ThoroughWiz
Guides · Wagering Guide v1

The $20 Ticket: Wagering Guide v1 and Its Backtest

A paper guide, written October 26, 2026, and published here as filed. It fixes three ticket templates at a flat $20.00 per race, built from the card's own selections, and then grades those templates against the settled archive of October 13–25, 2026. Nothing in the guide was wagered. The templates and every backtest figure below are reproduced unchanged from the lab's files, and the guide and its backtest are AI-produced, like everything else on this site.

Publication note: the guide's status line, when written, read "nothing here is published." It is published now, on November 15, 2026, with its text, templates and figures unchanged. The backtest report the guide points to is published in full in the second half of this page, also unchanged. This is a paper study: it records what a set of fixed tickets would have returned on races already run. It is not advice to wager, and no real-money play follows from it.

1. The frame

2. The card's vocabulary (the ticket inputs)

Every horse named below comes off the day's card and its filings — the same words the selections sheet uses:

3. Template T1 — the pace-price race

Fires when: the race has an in-band pace-price pick (S at 2/1–7/2). S is the center of the ticket; the favorite is the hedge.

BetTicketAmount
WinS$6.00
PlaceS$4.00
ShowS$1.00
Exacta ($1)S over F$2.00
Exacta ($1)F over S$1.00
Trifecta ($0.50)S / F / W and F / S / W$1.00
Win (saver)F$3.00
PlaceF$2.00
Total$20.00

Win + place on S is $10 — the largest share of the ticket. The exacta runs both ways with the favorite-on-top direction weighted smaller ($1 against $2). W, the third wheel, is the filed slot 2 when it is a different horse from S and F; otherwise it is the ML second choice among starters excluding S and F (then the third choice, and so on).

Worked example — Finger Lakes R2, October 13, 2026. S is #6 Blue Eyed Scout (3/1, the scenario's late-graph leader at a working price); the favorite is #5 Tekila (5/2); no slot 2 was filed, so the third wheel is #7 Diva Banker (7/2), the shortest price left after S and F.

4. Template T2 — the pair race

Fires when: slot 1 and slot 2 are both filed and the race is not a T1 race. The two poles share the ticket, slot 1 weighted.

BetTicketAmount
Winslot 1$6.00
Winslot 2$3.00
Placeslot 1$4.00
Placeslot 2$2.00
Exacta box ($1)slot 1 / slot 2 both ways$2.00
Trifecta ($0.50)slots 1–2 in the top two spots, both orders, with both filed backups (slot 1's backup and slot 2's backup) in third — 4 combinations$2.00
Showslot 2$1.00
Total$20.00

Worked example — Albuquerque R5, October 21, 2026. Slot 1 is #1 Mr. Biz (5/2), slot 2 is #3 Tapitboy (9/2); the filed backups are #4 Targaryis (slot 1's) and #2 Fast Service (slot 2's).

5. Template T3 — the single race

Fires when: slot 1 files alone (no slot 2, no in-band S). The anchor carries the ticket, win-heavy.

BetTicketAmount
Winslot 1$10.00
Placeslot 1$5.00
Showslot 1$2.00
Exacta ($1)slot 1 over the ML second choice$2.50
Trifecta ($0.50)slot 1 / ML second choice / slot 1's backup — the single combination$0.50
Total$20.00

Worked example — Delta Downs R2, October 14, 2026. Slot 1 is #3 Essay (6/1); the ML second choice is #2 Steady On (7/2); slot 1's filed backup is #8 Castle Island (3/1 — also the favorite).

6. The counter overlay and the multi-race ladder

Counter overlay — fires when: the race carries a filed §16 counter leg. Add a $2 win bet on the counter horse, funded by trimming the template's win amounts — T1: S's win $6 → $5 and the favorite saver $3 → $2; T2: slot 1's win $6 → $5 and slot 2's win $3 → $2; T3: slot 1's win $10 → $8. The race total stays $20.00. If the counter horse is already one of the template's win horses, the $2 stacks onto that horse.

Worked example — Belmont R7, October 25, 2026 (T1 + counter). S is #7 Central to Success (7/2); the favorite is #3 Material Girl (2/1); the third wheel is #4 Liberty's Secret (5/2) — who is also the counter leg, so the counter's $2 win and the trifecta's third spot name the same horse, independently.

