TrackSmart AI: Cumulative Performance Report

NYRA & Regional Tracks — October 25 Update

TrackSmart AI | Prepared for Saratogabets | Week Ending October 25, 2026

 

1. Overview & Methodology

This report covers the cumulative reporting period of December 26, 2025 through October 25, 2026, spanning Belmont at the Big A (BAQ), Finger Lakes (FL), Aqueduct (AQU), Laurel Park (LRL), Santa Anita Park (SA), Penn National (PEN), Saratoga Harness (SARH), Woodbine (WO), Saratoga (SAR), Del Mar (DMR), Hollywood Casino at Charles Town Races (CT), Kentucky Downs (KD) and Keeneland (KEE).

The headline metric is the Top 4 Strike Rate: the percentage of races in which the eventual winner appears among the AI's four ranked selections. This week added 106 races run across Finger Lakes, Belmont Park and Keeneland from October 19 through October 25, bringing the tracking dataset to 2,601 races.

2. This Week at a Glance

Volume Expansion

The tracking dataset has grown to 2,601 total races with the addition of 106 races across Finger Lakes, Belmont Park and Keeneland from October 19 through October 25.

Strike Rate Resilience

The cumulative Top 4 Strike Rate improved slightly to 77.3% (2,010 for 2,601), up from 77.1%. The model delivered 90%+ hit rates on multiple formful cards, including a perfect 8-for-8 at Finger Lakes on October 19 and October 21, and 10-for-11 at Belmont on October 24.

Value Normalization

The average payout for AI hits held steady at roughly $8.20. Deep value identification and pace mapping again caught sizable overlays, including Juliet On Approach ($33.46), Me Governor ($25.32), Obstreperous ($20.10), Danneel ($18.76), Joyful Flight ($18.66) and Willy D's ($18.02).

System Integrity & Scratch Protocol

The automated scratch adjustment protocol remained a critical storyline. Mechanically moving "Next-Man-Up" alternates into active slots protected win rates and elevated unheralded horses such as Jackson Heights, Stress Relief and Borax into actionable ticket structures when favorites scratched.

3. Daily Performance Breakdown

Date

Track

Top 4 Strike Rate

Total Races

1st Pick

2nd Pick

3rd Pick

4th Pick

Avg Payout (AI Hits)

Prev. Totals (thru Oct 18)

MULTI

77.1%

2495

716

501

416

290

~$8.20

October 19

FL

100.0%

8

2

3

2

1

$7.26

October 20

FL

75.0%

8

3

1

0

2

$7.62

October 21

BAQ

87.5%

8

1

4

1

1

$7.50

October 21

FL

100.0%

8

2

4

2

0

$7.20

October 21

KEE

50.0%

8

1

3

0

0

$14.78

October 22

BAQ

75.0%

8

3

1

2

0

$5.44

October 22

KEE

66.7%

9

3

2

1

0

$7.13

October 23

BAQ

77.8%

9

2

3

1

1

$11.81

October 23

KEE

90.0%

10

4

2

1

2

$8.51

October 24

BAQ

90.9%

11

5

2

2

1

$6.00

October 24

KEE

80.0%

10

3

5

1

0

$9.84

October 25

BAQ

88.9%

9

4

2

0

2

$8.43

NEW TOTALS

MULTI

77.3%

2601

749

533

429

300

~$8.20

Note: Prior totals reflect cumulative data up to October 18; daily breakdowns from previous reports are condensed into the Prev. Totals row. Late scratches shifted runners up in the rankings across cards this week. Oct 24 Keeneland rank totals include a dead heat between the 1st and 2nd AI choices.

4. Winner Distribution by Rank (Cumulative)

AI Rank

Wins

% of Hits

Trend Note

Top Pick

749

37.2%

Pace Flow Precision: consistent mapping of race shapes produced dominant top-pick success, with flawless 100 TPN ratings for winners like Juliet On Approach ($33.46), Ground Support and Lion Lake at Belmont.

2nd Selection

533

26.5%

Class Elevation & Adaptability: a reliable safety net that captured deep value finds like Me Governor ($25.32), Obstreperous ($20.10) and Joyful Flight ($18.66).

