TrackSmart AI - Cumulative Performance Report - NYRA Tracks

Reporting Period: December 26, 2025 – April 4, 2026

Track: Aqueduct (AQU)

2. Executive Summary: April 4 Update

   • Volume Expansion: The tracking dataset has grown to 420 total races with the addition of 28 races run across the April 2, April 3, and April 4 cards.

   • Strike Rate Resilience: The cumulative Top 4 Strike Rate held perfectly steady at 77.6%. The AI opened the week with a dominant 87.5% strike rate on Thursday and maintained a highly consistent 75.0% through Friday and Saturday's cards.

   • Value Normalization: The Average Payout for AI hits adjusted upward from ~$7.73 to ~$7.81. This was driven by an influx of chaotic deep value hits over the weekend, highlighted by Lika Rolling Stone ($25.64) and Mo Attitude ($17.06).

   • Down-Ballot / Alternate Strength: The rule to slide alternate selections (5th/6th rankings) into the active Top 4 when multiple scratches occur was the MVP of the week. This structural depth proved incredibly lucrative, catching multiple massive winners that were elevated into playable matrix slots by end-users.

3. Daily Performance Breakdown

3. Winner Distribution by Rank (Cumulative)

4. Cumulative Statistics Summary

5. Aggregate Observations: Apr 2 - Apr 4 Insights

   • Adaptability to Scratches (The Move-Up Rule): The algorithmic rule to automatically move alternates into the Top 4 playable matrix in the event of scratches was highly validated this week. Rampant scratches elevated massive payouts like Lika Rolling Stone ($25.64), That's Funny ($6.26), and El Grande O into playable positions. Observation: The AI's static deep-ballot (5th-8th) rankings are highly robust and critical for protecting exotic structures on days with high scratch volume.

   • Chaos Diagnostics vs. Trip Executions: The AI's macro-level pace mapping remains elite—it successfully flagged "High Chaos" and "Pace Meltdowns" prior to several massive upsets (e.g., Race 4 and Race 8 on Apr 4). However, it occasionally misdiagnosed the specific beneficiary. In Race 8, the AI correctly predicted a destructive pace duel, but missed 15-1 Fiddling Felix because it didn't anticipate the jockey entirely abandoning the horse's usual early-speed run style to execute a deep-closing trip. Corrective Action: Implement dynamic run-style flexibility algorithms that account for alternative trip scenarios and jockey intent when a destructive pace duel is heavily projected.

   • Layoff Penalties vs. Class Proven Runners: On April 2, the AI correctly mapped a slow pace but entirely dismissed Hello Beauty ($10.08), heavily penalizing the horse for declining numbers off a layoff while ignoring back-class. Corrective Action: Reduce the layoff penalty weight for class-proven runners, or implement a high-percentage connection modifier (e.g., Flavien Prat) to prevent logical horses from dropping below the fringe threshold.

   • Bottom-Tier Maiden Volatility: The AI completely dismissed 7-1 Icy Legs on April 3 based on poor raw speed numbers. Corrective Action: Soften the raw speed figure gap penalties in basement-level Maiden Claiming dirt routes. Marginal trip improvements in these fields can yield massive form reversals that rigid historical data struggles to capture.

TrackSmart AI: Cumulative Performance Report - NYRA Tracks

Reporting Period: December 26, 2025 – April 19, 2026

Track: Aqueduct (AQU)

Executive Summary: April 19 Update 

• •  
  • Volume Expansion: The tracking dataset has grown to 484 total races with the addition of 33 races run across the April 16, April 17, April 18, and April 19
  • Strike Rate Resilience: The cumulative Top 4 Strike Rate surged to 78.1% (from 77.6%), driven by an elite 84.8% strike rate (28 for 33) over the four days. This run was highlighted by a spectacular, flawless 100% strike rate (9 for 9) on April 18.
  • Value Normalization: The Average Payout for AI hits adjusted slightly downward from ~$7.71 to ~$7.66. The algorithm locked onto logical winners and heavy favorites seamlessly, while still capturing solid mid-priced value like Ez Roll ($16.12) and Pass the Hat ($14.36) to maintain a strong
  • Down-Ballot / Alternate Strength: The algorithmic rule to slide alternate selections into the active Top 4 when scratches occur proved to be an absolute powerhouse this This protocol flawlessly preserved sequences by auto-elevating winners like Two's a Crowd ($10.30), Bourbon Betty ($7.02), Kenny Be ($6.22), and Coquito ($2.32) directly into winning matrix slots.

Daily Performance Breakdown

(Note: Prior totals reflect data up to April 12. In previous documentation, 4th pick was listed as 42 for the initial prior totals, then bumped to 43 on Apr 12).

