WinPred in the clubhouse leader and, even more unusually, the three most directly Bookmaker-based Probability Predictors fill the last three places on the Leaderboard. That, almost certainly, will change.

The Log Probability Score of the Line Fund algorithm is the final figure shown. Based on historical performances, a -0.0003 result for it is very acceptable, and similar performances throughout the remainder of the season would be almost certain to drive significant Line Fund profitability.