The ladder: Daily Double through Pick 6. A window is a run of consecutive race numbers at one track on one card where every race in the run is a picked race in the kept universe. A window of length 2 carries a Daily Double ticket, 3 a Pick 3, and so on up the ladder. Every starting race of a long enough run carries its own ticket for each ladder it reaches (a 4-race run carries Doubles from its first three races, Pick 3s from its first two, and a Pick 4 from its first).

The headline pick of a leg is the race's pace-price horse if the race is a T1 race, otherwise its slot 1.

Funding rule. A multi-race ticket is paid for out of its first leg race's $20 — that race's single-race template is trimmed by the ticket's cost, and later legs' budgets are untouched. When several tickets start at the same race they are funded longest ladder first (Pick 6, then 5, 4, 3, then the Double). Trims come off the template in this order: the secondary place bet (T1's place on F / T2's place on slot 2), then show, then the trifecta, then the exacta, then the secondary win (T1's saver / T2's slot-2 win), then the primary place, then — last — the primary win, which is never trimmed below $2. If a charge still cannot be covered, that ticket is not played (and is counted as unfunded, never as a loss).

Order of construction for every race: build the template → any component that cannot be built from the filings (a missing backup, above all) is dropped and its amount rolls into the primary win bet → apply the counter overlay → apply multi-race funding trims.

Worked example — Finger Lakes R1–R3, October 13, 2026. Three consecutive picked races: a Pick 3 window and a Double window, both starting at R1.

Scratches and refunds (single-race tickets). The tickets are built on the filed picks. A backed horse that is scratched (in the chart's scratched list, or absent from the starters) refunds: a win/place/show bet on it is returned in full, and any exacta or trifecta combination containing it is refunded. Multi-race tickets get no substitution and no refund — a scratched selection is simply a horse that cannot win its leg, and the ticket stands or falls with its remaining combinations.

7. What the archive can settle (Step 1 inventory)

The backtest population is every picked race in the settled archive, October 13–25, 2026, in the §17 kept universe (the same filters the lab's studies use: final result, no CT/CTD/FE/GP, no synthetic, no maidens, at least 5 starters, a usable morning line):

Races
Picked races in the archive438
Excluded — no final result26
Excluded — §17 tracks (CT/CTD/FE/GP)53
Excluded — synthetic surface34
Excluded — maiden race18
Excluded — fewer than 5 starters11
Backtest population296
— T1 (pace-price)127
— T2 (pair)30
— T3 (single)139
— with a counter overlay8

What each kept race's record carries:

DataCoverage
Finish order, top 3 (results)296 / 296
Finish order, 4th (superfecta combination)287 / 296
Exacta payout + winning combination287 / 296
Trifecta payout + winning combination294 / 296
Superfecta payout + combination (no template uses it)287 / 296
Place/show prices for the 2nd/3rd finishers (results)296 / 296
Full win/place/show prices for all three finishers (Equibase charts, diagonal-validated against results)271 / 296

Multi-race pools are recorded in the results at the last leg's race (verified against 829 doubles in the archive: the double filed under race N pays the double ending at N). Windows among the kept races, and how many carry the pool's payout at the window's last leg:

LadderWindowsPayout on record at last leg
Daily Double9185
Pick 33634
Pick 4154
Pick 572
Pick 631

Settlement discipline, stated before any ticket was settled: a ticket component settles from the official payouts. A component whose price cannot be known from the record — a place bet on a horse that won in a race whose chart prices did not validate, a winning exacta whose payout is one of the 9 missing, a winning ladder ticket whose pool payout is not on record — is UNSETTLED: counted, with its stake reported, and never filled in with a zero or an estimate. A component that cannot win on the official finish order settles as a loss whether or not its pool's payout happens to be on record.

8. How this was tested

The templates above were applied mechanically to every picked race in the ThoroughWiz archive, October 13–25, 2026 (the kept universe of §2: 296 races), and settled against the official results and payouts on file — win/place/show prices (including the full win/place/show table read from the official charts), exacta and trifecta payouts, and the end-keyed daily-double and Pick 3–6 payouts. A component whose payout the archive does not carry is reported as UNSETTLED — counted, its stake shown, never scored as a loss. The full tables — per template, per component, per ladder, per day, and the two baselines — are in the backtest report, published in full directly below.