3. Winner Distribution by Rank (Cumulative)

4. Cumulative Statistics Summary

5. Aggregate Observations: Apr 16 - Apr 19 Insights
   • Flawless Day & Structural Versatility: The AI put on a spectacular clinic on April 18, hitting 9 of 9 winners. The algorithm adapted beautifully to all track conditions, hitting perfectly on stamina reserves and class shifts in turf races, while equally isolating lock-down heavy favorites and mid-priced early speed setups.
   • Scratch Protocol Perfection: The AI's systemic ability to handle scratches saved significant ROI. By mechanically elevating alternate selections to fill voids, the system directly engineered winning hits for Coquito ($2.32), Kenny Be ($6.22), Bourbon Betty ($7.02), Two's a Crowd ($10.30), and El Paco ($6.28).
   • Meltdown Blind Spots (Deep Closers): In severe pace meltdowns ("Suicidal/Contested Duels"), the AI properly predicted the front-runner collapse but struggled to isolate the correct deep closer, resulting in missed longshots like Timaeus ($46.40) and Blown Cover ($9.08). Corrective Action: Loosen base speed pars for deep closers when a "High Chaos / Meltdown" scenario is flagged, and ensure outside-drawn closers in turf sprints are accurately
   • Maiden Turf Sprint Volatility: Lightly raced horses and First-Time Starters (FTS) setting the pace in maiden turf events broke the model's speed pars (e.g., Moonlight Drive at $14.92, All of It at $13.28). Corrective Action: Adjust the turf sprint FTS algorithm to grant higher baseline E1/E2 projections for outside-drawn debut runners with steady gate Broaden the turf-pedigree algorithm to account for elite jockey/trainer shifts (e.g., Clement/Franco) on non-first-time starters.
   • Stagnant Speed Over-Penalization: On April 17, the AI relegated wire-to-wire winner Double Act ($32.56) to "Fringe" status due to stagnant speed figures, despite elite barn connections. Corrective Action: Increase the "Need-the-Lead" probability and front-running upgrade for elite trainees (like Bill Mott) in maiden turf routes when there is a clear lack of dominant early speed

TrackSmart AI: Cumulative Performance Report - NYRA Tracks

Reporting Period: December 26, 2025 – April 26, 2026

Track: Aqueduct (AQU)

2. Executive Summary: April 26 Update
   • Volume Expansion: The tracking dataset has grown to 511 total races with the addition of 27 races run across the April 24, April 25, and April 26 cards.
   • Strike Rate Resilience: The cumulative Top 4 Strike Rate dipped slightly to 77.7% (from 78.1%), driven by a 70.4% strike rate (19 for 27) over the volatile three-day stretch. The AI model successfully navigated heavy weather disruptions and massive scratch scenarios to remain highly consistent.
   • Value Normalization: The Average Payout for AI hits remained remarkably stable, adjusting a fraction of a cent upward from ~$7.66 to ~$7.67. While wet tracks yielded several chalky outcomes, the algorithm perfectly offset this by sniffing out excellent mid-priced value on April 24, including Factually Correct ($12.16) and Sounds Like a Plan ($11.18).
   • Down-Ballot / Alternate Strength: The algorithmic rule to slide alternate selections into the active Top 4 when scratches occur proved vital during a weekend dominated by off-turf surface This protocol flawlessly preserved sequences by auto-elevating winners like In the Dark ($12.26) and Annexperience ($7.32) directly into winning matrix slots.

2. Daily Performance Breakdown

(Note: Prior totals reflect data up to April 19. In previous documentation, 4th pick was listed as 46 for the initial prior totals. April 24 Race 1 dead heat counts as a single race win attributed to the top pick).

