After an extensive multi-month development cycle, the R9 report engine with AI has successfully completed its first fully stable, end-to-end test, marking a major milestone in the evolution of our handicapping analytics platform. This achievement follows a long period of refinement in which individual components of the system worked well on their own but failed to operate consistently together. Throughout the process, fixing one issue often led to another element breaking or reverting — a challenge common to highly complex, multi-layered systems.
Those setbacks made the breakthrough all the more significant. With the core foundations rebuilt and every section of the report now drawing its information from a unified and consistent source, the system has finally reached the level of reliability and performance we set out to create. The first full live test at Churchill Downs confirmed this success, with the engine not only running flawlessly but also producing several high-value selections, including multiple double-digit winners that showcased its analytical power.
Today, we are pleased to announce that the R9 engine is fully stable, fully synchronized, and ready for full deployment. This marks the completion of a long and demanding phase of development — and the beginning of a stronger, more accurate generation of performance reporting for our users.
Here is the PDF of the CDX1123 card.
Here are the full results at CDX.
https://www.equibase.com/static/chart/summary/CD112325USA-EQB.html
Race 3 — Grade: B+
The system identified the 8-1 winner (#7 Scot’s Law) within its key value structure and correctly placed the second- and third-place finishers in its projected rankings. This race produced a longshot winner and validated the model’s chaos-detection layer, which is designed to uncover exactly this type of overlooked horse.
Race 4 — Grade: A
The report’s Primary Selection (#5 Groveland) won at 5-1. The system also correctly identified the runner-up (#7 Raising Kane) within its supplemental selections. This was a clean, high-confidence win.
Race 5 — Grade: A+
Perhaps the strongest performance of the day.
The report listed #12 Lord Majesty (a 14-1 winner) in both its Tertiary and Value categories. When a horse is flagged in multiple deep-value slots and wins at a huge price, it demonstrates exceptional model reach. This was a textbook strike by the system.
Race 6 — Grade: C-
This was the system’s weakest race, missing the winner entirely. The second-place finisher was included in the selections, but the top of the board did not match the model’s expectations. This race did not materially affect overall ROI due to the strength of the surrounding races.
Race 8 — Grade: B+
The Secondary selection (#6 Thoughtthatcounts) won at 7-2, with the Exacta and Trifecta structures aligning naturally from the report’s ordering. While not a Primary winner, this was a solid and profitable outcome consistent with expected performance in mid-range allowance races.
Race 9 — Grade: A+
Another standout.
The Primary selection (#5 JoeJoe Go) won at10-1, the Secondary selection, #9 Mount Horeb at 3-2, finished second, and a Value Pick finished third — producing a near-perfect read of the race. This reflects the system’s ability to identify overlooked turf runners and build an accurate race structure around them.
Summary
Across the covered races, the system produced:
- Two major double-digit winners (14-1, 10-1)
- Two Primary winners at 5-1 and 10-1
- Five winners identified within the top/value stack
- One miss (Race 6), with no impact on overall profit
Overall Grade: A (+ 183% ROI with primary selections)
Below is video of the 9th race at Churchill, in which the primary, #5 JoeJoeGo at 10-1, won over #9 Mount Horeb at 3-2. The program assigned JoeJoeGo theoretical odds just under 3-1.
To see the above image in full resolution, click on it, then on the encircled “i” in the bottom right hand corner. Scroll down and click on the “view full size” link in the center.
Pro-Handicap Analytics 