Saratoga & Del Mar · 2026 Meets · Model v0 · the anti-tout
Transparent model race-cards

Saratoga — Sun Jul 26, 2026

4 value plays today · 6 races we're passing
Edge of the Day
1MAIDA
Race 8 · 6 furlongs dirt
24%model3/1fair5/1boardWINpaid at the close
Our biggest edge on the card today (+11 pts over the board) — the price may move either way into post; you’re paid the close. Chosen by rule: the flag with the most value, graded like any other.
Public record11–35flags, W–L-24.7uflat-$2-26.9%flat-bet ROI, n=46every call hash-stamped before post →

We lose most single flags — that’s expected. The edge is price: when we’re right, the board paid more than fair. Our 30% picks win about 30% (see the calibration curve), and every call is graded against the closing odds — wins and losses.

No set price. On the tote you’re paid the closing odds, so the edge isn’t a line you beat early — it’s the gap between our fair number and where the crowd lands. We grade each flag against the actual close.

Our last miss: SAR R11 Tuthilltown, flagged at 12/1, didn’t win — on the record like every call.

Today at SaratogaPartly Cloudy, 58°F, wind 1 mph · rain 0% — no off-track flagNWS forecast
How this card is made & how to verify it

Edge Card v0 is a transparent heuristic, not a trained model: Prime-Power softmax blended 65/35 with the de-vigged morning line, plus published rules-based pace and trainer-form adjustments (the trainer angle is used as a model input; the raw figure is never published). Constants on the methodology page. Structured analysis, not proven edge; tracked publicly either way.

It reads Brisnet figures, pace, and trainer form — not jockey, trip, or track bias yet; those arrive with the trained v1 model. The trainer angle is used as a model input only; we never publish the raw figure. Fair odds are 1 ÷ our win probability.

The hold/shorten read on each flag is an early signal from its committed board price — long-priced flags the crowd leaves alone tend to hold, short ones get bet down. It’s a heuristic on a small sample; every call is graded on the record.

Model v0-heuristic-1.4 · cfg aab788cb · code 3c797c7f · takeout (SAR) win 16% / exacta 18.5% · card SHA-256 99d2c0a8398a… · Scratches as of 12:02 AM ET (lock time — no scratches posted yet; first post 13h away). Late scratches void at scoring; the card is never edited.

Powered by Brisnet data
Flags 4 of 10 races
New here? How to read a flag
Model
Our estimated win chance for the horse.
Fair
The odds that chance implies — the price where a bet breaks even.
Board
The morning line, the track's early guess at the odds.
Win bet
Bet the horse to win. Pari-mutuel pays the closing price no matter when you bet, so you're backing the horse, not a fixed price.
Hold / bet down
Whether the price should stay long or get bet down by post — how much of the edge is likely to survive.
The rest of the card
Most races. We still show our top pick and a read on it, but there's no overlay worth a bet.

Odds note: 7/2 is the same as +350.

Race 5 · 1 1/16 miles · turf · Alw 105000n1xFlag
2CHARLOTTESUNIVERSE
Model
Fair
Board
21%
4/1
6/1
our win %
break-even
the morning line
We flag it because the board (6/1) pays more than our fair price (4/1). Same horse, longer price.
The market
11%
Edge Card
21% +10 pts
▸ WIN BET · #2 to win · our fair 4/1, paid at the close
Price may move A mid board price — it can hold or shorten into post. You are paid the closing price either way, so bet the horse and take the number you get.
Ranks #1 in our power figure blend. Contested pace likely (5 early-speed types)
Data · Brisnet
The model’s read ▾

We make CHARLOTTESUNIVERSE a 21% chance in the 1 1/16 miles turf; the line implies about 11%, and that 10-point gap over the board is the flag.

It tops our power blend for the race; on the numbers it’s the most likely winner, whatever the odds say.

The race sets up with contested early pace (5 speed types), a shape that tends to favor closers.

To play it: a straight win bet on #2. Our fair is 4/1. On the tote you’re paid the closing price no matter when you bet, so you’re backing the horse, not a fixed price — the hold/shorten read above says how much of the edge is likely to survive to post.

Fair-value exotics ▾

Model-fair reference prices, not a +EV call. Exotics take ~20% off the top, so a pool is an overlay only when the tote's probable pays more than the fair return.