The backtest report, as filed

Wagering Guide v1 — Backtest Report. Run 2026-10-26. Paper only — nothing wagered, nothing published (at the time of the run). Templates were fixed in the guide (amounts parsed from its §9 machine-readable appendix) before any ticket was settled; no template or amount changed after any payout was seen.

Method in brief

Headline

LineSettled stakeReturnedROI
Guide overall$5,871.50$5,475.86−6.7%
Unsettled (excluded above)$48.50 stake · 23 components—0.82% of constructed

Per template (single-race components, net of ladder funding trims)

TemplateRacesComponentsSettled stakeReturnedROIUnsettled
T1 Pace-Price1271,145$2,365.50$2,143.22−9.4%14 comps · $29.50
T2 Pair30290$516.50$326.30−36.8%1 comp · $0.00
T3 Single139653$2,679.50$2,788.58+4.1%7 comps · $15.00

Per component type

ComponentBetsSettled stakeReturnedROIUnsettled
Win461$2,824.50$2,796.82−1.0%—
Place453$1,490.50$1,363.29−8.5%9 · $29.00
Show296$319.50$281.49−11.9%13 · $15.50
Exacta453$743.50$647.21−13.0%—
Trifecta425$183.50$169.31−7.7%—

Unsettled place/show components are the known archive gap: in the 25 kept races without a validated chart WPS table, a winning place-on-the-winner / show-on-the-winner-or-runner-up price is not on file. The one T2 unsettled component is a $0.00-stake remnant left by funding trims — counted for completeness, no money in it.

Multi-race ladders (each reported separately)

LadderWindowsTickets playedUnfundedSettled stakeReturnedROIUnsettled
Daily Double ($2)91910$256.00$217.76−14.9%1 · $4.00
Pick 3 (50¢)36360$25.50$0.00−100%—
Pick 4 (50¢)15150$13.00$0.00−100%—
Pick 5 (50¢)770$7.50$0.00−100%—
Pick 6 ($2)330$8.00$0.00−100%—

Every window's ticket was fundable from its first-leg race's $20 under the guide's trim order. No Pick 3–6 ticket swept its legs; the one unsettled DD is a covering ticket whose recorded combination does not reconcile to a single pool sequence (the concatenated-sequence data defect noted in the guide's §7).

Per day (all components + ladders, settled basis)

DateSettled stakeReturnedROIUnsettled stake
2026-10-13$320.00$210.78−34.1%$0.00
2026-10-14$420.00$432.50+3.0%$0.00
2026-10-15$280.00$235.21−16.0%$0.00
2026-10-16$369.00$328.10−11.1%$11.00
2026-10-17$633.00$610.28−3.6%$7.00
2026-10-18$400.00$371.99−7.0%$0.00
2026-10-19$180.00$133.50−25.8%$0.00
2026-10-20$345.50$410.58+18.8%$14.50
2026-10-21$596.00$575.13−3.5%$4.00
2026-10-22$298.00$244.47−18.0%$2.00
2026-10-23$498.00$545.57+9.6%$2.00
2026-10-24$854.00$626.42−26.6%$6.00
2026-10-25$678.00$751.34+10.8%$2.00

Baselines (same 296 kept races)

BaselineLegsWinsStakedReturnedROI
Flat $2 win on slot 1 as filed290 (6 refunds)102$580.00$605.54+4.4%
Pair-accounting headline (effective legs, production code)347114$694.00$696.28+0.3%

Harness check: the production pair-accounting code, unfiltered, reproduces the corrected published books exactly — Oct 24: 62 legs · $124 → $103.86 · −16.2%; Oct 25: 43 legs · $86 → $102.28 · +18.9%.

Memo lines

Defects found and fixed during the run

  1. Cross-date window collision (caught before any number was reported). The first run keyed multi-race windows by race id alone; race ids repeat across dates, so windows from other days were charged against — and settled into — same-id races on the wrong day (91 real DD windows ballooned to 282 played tickets). The $20-per-race assertion still passed, which is why the ticket-count-vs-window-count check mattered. Fixed by keying windows (date, race id); after the fix, tickets played equal windows exactly on every ladder (91/36/15/7/3) and every figure in this report comes from the corrected run.
  2. Template race counts in the first run's report block counted bare race ids (the same collision); the underlying stakes and returns were unaffected. Counts here are distinct (date, race) pairs: 127/30/139 = 296.

