Refine Horse-Racing AI Scoring
Budget: $250 – $750 AUD
My horse-racing app is already live on the Base44 platform, but the AI module that converts form-guide data into race scores still isn’t firing on all cylinders. Right now it mis-scores runners and, more importantly, fails to read horse performance history in a way that reflects actual race outcomes. I need an experienced data-driven developer to step in, tighten the existing logic, and teach the system to read the form properly.
Scope of work
• Inspect the current Base44 code, database structure, and the model I built for calculating scores.
• Rework the analysis layer so the AI digs deeper into each horse’s performance history, weighting recent runs, margins, distance/class changes, and lay-off periods more intelligently.
• Ensure the refined logic flows cleanly into the point-allocation routine so the final rankings are accurate and repeatable.
Acceptance criteria
– A sample of past meetings (I’ll supply the CSVs) must return scores that match my manual benchmark to within 1 %.
– The new code runs inside the existing Base44 environment with no broken views or performance hits.
– Clear inline documentation explains every rule adjustment so I can tweak parameters later without recoding.
If you’re comfortable wrangling racing data, ML-style feature engineering, and the quirks of Base44’s scripting language, I’d like to hand this over quickly and see reliable numbers on my dashboard within the week.
Scope of work
• Inspect the current Base44 code, database structure, and the model I built for calculating scores.
• Rework the analysis layer so the AI digs deeper into each horse’s performance history, weighting recent runs, margins, distance/class changes, and lay-off periods more intelligently.
• Ensure the refined logic flows cleanly into the point-allocation routine so the final rankings are accurate and repeatable.
Acceptance criteria
– A sample of past meetings (I’ll supply the CSVs) must return scores that match my manual benchmark to within 1 %.
– The new code runs inside the existing Base44 environment with no broken views or performance hits.
– Clear inline documentation explains every rule adjustment so I can tweak parameters later without recoding.
If you’re comfortable wrangling racing data, ML-style feature engineering, and the quirks of Base44’s scripting language, I’d like to hand this over quickly and see reliable numbers on my dashboard within the week.