This shows that the A/E is a much better tool to use for winner finding as it is able to account for rogue results.

Flaws With the ROI Stat

Today Franny Norton had three winners (from just three rides) at Southwell. The odds of those winners were 8/1, 20/1 and 100/1!

Results such as this can have a big impact on stats and systems because of the extreme high prices. For this reason it is much better to rely on the A/E stat rather than ROI% when creating horse racing systems.

Prior to today Norton's all weather stats were:

Norton returned a loss of around 15% prior to today but now look at the stats when todays results are included.

The strike rate has changed by 0.2%, the loss backing his runners to a level £1 stake has dropped by £128 and the ROI dropped by 6.1%. The changes to the level stakes profit and the ROI% are huge and can cause a big swing in a systems performance.

Notice the change in the A/E - an increase of just 0.02 from 0.93 to 0.95. This shows that the A/E is a much better tool to use for winner finding as it is able to account for rogue results.

Another example is with the sire of Norton's 100/1 winner. Prior to today Sire Singspiel(IRE) returned a loss of just 2.2%. But after the 100/1 win by his offspring Sweet Child O'Mine that loss turns into a profit of 11.3%!

Now imagine a punter researching a list of sires to follow on the all weather. It is quite likely he would spot that Singspiel has returned a greater than 10% profit on the sand and may choose to follow his runners. Of course, if you take out the rogue 100/1 winner the stats are not positive.

Filter out Big Prices

If you do have to use the ROI% stat then try filtering out the outliers from your research. Put a price cap on all runners of say 66/1 or lower if the sample size permits.

Related Articles and Resources

• Stats Guide - Info on the Return on Investment (ROI%) Stat
• Stats Guide - The A/E Value Index Stat

Comment on this Article in the Message Board

• Why the A/E stat is more important than ROI%

          

Article created 11-Dec-09. Stats may have changed since.