3. Winner Distribution by Rank (Cumulative)

4. Cumulative Statistics Summary

5. Aggregate Observations: Apr 24 - Apr 26 Insights
   • Wet Track Bias & Off-Turf Chaos: Heavy rain and muddy/sloppy sealed tracks caused massive disruptions over the weekend. While the AI model performed superbly on the planned main-track dirt races, its blind spot was badly exposed in off-the-turf races (e.g., Races 5 & 6 on Apr 26). Massive scratches rendered the pre-race turf pace models entirely obsolete.
   • MTO & Alternate Failsafe Needs: The AI entirely missed chaos winners like McDiesel ($46.12) and So Vain ($26.50) because it bypassed Main-Track-Only (MTO) and Also-Eligible runners, assuming they would Corrective Action: Implement a dynamic "MTO / Surface Switch" contingency protocol to automatically evaluate alternate entrants and re-handicap speed pars the moment track conditions downgrade to "Off-Turf."
   • Handling Late Scratches: The AI’s robust depth proved highly effective. The Top 3 scratch rule cleanly preserved multiple hits across the three Most notably, it caught In the Dark (Race 5, Apr 24) and Annexperience (Race 8, Apr 25) after massive waves of scratches bumped them up the matrix directly into winning slots.
   • Pace Meltdown and Tactical Speed Tweaks: The model struggled to correctly weight early/tactical speed in specific wet and turf route scenarios. It overly penalized "regressing" speed figures for horses that ultimately secured forward placement on sloppy tracks (e.g., Unbroken Chain, $10.60). Furthermore, tactical mid-priced pressers compromised expected "lone speed" runners (e.g., Papagiorgio, $11.12). Corrective Action: Suppress regression penalties for dangerous early speed on wet/sealed tracks, increase the weighting floor for ground-saving tactical speed on turf routes, and better respect pressers against assumed lone leaders.
   • Forgive Last Out & Layoff Parameters: The AI heavily penalized horses for single massive regressions or form drop-offs after returning from long Corrective Action: Implement a "2nd-off-the-layoff" improvement bump for horses who previously ran against tougher competition, and apply a "forgive last out" cap on penalties if previous speed figures are firmly on par with the class level.

TrackSmart AI: Cumulative Performance Report - NYRA & Regional Tracks

Reporting Period: December 26, 2025 – May 3, 2026

Track: Belmont at the Big A (BAQ) / Finger Lakes (FL) / Aqueduct (AQU)

Executive Summary: May 3 Update

• • Volume Expansion: The tracking dataset has grown to 566 total races with the addition of 55 races run across Finger Lakes and Belmont at the Big A from April 27 through May
  • Strike Rate Resilience: The cumulative Top 4 Strike Rate surged to 79.2% (from 77.7%), driven by an extraordinary 92.7% strike rate (51 for 55) over the past week. The AI model successfully navigated track surface changes and achieved rare perfect cards on May 1 (9-for-9) and May 3 (12-for-12).
  • Value Normalization: The Average Payout for AI hits remained incredibly stable, hovering around ~$7.65. While predictable, chalk-heavy cards at Belmont slightly depressed daily averages, the algorithm perfectly offset this by sniffing out massive longshots like Reiterate ($46.24) and Mia Nipotina ($25.24) at Finger
  • Down-Ballot / Alternate Strength: The dynamic algorithmic rule to slide alternate selections into the active Top 4 when scratches occur proved completely This protocol flawlessly preserved sequences by auto-elevating winners like April Storm ($2.96), Sharpaz ($3.60), and Graded Stakes winner Antiquarian ($6.34) directly into winning matrix slots.

Daily Performance Breakdown

(Note: Prior totals reflect cumulative data up to April 26. Daily breakdowns from previous reports have been condensed into the Prev. Totals row. Live scratch protocols dynamically shifted runners up in rankings across multiple races this week, securing numerous top-tier hits.)

 Winner Distribution by Rank (Cumulative)

Cumulative Statistics Summary

Aggregate Observations: Apr 27 - May 3 Insights

• • Dynamic Scratch Protocol Mastery: The automated contingency rule to bump alternate selections up following late scratches of top contenders proved entirely critical this week. It directly preserved perfect and near-perfect tickets across both tracks, elevating winners like April Storm, Sharpaz, Baby Sassicaia, and Antiquarian immediately into primary matrix slots.
  • Pace & Track Bias Adaptability: The AI effectively modeled regional track conditions, recognizing the heavy advantage for forwardly placed ("Lone Speed") runners at Finger Lakes' abbreviated 5F sprints, while accurately predicting precise pace meltdowns for deep closers at Belmont (e.g., Grammy Girl in the Vagrancy-G3).
  • Class Drop & Shipper Blind Spots: Despite an astronomical overall accuracy this week, the model occasionally undervalued the severe class relief of major-circuit shippers dropping into weaker maiden fields (e.g., Juniors Pal at Finger Lakes, $27.70). Corrective Action: Implement stronger algorithmic weights for class-drops from major circuits (like AQU) into regional maiden events.
  • Turf Routing Mud Meltdowns: The AI thrived on Belmont turf routes, consistently identifying top-tier Late Pace (LP) closers and potent First-Time Starters (e.g., King's Remark, $15.90; Midnight Concerto, $13.56). However, its blind spot was exposed during muddy track chaos (e.g., Tarpaulin, $20.58). Corrective Action: Apply pace-meltdown logic triggers or heavier pedigree modifiers for deep closers on sealed/muddy tracks to account for faster-decaying front-end speed.