Before you scroll any further, there's one question I'd like you to answer in your head.
What would you do with an extra £250 landing in your current account by teatime today?
Not next Saturday, and not on the afternoon your lucky horse finally turns up, but today.
I ask because that question used to be pure fantasy for me, the sort of thing you daydream about on a slow train home.
These days a figure like that reaches my account on a regular basis, and it's been doing so for well over a year.
It won't happen on every day of the week, and I'd never pretend that it does.
But it happens on enough days, and in big enough amounts, that a betting bank which began at a modest £1,000 has grown into something that's reshaped how I think about money.
I've never owned a racehorse, ridden one or worked in a training yard
What I have done is spend most of my working life teaching computers to spot trouble before it happens, and for the last 3 years I've pointed that skill at British and Irish racing.
I'm a 52-year-old former data analyst from Hertfordshire who got tired of losing, and decided to do something about it with the tools I knew best.
It's Tuesday the 17th of November 2026 as I write this, and the jumps season is now in full cry after the last of the turf Flat racing wound up at Doncaster.
The app has worked through both codes this year without missing a beat.
Over £63,500 In Profit Across The Last 12 Months
September closed at £5,875.35, with the St Leger meeting at Doncaster leading the way.
August, with its packed weeks of festival racing, brought in £5,215.65
October came in lighter at £3,562.15, by design rather than bad luck, and I'll show you the reason for that shortly.
And across the first 7 days of November, the app has already put another £1,487.60 back into my account.
That came during the first big jumps weekends of the winter, which is the time of year the app tends to enjoy most.
Right now, with the winter programme building, the app is showing no sign at all of easing off.
Over the next few minutes I'll show you how this came about, and why it keeps working when every other approach you've tried has fallen apart.
Then I'll show you how to start copying every bet I place from tomorrow morning.
Before any of that, I want to explain why so many decent judges of a horse never see a regular profit.
Why Regular Profits Are Out Of Your Reach
Most punters reading this page aren't losing for lack of knowledge, and I'd like to make that clear from the outset.
You know your racing, and you probably know it better than most of your friends do.
You understand the form, and over the years you've paid good money for advice that promised the earth and delivered very little.
The real problem is structural rather than personal
I spent 15 years working inside an insurance company, and the people who made money there thought about risk in a way almost no punter ever does.
An underwriter never asks whether one particular house will burn down this winter.
He applies one fixed method to thousands of policies, and he lets the numbers settle out over the year.
An underwriter also refuses to fall in love with a policy
A nice garden or a friendly voice on the phone counts for nothing with him, and what matters is what the figures say about houses like it.
Punters fall in love with horses all the time, and the bookmakers have built very comfortable businesses on the back of it.
He knows a few of those houses will catch fire, and he's already allowed for it in the price, so a bad week doesn't send him into a panic.
The average punter does the reverse, putting everything on one opinion about one race and judging himself on whether that one horse wins.
When it loses, he tears up his method and starts again with a new one the following week.
When it wins, he grows bolder and starts backing horses in races he hasn't studied at all.
Neither habit can ever add up to a regular profit, however much racing knowledge sits behind it.
I know this because I was that punter for the best part of 20 years, and I'll tell you what those years cost me before long.
Most Punters Only Ever Read A Fraction Of The Story
There's a second problem, and it comes down to sheer volume.
Every race run in Britain and Ireland goes into the public record, from the big Saturday handicaps to a Monday evening seller at Southwell.
Each runner drags years of history behind it, covering every trip it has tried and every rider who has ever sat on its back.
Most punters glance at the last 3 runs on the race card and settle on a view inside a couple of minutes.
On a busy Saturday there can be more than 300 declared runners across Britain and Ireland, and even a retired man with all day to spare couldn't give each of them 5 minutes.
Add up a full week and you're looking at well over 1,000 runners, every one with a history that deserves reading
That's the plain reality of betting on horses today, and no amount of system-hopping is going to shift it an inch.
Everyone, the bookmakers included, leans on the obvious clues, such as the fashionable trainer or the yard with a tidy record at the track.
The runner nobody is talking about, the one whose chance is hidden in a mix of less obvious numbers, slips straight past them.
These are the races where a horse is much better suited to the day's conditions than anyone has noticed.
That's Where Monthly Profit Lives
You can't find those runners by hand.
Not reliably, and certainly not across a full day of racing spread over 5 or 6 meetings.
A well-built AI betting app can read every one of them, and it has the job finished before 7 o'clock in the morning.
It's entirely possible to pull real, repeatable money out of betting on the horses
The people doing it aren't reading the same tips column as you, and they're not chasing services that always seem to go silent the moment the wheels come off.
They have a systematic, computational advantage, and today I'm offering you a way to borrow mine.
That word borrow matters, because you won't have to build or learn a thing to use it.
My own mornings look nothing like they used to
I'm usually walking back from the newsagent in Harpenden with the paper under my arm when the app finishes its run.
By the time I've read the back pages, the selections are sitting in my inbox, and the bets are on a few minutes later.
Then I leave them alone until the evening, which is something the old me would have found impossible.
So before I tell you anything else, let me tell you what's been happening in my own corner of the world.
At the start of 2026 I let 8 ordinary people follow my daily bets as a real-money test
I wanted to know whether the results would stand up for people with no background in data or racing analysis.
