Skip to content
Trainer profile

Padraig Roche

1,050 runners in the lifetime window · 123 individual horses
As at 5 Nov 2026 · KTB archive: 2005–present, updated daily
Last runner 2 days ago
from 1 Jan 2008 — the aggregate floor: every horse in the window has a complete captured career behind it
95–1,050
Wins / Runs
9%
Strike rate
n=1,050
265 (25%)
Placed
1.05
n=1,050 · 0.84 incl. margin
−222.28
£1 level stake
n=1,050 at SP · −21.2%
£1.58m
Prize money (window)
Strike by season
3.9%
KTB Yard Profile

Sends out around 60 runners a season, winning with 9% of them — beating the margin-free market read across the lifetime window (Fair A/E 1.05). Horses with headgear back on after a bare run strike at 12% (43 runs).

Deterministic profile computed from 1,050 archived runs (lifetime window) · updated 5 Nov 2026
KTB at a glance Fresh horses: 9.2% 25% when favourite Aidan Kelly: 9.3% from 215 New recruits: 9.8% Headgear returns: 11.6% Claiming riders: 9.5% Sends off the favourite in 10% of races
The read

The record is strongest at Fairyhouse (16.9%, n=83), clear of the yard's 9% baseline. Debutants struggle at 1.4% (n=72, Fair A/E 0.37). Horses returning fresh after 61+ days are the standout value angle: 9.2% strikes only just above baseline, but Fair A/E 1.22 and +26.74 to a £1 stake over 283 runners — a record the market has persistently underpriced. Aidan Kelly rides 20% of the entries and matches the yard's average almost exactly (9.3%, A/E 1.31) — the partnership beats the margin-free market read. The record reads poorly over 1m – 1m1f — 2.4% over 41 runners, 6.6pts below the baseline. 6 Group and Graded wins sit in the archive record — the biggest prize the yard has landed: the Ballymore Champion Four Year Old Hurdle at Punchestown (May 2026, £73,625). The stable star of the record: Walking On Glass (IRE) — 8 wins from 69 runs for the yard.

Written from the figures on this page — every claim carries its sample. Descriptive, not a tip.

Yard patterns

the angles behind the record · vs 9% baseline
Fresh · 61+ days off
9.2%+0.2
n=283 · A/E 1.22 · £1 +26.74
New recruits · first run for the yard
9.8%+0.8
n=51 · A/E 1.37 · £1 +31.00
Debutants · first career start
1.4%−7.6
n=72 · A/E 0.37 · £1 −38.00
First-time headgear · true-first definition
2.6%−6.4
n=39 · A/E 0.38 · £1 −10.00
Headgear returns · back on after a bare run
11.6%+2.6
n=43 · A/E 1.44 · £1 +12.50
When favourite
25%
n=104 · A/E 0.97 · £1 −22.73
Claiming riders · a claimer in the saddle
9.5%+0.5
n=464 · A/E 1.13 · £1 −63.02
Group / Graded races · pattern & listed
9.5%+0.5
n=63 · A/E 0.96 · £1 −28.12

Each tile is the record in that situation vs the 9% baseline — descriptive splits with the sample always shown, for you to weigh. Tiles under 30 runs render muted. Headgear figures use the true-first definition (first gear appearance in the captured career); returns are gear back on after a bare run. Claiming riders are runs where the booked rider carried a weight claim. Group / Graded covers pattern and listed races. The same beliefs, tested across every yard and every GB and Irish race since 2005: Racing myths.

Notable angles

vs their own baseline
With Mark Walsh 20.4% vs 9% · n=147
With C&D winners 22% vs 9% · n=50
On last-time-out winners 17% vs 9% · n=94
At Fairyhouse 16.9% vs 9% · n=83
With debutants 1.4% vs 9% · n=72
Over 1m – 1m1f 2.4% vs 9% · n=41
Descriptive splits ranked by divergence × sample — insight, not instruction. Every angle carries its N; nothing under 30 runs qualifies. Rows within 2pts of the baseline appear only when the market-relative read (Fair A/E) diverges, and show it.

Record by conditions

same window as header

Race type

TypeRWSR£1
Hurdle 504 46 9.1% −177.84
Chase 273 34 12.5% +26.53
Flat 250 13 5.2% −54.67
NH Flat 23 2 8.7% −16.30

Going

GoingRWSR£1
Soft 402 36 9% −91.20
Good 278 25 9% −55.90
Heavy 177 13 7.3% −86.47
Good to Soft 151 18 11.9% +0.79
Standard (AW) 24 1 4.2% +5.00
Good to Firm 18 2 11.1% +5.50