TicketCostHitFair returnAfter ~20%
Key exacta #2 / #1, #7 value$2.008.1%$24.83$19.86
Keyed trifecta #2 / #1, #7 / #1, #7, #5, #9 longshot$3.004.6%$65.33$52.26
Full field ▾
#HorseModelFairBoard
2CHARLOTTESUNIVERSE 21%4/16/1
1SHE TAKES CASH15%11/24/1
7LIVIN THE BREEZE15%11/25/2
5TONGUE TWISTER11%8/16/1
9MARVELOUS MADISON10%9/17/2
3FEDUPWITHTHEFED8%11/112/1
6LAST GLORY7%15/115/1
8FABULOUS FOX6%15/115/1
4FULL POUR4%30/120/1
10MIAMI STYLE3%30/130/1
Race 6 · 1 1/16 miles · turf · Md 55000Flag
7SAMARTINA
Model
Fair
Board
23%
7/2
7/2
our win %
break-even
the morning line
The board and our fair price round to the same number (7/2), so the gap is finer than fractional odds show — it is +5.7 points of win probability, not a longer price.
The market
17%
Edge Card
23% +6 pts
▸ WIN BET · #7 to win · our fair 7/2, paid at the close
Likely bet down A short board price the late money bets down. Still a win play, but expect a shorter number by post — pari-mutuel pays the closing price, so the edge is thinner here. Size it smaller.
Ranks #1 in our power figure blend. Contested pace likely (3 early-speed types)
Data · Brisnet
The model’s read ▾

We make SAMARTINA a 23% chance in the 1 1/16 miles turf; the line implies about 17%, and that 6-point gap over the board is the flag.

On the figures it’s the one we rate highest in here, so a longer price reads as the market underrating it.

The race sets up with contested early pace (3 speed types), and this one profiles to come from off it — we pushed its number up for that (×1.1).

To play it: a straight win bet on #7. Our fair is 7/2. On the tote you’re paid the closing price no matter when you bet, so you’re backing the horse, not a fixed price — the hold/shorten read above says how much of the edge is likely to survive to post.

Fair-value exotics ▾

Model-fair reference prices, not a +EV call. Exotics take ~20% off the top, so a pool is an overlay only when the tote's probable pays more than the fair return.

TicketCostHitFair returnAfter ~20%
Key exacta #7 / #6, #2 value$2.008.2%$24.50$19.60
Keyed trifecta #7 / #6, #2 / #6, #2, #8, #1 longshot$3.004.3%$69.75$55.80
Full field ▾
#HorseModelFairBoard
7SAMARTINA 23%7/27/2
6OUTRO14%6/19/2
2MAJESTIC MOONLIGHT13%13/26/1
8DEVILISH GRIN11%8/16/1
1PLAYING GAMES9%10/15/1
11JUST TELL ANNE8%12/16/1
12MAPPY7%12/112/1
5BESIDE YOU6%15/112/1
4SWEET MACKENZIE5%20/115/1
10ADDAGIRL ADDIE2%50/130/1
3PAY THE BILLS1%99/150/1
9RIYAH AL NIL1%99/150/1
Race 8 · 6 furlongs · dirt · HnrblMs-G2Flag
1MAIDA
Model
Fair
Board
24%
3/1
5/1
our win %
break-even
the morning line
We flag it because the board (5/1) pays more than our fair price (3/1). Same horse, longer price.
The market
13%
Edge Card
24% +11 pts
▸ WIN BET · #1 to win · our fair 3/1, paid at the close
Price may move A mid board price — it can hold or shorten into post. You are paid the closing price either way, so bet the horse and take the number you get.
Ranks #2 in our power figure blend; projects as the lone early speed. Lone speed
Data · Brisnet
The model’s read ▾

This is the widest gap on today’s card. We make MAIDA a 24% shot in the 6 furlongs dirt, where the morning line — take stripped out — sits near 13%. When our number is 11 points clear of the board, that disagreement is the whole reason to be here.

It sits second on our power blend, close enough to the top that the price still looks generous.

It profiles as lone speed — likely uncontested on the front end, a real edge we bumped it for (×1.15).

To play it: a straight win bet on #1. Our fair is 3/1. On the tote you’re paid the closing price no matter when you bet, so you’re backing the horse, not a fixed price — the hold/shorten read above says how much of the edge is likely to survive to post.

Fair-value exotics ▾

Model-fair reference prices, not a +EV call. Exotics take ~20% off the top, so a pool is an overlay only when the tote's probable pays more than the fair return.

