Zedge Metric Update 101
As part of the Zedge/ZedRacer merge, we took the opportunity to completely overhaul how we derive our key horse trait metrics. With the huge changes to the racing meta (no more down-classing and level-added speed), we had no choice but to modernize the way we calculate, and scale, Base Ability (BA), Distance Preference (DP), and Variance (VAR).
Up until this point, we have used an ELO-based system to derive placement-based times for the foundation of our metrics. It did the job from a relative perspective but the times were never "real" enough to provide members any speed metrics. We've moved to a normalized speed based system that not only allows for more accurate data/metrics, but allows us to include robust speed data in the tool (which we'll continue to build out).
There are two major reasons you're horse's scores are different:
1) Our new speed based system is far more accurate and our logic is much improved.
2) Our 0-100 scales have been tweaked in order to match 3.0 trait ranges, to accommodate the upward mobility in mean speed with leveling, and to provide greater granularity at both ability extremes.
Let's dive into the 3 major metrics and explain some of the changes to the process/logic and why you may be seeing changes in your horse traits.
BASE ABILITY (BA)
The scale for BA had to be extended out to accommodate a ceiling BA of a 700 level horse with no DP. That would likely lower BA values for most mid-range, non-leveled horses. But, because BA is no longer passed in breeding, the need to maintain a "base" starting level BA went away and the new BA is now your current BA with the level speed added. That should lead to higher BA values. How much your BA score has changed will be based on how your horse falls in those offsetting forces.
DISTANCE PREFERENCE (DP)
The major change to DP, besides some complicated methods to derive a DP score when we only have races at one distance or in one funnel, is the rescaling around 3.0-viable outcomes. As we've done with VAR, we've reset the 0-100 scale so that 100 is roughly the highest value attainable (in high sample) in the current breeding algo. To accomplish this, we've extended the scale from 101-105 to make room for the Breeding 1.0 outliers. If your DP score has changed, it's either due to more accurate methodology or the pushing upwards of scores due to 100 representing a lesser value than it did in the old system.
This is where you'll likely see the biggest differences from the old system. Firstly, using normalized speeds is SO much more accurate for calculating standard deviations than our old ELO-based system. Secondly, like DP, the scale was extended to 105 in order to keep the 0-100 scale reserved for the current possibilities in 2.0-3.0. That change allowed us the ability to get more granular in our distinction between values at the top and bottom of the VAR scale. With DU and 1v1 becoming a popular format since we last set our scales, this was important. It also allowed us to align our 50-99 scores with the corresponding speed SDs people are familiar with on other popular tools. For instance, an 87 score now equals a meter per minute standard deviation of 8.7.
The result of these changes is generally an increase in the VAR scores for lower VAR horses (there was a huge range of SDs that scored 0 in the old system that might extend all the way to 15 now), and a decrease in scores for horses with an old 70-95 VAR. In most cases it's not newly discovered ability, but just the new scale.
We'll always continue to refine our metrics to give you the most useful data to make better ZED decisions. We understand that these new scales will be a tough adjustment at first (it was for us as well), but we are confident it's a MUCH better system moving forward.
As always, if you have any specific questions, you can shoot me a DM or message me in the ZEDGE discord: https://discord.gg/Stah4QNT