Courses

CourseRWSR£1
Punchestown 146 9 6.2% −83.09
Naas 101 4 4% −77.37
Fairyhouse 83 14 16.9% +23.00
Navan 67 3 4.5% −36.00
Curragh 60 3 5% −39.67
Limerick 49 5 10.2% −27.30
Galway 48 2 4.2% −4.00
Wexford 47 7 14.9% −6.52
Leopardstown 46 4 8.7% +54.83
Gowran Park 42 4 9.5% −7.00
Thurles 40 5 12.5% −4.00
Kilbeggan 37 5 13.5% −4.59
Clonmel 31 3 9.7% −9.07
Listowel 28 5 17.9% +13.50
Tipperary 28 1 3.6% +6.00

Class

ClassRWSR£1
Class 1 67 7 10.4% −21.12
Class 2 2 0 0% −2.00
Class 3 1 0 0% −1.00
Class 4 1 0 0% −1.00

Market position

In the bettingRWSRA/E£1
Favourite 123 28 22.8% 0.93 −32.23
2nd–3rd fav 208 33 15.9% 1.06 −30.55
4th–6th fav 233 19 8.2% 1.13 −45.50
7th+ in betting 486 15 3.1% 1.15 −114.00
Fair A/E per band is the tell: near 1.00 the market had it priced right; well above, the wins came where the market didn't expect them.

Rows below n=30 are shown faded — a small sample is a fact, not a trend. Every figure carries its N.

Jumping record

completion over obstacles · same window as header
85.8% completed · 777 obstacle runners (chases & hurdles)
CodeRCompFURPUBDOth
Chase 273 78% 7 11 41 1 0
Hurdle 504 90.1% 12 4 34 0 0
Chases and hurdles only — bumpers have no obstacles, so completion is not a question there. F fell · UR unseated rider · PU pulled up · BD brought down · Oth covers refusals, slip-ups, run-outs and the rare disqualification. Non-runners are excluded throughout. Rows under n=30 are shown faded. A descriptive record, not a tip.

Horses trained

same window as header · most-raced first
HorseRWPlLast ran
Walking On Glass (IRE) 69 8 22 30 Apr 26
Outside The Door (IRE) 41 7 14 4 Jul 26
Brazil (IRE)from A P O'Brien 32 4 6 23 Nov 25
Back Down Under (IRE) 28 2 14 10 Oct 26
Bay Hill (IRE)from C Roche 26 5 12 29 Jul 21
The Virginian (IRE)from G M Lyons 25 1 3 24 Aug 25
Ernie From Nurney (IRE) 23 2 6 7 May 26
Marshalled (IRE)then Ben Haslam 22 4 9 29 Dec 22
Demi Plie (IRE)from C Roche 22 5 13 31 May 21
Alice Johnston (IRE)from C Roche 21 1 7 22 Oct 20
One Horse Army (IRE) 20 3 4 16 May 26
Steer Clear (IRE)from C Roche 19 2 5 3 Oct 21
Showing the 60 most-raced of the 123 horses sent out in this window — each ran at least 6 times. Horses with fewer runs aren't listed; search finds any horse since 2005. Runs, wins and places only: a strike rate or a return on a handful of runs is noise, so none is quoted per horse — each horse's full record is on its own page.
Where they came from · where they went next

Of the 60 horses listed, 30 arrived from another yard in our archive, most often from C Roche (11) and A P O'Brien (3); 30 had their first archived run for this yard.

14 have since run for another yard, most often for Ben Haslam (3) and Ross O'Sullivan (2); 46 have no later run for another yard in our archive.

Read from each listed horse's whole archived career (2005 onwards), not only this window, and counted over the horses listed above, not the whole string. “First archived run” means our archive holds nothing earlier, not that the horse was unraced. A horse with no later run elsewhere may still be here, be off the track, or have run somewhere our archive does not cover — none of that is asserted.

Current form

significance-gated

Last 14 days

Small sample
0 from 3
Only 3 runners in the last 14 days — too few to read against the baseline (gate: n≥15). The card states the sample and stays neutral.

When the yard fires

strike by month, lifetime
JFMAMJJASOND
71571081361271484
A jumps yard: the season runs Oct–Apr. Muted months are small samples (under 30 runs) — read them with the caution they deserve.

Jockey bookings

same window · biggest samples first
Go-to riderAidan Kelly
20% of all rides
9.3% SR · A/E 1.31
RiderRWSR£1
Aidan Kelly 215 20 9.3% −15.15
Mark Walsh 147 30 20.4% +33.86
Peter Smithers 116 8 6.9% −47.40
Jody McGarvey 104 4 3.8% −65.50
Leigh Roche 78 3 3.8% −36.00
Miss A B O'Connor 59 7 11.9% −19.42
Darragh O'Keeffe 41 4 9.8% −23.25
Simon Torrens 35 2 5.7% +12.50
Declan McDonogh 32 1 3.1% −24.00
Muted rows are under 30 rides and sit behind "Show all".