TicketCostHitFair returnAfter ~20%
Key exacta #1 / #6, #4 value$2.0016.4%$12.21$9.77
Keyed trifecta #1 / #6, #4 / #6, #4, #2, #3 value$3.0014%$21.47$17.18
Full field ▾
#HorseModelFairBoard
6GRAND JOB41%3/21/2
1MAIDA 24%3/15/1
4HAULIN ICE12%7/14/1
2KAPPA KAPPA10%9/110/1
3STRIKER HAS DIAL8%12/110/1
5GOING STEADY4%25/130/1
7DO GOODER2%50/130/1
Race 10 · 1 3/8 miles · turf · Md Sp WtFlag
10PHANTOM BLUE
Model
Fair
Board
33%
2/1
5/2
our win %
break-even
the morning line
We flag it because the board (5/2) pays more than our fair price (2/1). Same horse, longer price.
The market
22%
Edge Card
33% +11 pts
▸ WIN BET · #10 to win · our fair 2/1, paid at the close
Likely bet down A short board price the late money bets down. Still a win play, but expect a shorter number by post — pari-mutuel pays the closing price, so the edge is thinner here. Size it smaller.
Ranks #1 in our power figure blend. Even pace
Data · Brisnet
The model’s read ▾

PHANTOM BLUE is one of the more likely winners we flag today — a 33% chance in the 1 3/8 miles turf the board still prices like a longer shot (about 22% once you strip the take). The gap is smaller in points (11) but it’s on a horse we genuinely rate to win.

Our power-rating blend makes it the best of this field — which is why the board’s price looks a step generous.

To play it: a straight win bet on #10. Our fair is 2/1. On the tote you’re paid the closing price no matter when you bet, so you’re backing the horse, not a fixed price — the hold/shorten read above says how much of the edge is likely to survive to post.

Fair-value exotics ▾

Model-fair reference prices, not a +EV call. Exotics take ~20% off the top, so a pool is an overlay only when the tote's probable pays more than the fair return.

TicketCostHitFair returnAfter ~20%
Key exacta #10 / #7, #9 value$2.0015.2%$13.12$10.50
Keyed trifecta #10 / #7, #9 / #7, #9, #2, #8 value$3.009.9%$30.37$24.30
Full field ▾
#HorseModelFairBoard
10PHANTOM BLUE 33%2/15/2
7WHITETHORN18%9/22/1
9MARY LOIS13%7/16/1
2AMERICAN DEBUTANTE10%9/18/1
8ONEHOTMESS8%11/112/1
5DON'T LOOK NOW5%20/112/1
1EENYMEANYMIGHTYMO5%20/112/1
4ISADORA DUNCAN4%20/112/1
6ABUNDANT LOVE2%30/115/1
3RAGING BLOOM1%50/130/1
Multi-race tickets the whole card, keyed on our reads
Late Pick 4 (R7–R10)
BuildCostHitFair returnAfter ~20%
Conviction$0.500.3%$164.34$131.47
R7 #3 · R8 #1 · R9 #4 · R10 #10
Coverage$2.000.86%$231.86$185.49
R7 #3, #4 · R8 #1 · R9 #4, #5 · R10 #10
Late Pick 5 (R6–R10)
BuildCostHitFair returnAfter ~20%
Conviction$0.500.07%$715.13$572.10
R6 #7 · R7 #3 · R8 #1 · R9 #4 · R10 #10
Coverage$2.000.2%$1008.96$807.17
R6 #7 · R7 #3, #4 · R8 #1 · R9 #4, #5 · R10 #10
Early Pick 5 (R1–R5)
BuildCostHitFair returnAfter ~20%
Conviction$1.000.07%$1427.80$1142.24
R1 #6 · R2 #5 · R3 #2 · R4 #3, #9 · R5 #2
Coverage$16.000.73%$2203.80$1763.04
R1 #6, #4 · R2 #5, #4 · R3 #2, #1 · R4 #3, #9, #4, #2 · R5 #2
Pick 6 (R5–R10)
BuildCostHitFair returnAfter ~20%
Conviction$0.200.01%$1377.91$1102.33
R5 #2 · R6 #7 · R7 #3 · R8 #1 · R9 #4 · R10 #10
Coverage$0.800.04%$1944.04$1555.23
R5 #2 · R6 #7 · R7 #3, #4 · R8 #1 · R9 #4, #5 · R10 #10
Rolling Pick 3 (R4–R6)
BuildCostHitFair returnAfter ~20%
Conviction$1.001.56%$63.91$51.13
R4 #3, #9 · R5 #2 · R6 #7
Coverage$2.002.79%$71.58$57.26
R4 #3, #9, #4, #2 · R5 #2 · R6 #7
Rolling Pick 3 (R5–R7)
BuildCostHitFair returnAfter ~20%
Conviction$0.500.78%$64.46$51.57
R5 #2 · R6 #7 · R7 #3
Coverage$1.001.34%$74.54$59.63
R5 #2 · R6 #7 · R7 #3, #4