Limits

Reading the test honestly

This is one 13-day archive, graded after the fact; it is a record of what these exact tickets would have done, not a promise about any future card. The win/place foundation (§1) held up best — the win line ran −1.0% and the anchor-heavy T3 template was the only positive template — while the small exotic and ladder pieces cost what they usually cost. The counter overlay's $2 win bets (a memo line inside the templates above) went 0-for-7 with one refund on this archive.

No Pick 3–6 ticket swept its legs on this archive — at these ticket sizes that is the expected shape of the bet, and the combined ladder stake was $310.00 of the $5,920.00 constructed.

Appendix — the template definitions (machine-readable)

The backtest parses its amounts from this block; the prose above and this block are the same templates stated twice. Reproduced from §9 of the guide, unchanged.

{
  "budget": 20.00,
  "templates": {
    "T1": {"win_s": 6.00, "place_s": 4.00, "show_s": 1.00,
           "exacta_s_over_f": 2.00, "exacta_f_over_s": 1.00,
           "trifecta_total": 1.00, "trifecta_base": 0.50,
           "win_fav_saver": 3.00, "place_fav": 2.00,
           "primary_win": "win_s",
           "trim_map": {"place_secondary": "place_fav", "show": "show_s",
                        "trifecta": "trifecta_total", "exacta": ["exacta_s_over_f", "exacta_f_over_s"],
                        "win_secondary": "win_fav_saver", "place_primary": "place_s",
                        "win_primary": "win_s"}},
    "T2": {"win_1": 6.00, "win_2": 3.00, "place_1": 4.00, "place_2": 2.00,
           "exacta_box_total": 2.00, "exacta_base": 1.00,
           "trifecta_total": 2.00, "trifecta_base": 0.50,
           "show_2": 1.00,
           "primary_win": "win_1",
           "trim_map": {"place_secondary": "place_2", "show": "show_2",
                        "trifecta": "trifecta_total", "exacta": ["exacta_box_total"],
                        "win_secondary": "win_2", "place_primary": "place_1",
                        "win_primary": "win_1"}},
    "T3": {"win_1": 10.00, "place_1": 5.00, "show_1": 2.00,
           "exacta_1_over_ml2": 2.50, "trifecta_total": 0.50,
           "trifecta_base": 0.50,
           "primary_win": "win_1",
           "trim_map": {"place_secondary": null, "show": "show_1",
                        "trifecta": "trifecta_total", "exacta": ["exacta_1_over_ml2"],
                        "win_secondary": null, "place_primary": "place_1",
                        "win_primary": "win_1"}}
  },
  "counter_overlay": {"win_counter": 2.00,
    "trim": {"T1": [["win_s", 1.00], ["win_fav_saver", 1.00]],
             "T2": [["win_1", 1.00], ["win_2", 1.00]],
             "T3": [["win_1", 2.00]]}},
  "ladders": {
    "daily_double": {"base": 2.00, "legs": 2, "slot2_rule": "always_add"},
    "pick3": {"base": 0.50, "legs": 3, "slot2_rule": "add_if_headline_favorite", "cost_cap": 10.00},
    "pick4": {"base": 0.50, "legs": 4, "slot2_rule": "add_if_headline_favorite", "cost_cap": 10.00},
    "pick5": {"base": 0.50, "legs": 5, "slot2_rule": "add_if_headline_favorite", "cost_cap": 10.00},
    "pick6": {"base": 2.00, "legs": 6, "slot2_rule": "add_if_headline_favorite", "cost_cap": 10.00}
  },
  "funding_order": ["pick6", "pick5", "pick4", "pick3", "daily_double"],
  "trim_order": ["place_secondary", "show", "trifecta", "exacta",
                 "win_secondary", "place_primary", "win_primary"],
  "win_primary_floor": 2.00
}

An AI-run operation

The Daily Pace Card is run by AI. The picks, the pace sheets, and the podcast are all produced by AI. The system behind them was built and tested using many of today's leading AI systems in a strict deterministic environment set up as a handicapping lab, and it has been independently audited by other Large Language Models.