Career highlights

lifetime archive, 2018→
95
Archived wins
6
Pattern / graded wins
£1.58m
Prize money
2024
Best season · 16 wins
Saratoga (IRE) — Ballymore Champion Four Year Old Hurdle (Grade 1) Punchestown · May 2026 · Grade 1
£73,625
Brazil (IRE) — Guinness Handicap (Premier Handicap) Galway · Aug 2023
£64,900
Saratoga (IRE) — McCoy Contractors Juvenile Handicap Hurdle (Premier Handicap) (Fred Winter Juvenile Hurdle) (GBB Race) Cheltenham · Mar 2026 · Class 1
£45,016

Sires in the string

progeny record · win % by ground and trip
Flemensfirth (USA) 10.3% · 12–116
Progeny have won 11.9% of 17,287 runs since 2008 (1,595 horses).
Big Bad Bob (IRE) 11.6% · 8–69
Progeny have won 8.8% of 4,568 runs since 2009 (314 horses).
Milan (GB) 12.1% · 8–66
Progeny have won 11.3% of 16,871 runs since 2008 (1,463 horses).
Galileo (IRE) 10.9% · 5–46
Progeny have won 14.5% of 18,201 runs since 2008 (1,776 horses). Strongest over 6f – 7f (all codes): 18.2% from 1,383. Weakest over 2m5f – 3m (all codes): 9.4% from 890.
Walk In The Park (IRE) 11.1% · 5–45
Progeny have won 13.4% of 7,030 runs since 2011 (704 horses).
Leading Light (IRE) 3.3% · 1–30
Progeny have won 8.2% of 2,214 runs since 2018 (226 horses). Weakest on Heavy going: 4.6% from 241.
Top line: the yard's own record with each sire's runners (win %, wins–runs), same window as the header; faded under 30 runs. Beneath it: every run by that sire's progeny in our archive since 2008, as at 5 Nov 2026 — what they have done, not a forecast. A going or a distance is named only where at least 100 progeny runs sit 3 or more points from the sire's own win rate. Distance bands cover Flat and jumps together.

The record

latest 50 of 1,050 · newest first

The latest 50 of 1,050 archived runners, newest first. The full record runs from 13 January 2018 to 3 November 2026.

DateCourseHorseRaceFinishSP

Drawn from the whole archive (2005 onwards), so it can include runs outside the window the header statistics cover. Non-runners are not listed. A finish shown in letters is a non-completion (F fell · UR unseated · PU pulled up · BD brought down).

KTB’s figures

the forward record · since 5 Oct 2026

KTB’s figures on this yard’s runners since 5 Oct 2026: the KTB Selection in 3 runs (won 0) · top-rated on KTB Rating in 1 run (won 0) · fastest on KTB Speed in 1 run (won 0) · to 3 Nov 2026.

Each pick was a KTB figure’s choice before the off — the published KTB Selection, the top-rated horse on KTB Rating, the fastest on KTB Speed — settled against the result the next morning and never worked out again; “did not run” is counted apart. GB and Irish races only. A record of weeks so far: read it as a count, not a verdict.

Quick answers

What is Padraig Roche's strike rate?
Padraig Roche has won 95 of 1,050 runners since 2008, a strike rate of 9% (as at 5 November 2026). Figures are computed from our own results archive.
Which course does Padraig Roche do best at?
Fairyhouse by strike rate: 14 wins from 83 runners (16.9%) since 2008. Read from the 15 courses with the most runners on the same basis, and only those with at least 30 — a small sample cannot top the list, and a rarely-visited course is not in the comparison.
Who rides most for Padraig Roche?
Aidan Kelly, with 215 runners since 2008 — 20% of the total. That pairing has won 20 (9.3%).
Is Padraig Roche in form?
Too few runners in the last 14 days to read — 3 runners, which is not a sample we will draw a conclusion from.
What is Padraig Roche's Fair A/E?
1.05, from 1,050 runners with a computed fair probability since 2008. Fair A/E compares actual wins with a margin-stripped market expectation, so 1.00 is par: above it the wins came at bigger prices than the margin-free market implied, below it they did not.
Fair A/E compares wins with a margin-stripped market expectation. Favourite-heavy samples — top yards, leading riders — naturally read a few points warm, so 1.05 means "beat the margin-free market read", not "the market underrates this by 5%". Three sample sizes are reported and never blended: runs (n), the ROI basis (n at SP), and the fair-A/E basis (n with a computed fair probability).

Strike rate is wins over all runs (n). ROI is £1 level stakes at SP over runs with a parseable price (n at SP). Fair A/E is wins over margin-stripped market expectation, over runs with a computed fair probability (n fair). The three Ns can legitimately differ and are always shown. Figures as at 5 November 2026, computed from our own results archive (2005 onwards; aggregates from 2008).