Whole-card tickets keyed on our reads — a tight conviction build and a wider coverage build, singling the flags and spreading the open races. Reference construction, not a +EV call: the hit % assumes the legs are independent (v0 doesn’t model pace or scratches across races), and only the flagged legs are graded bets.

The rest of the card our read on every race — analysis, not a bet
R1Chalk6SHOWA20%4/13/1
No edge on our top pick — the market has it priced about right.
Full field ▾
#HorseModelFairBoard
6SHOWA20%4/13/1
4HIGH OFFICIAL17%5/17/2
2GENEVA LAKE16%11/24/1
8JERSEY GUY16%11/24/1
3HARBAUGH11%8/16/1
1KIPROS8%11/18/1
5LE PATRON6%15/112/1
7BLUE LAW6%15/112/1
R2Lean5SON OF A BIRCH25%3/15/2
Our top-rated horse. A small edge, under our bet line.
Full field ▾
#HorseModelFairBoard
5SON OF A BIRCH25%3/15/2
4VACATION DANCE16%5/19/2
6SHEFFLIN15%6/13/1
8NIAGARA SKYLINE14%6/16/1
2FLUID SITUATION10%9/16/1
7DOCTOR JEFF9%11/18/1
3TRUST ISSUES6%15/120/1
1BILLIONAIRES ROW5%20/18/1
R3Chalk2DON'T BE SALTY20%4/13/1
No edge on our top pick — the market has it priced about right.
Full field ▾
#HorseModelFairBoard
2DON'T BE SALTY20%4/13/1
1BEIRA18%9/24/1
3KYLE'S MOM13%7/19/2
7KATIE KING12%15/28/1
8ROYAL BOBBIE12%15/25/1
5ANGEL BELLA12%15/24/1
4RACING COLORS8%11/110/1
6BOURBON N LACE5%20/112/1
R4Scramble3EMIRATES ROAD17%5/15/2
Wide-open heat; no separation we'd trust.
Full field ▾
#HorseModelFairBoard
3EMIRATES ROAD17%5/15/2
9COUNTERSPY16%11/29/2
4ALTERNATE REALITY14%6/16/1
2FORT NELSON12%15/27/2
8BIG BLUE LINE10%9/18/1
7CASTLE CHAOS9%10/18/1
5FRIZZANTE8%11/18/1
1NOTED7%12/115/1
6CELESTIAL GLAZE7%15/115/1
R7Chalk3WILLINTORISKITALL16%5/17/2
No edge on our top pick — the market has it priced about right.
Full field ▾
#HorseModelFairBoard
3WILLINTORISKITALL16%5/17/2
4GOLDEN SYMPHONY12%15/26/1
10ART FAIR11%8/18/1
7SOUTH BEND TOM10%9/19/2
11EGYPTIAN10%9/110/1
6DISCOTHEQUE9%10/16/1
1MOMENTUM FILES8%11/18/1
2HOT PROPERTY7%12/18/1
8MASTER FREUD7%12/110/1
9BEARY FUNNY6%15/120/1
5CONNIVING3%30/130/1
R9Chalk4SENIOR OFFICER24%3/19/5
No edge on our top pick — the market has it priced about right.
Full field ▾
#HorseModelFairBoard
4SENIOR OFFICER24%3/19/5
5SACROSANCT15%11/26/1
2BRING THE SMOKE10%9/18/1
8TIP TOP THOMAS10%9/17/2
7OVER AND OLLIE8%11/110/1
6IGNITE THE LIGHT8%11/115/1
9ACOUSTIC AVE8%11/18/1
3TARANTINO8%12/115/1
10LIGHT MAN5%20/120/1
1STARS AND STRIDES4%25/112/1