3rd Selection

429

21.3%

Alternate Protection: benefited from "Next-Man-Up" overrides and scratch adjustments, catching overlays that rounded out the middle tiers in depleted fields.

4th Selection

300

14.9%

Rebounding Safety Net: elevated longshots in chaotic and depleted fields, catching hits that barely make the ticket and protecting exotic structures when favorites scratch.

 

Taken together, the top two selections account for 63.7% of all hits, while the third and fourth slots contribute the remaining 36.2%, showing that the back half of the ticket is where scratch protection and longshot value do their work.

5. Cumulative Statistics Summary

Metric

Stat

Description

Overall Accuracy

77.3%

(2010/2601 races) The winner landed in the AI's Top 4 in 2,010 of the 2,601 recorded races, holding stable across large field expansions and volatile weather.

Top 2 Accuracy

49.3%

(1281/2601 races) The primary or secondary choice won in nearly half of all modeled races.

Avg Payout

~$8.20

The average payout stayed resilient this week, fueled by pace-meltdown payouts and mid-priced value finds.

 

6. This Week's Structural Findings

Pace Architecture & Deep Flow Analytics

The AI's pace profiles were exceptionally sharp all week. It repeatedly identified the "Pace Beneficiaries" in its Deep Flow Analysis, pointing out stalkers and closers that capitalized on early battles, and scripted chaotic races accurately, isolating wide-trip closers or lone-speed locks depending on the projected tempo.

Scratch Protocol Validation

The mechanical rule of sliding alternates into the Top 4 after late scratches proved valuable again. Automated bumps salvaged multiple cards, moving lower-ranked horses up to win or hit the board, as with Jackson Heights, Stress Relief and Borax.

Reliability on Formful Days

When conditions were formful and logical (October 19 at FL, October 24 at BAQ and October 25 at BAQ), hit rates climbed to near-perfect levels (85%–100%). The "Class Capacity Edge" and "TS Power Edge" metrics separated true contenders from the pack efficiently.

Maiden & Juvenile Volatility

The model remains vulnerable in races with unexposed first-time starters or lightly raced juveniles stretching out in distance. It correctly penalized unproven horses on back-class and par times, but overnight improvement and pedigree-driven upsets occasionally overturned those projections.

7. AI's Final Analysis — Building the Path Forward

  1. Trust the pace read on formful cards. Where the shape is clear, treat the AI's top pick as a strong standalone anchor.
  2. Re-check the card after scratches. Alternates promoted by the Next-Man-Up rule hit the board at real prices, so rebuild tickets with the post-scratch order.
  3. Do not skip the third and fourth selections. They account for over a third of all winners and carry much of the overlay value.
  4. Be cautious with maiden and juvenile races featuring first-time starters or distance stretch-outs; consider wider tickets or smaller stakes.
  5. Use deep-value plays like Juliet On Approach and Me Governor as a reminder that second-choice and longshot slots can pay in exotics.

8. Data Notes & Caveats

  • Reporting period: December 26, 2025 – October 25, 2026; this week's additions cover October 19–25 only.
  • Prior totals reflect data through October 18, with earlier daily breakdowns condensed into the Prev. Totals row.
  • Oct 24 Keeneland rank totals include a dead heat between the 1st and 2nd AI choices.
  • Average payout is reported as approximately $8.20 cumulatively; daily figures are exact as shown.
  • Strike rate counts a hit when the winner appears anywhere in the AI's Top 4.

THE MACHINE (AKA) TrackSmart AI isn’t a tip sheet or a picks website — it’s a professional-grade handicapping engine that works directly with the Past Performances (PPs) you already use from DRF, Brisnet, and Australia's Racing and Sports. Each version of TrackSmart AI — DRF, Brisnet, Harness, and International — is custom-engineered to read and interpret every printed detail in those PPs. No guessing. No databases. No shortcuts. The AI performs the deep work for you — comparing figures, detecting trainer intent, modeling pace, and uncovering hidden value plays — with the speed and accuracy of a professional analytics team. You stay in control of your handicapping while AI handles the heavy lifting. Pace projections, class and form analysis, TPN rankings, and betting strategy — all built from the same data you’d normally spend hours studying.