I picked them from readers who'd answered a short post I'd written on a racing forum, choosing a mix of ages and a mix of experience.
The youngest was 29 and the oldest was 68, and only 2 of them had ever bet more than £20 on a race.
None of them were given any special treatment, and they received the email at the very moment new members will.
10 months on, all 8 of them are still following the bets.
I asked each of them to write to me after their first few weeks, and I haven't changed a word of what they sent.
Here's What The First Four Of Them Told Me
"£1,326.80 in 27 days, staking £10 a bet and never once going above it. I've had dearer hobbies over the years, but none of them ever paid me back."
Barry Lomax, Wigan
"George, I read your first email 3 times before I dared place a bet. 6 weeks later I'm £1,148.35 ahead, and these days I only need to read it once."
Janet Ellery, Whitby
"For years my betting ran on whatever my brother-in-law fancied after a few pints on a Friday night. The app has put £1,571.90 in my account since February, and I've stopped listening to his Friday tips altogether."
Russell Hambly, Bodmin
"The thing nobody warned me about was how little of my morning it would take, because I'm finished before the 8:15 bus goes past the window. That's £1,203.15 up after my first month, and a week in Pitlochry booked on the strength of it."
Moira Fenwick, Alnwick
How Did They Make That Kind Of Money?
Not because they're a room full of seasoned betting experts, because they certainly aren't.
None of them had made serious money from betting before they joined.
My AI betting app does the hard reading for them before most of them have opened the curtains.
All they do is copy what the email tells them, at stakes they're comfortable with, and then get on with their day.
Anyone Over 18 Can Make This Work
Age, background and racing knowledge make no difference to the email, which reads the same for a retired engineer as for someone who's never set foot on a racecourse.
If you're like most punters up and down the country right now, I suspect a few of these will land close to home.
You're sick to the back teeth of chasing newspaper tipsters and online services first thing every morning.
You've tried the form guides and the fancy systems sold off glossy websites.
But whatever you try, you can't crack it.
Every time you stumble onto a run of form, something goes wrong and it all collapses.
The Big Bookmakers Always Seem To Win
And you're left topping the account up again out of your wages, promising yourself it'll be different this time round.
I know how that feels, because I lived every bit of it for a very long time.
And you already know that blindly backing the short-priced favourites most tipsters push is a slow road to precisely nowhere.
Favourites win their share of races, but the prices on them are cut so fine that the winners rarely pay for the losers.
£63,521
Nov 2025 to Jan 2026£15,281.55
Feb to Apr 2026£17,311.05
May to Jul 2026£16,275.50
Aug to Oct 2026£14,653.15
Why Most Tipsters Eventually Let You Down
The trouble with most tipster services isn't that the person behind them doesn't know their racing.
The problem is that knowing racing isn't enough.
Human beings are emotional creatures who wobble under pressure
A tipster on a good run is bold and decisive, happy to back a selection at a decent price.
The same tipster after a couple of rough weeks starts second-guessing every call, drifting toward shorter prices and abandoning the very rules that made him profitable in the first place.
He'll call it refining the process, but what it really is, is fear leaking into the selections.
I saw the same thing in insurance, where an underwriter who'd just paid out on a big claim would turn away perfectly good business for months afterwards.
One bad memory was steering decisions that should have been made by the figures.
Tipsters are no different, and the worse the run gets, the more their selections start to look like an apology for the last one.
And then there's a commercial problem that almost nobody stops to think about.
A Tipster's Income Comes From Subscriptions, Not From Your Winning Bets
Which means his incentive is to keep you paying month after month, not to hand you the sort of profit that actually changes your life.
Once you see it that way, the endless monthly fees and the habit of blaming bad luck after every losing run start to make a lot more sense.
My AI betting app carries none of these weaknesses on its back
It works through every declared runner at every meeting each morning, giving the opener the same attention as the finale.
It has no subscriber count to protect, and no losing run it feels the need to hide.
There's just the data going in one end and the selections coming out the other.
A bad Thursday leaves no mark on Friday's selections, because the model has no memory of feeling hard done by.
Building something like that yourself is another matter entirely.
You've no interest in spending months learning to write code or stitch together machine learning models from scratch.
There's a long way between wanting a dependable advantage over the bookmakers and actually holding one in your hand.
This is where my AI betting app walks onto the stage
Built over years of patient trial and error, and no small amount of stubbornness.
I've Built A System That Finds Winners Most Punters Never Notice
Producing Over £63,500 In Profit Across The Last 12 Months
And it's only getting better as the months roll by.
Each new result adds to its history, and it tightens itself a fraction with every prediction it makes.
Let me be clear about what you're getting, so there are no surprises later on.
This isn't a get-rich-by-Friday scheme, and some weeks finish slightly down before the month comes good.
What it gives you is a set of bets each morning that have been through more checking than any human tipster could manage, backed by a record that's held up for over 2 years.
If you can follow the email for a full month without second-guessing it, the figures in this letter are what the method has produced.
Before I tell you any more, let me tell you a little about who I actually am
My name's George Davis, and I was born in St Albans in the spring of 1974.
I've been wrapped up in horse racing since I was barely tall enough to see over the counter of a betting shop.
And a betting shop is where this story starts, because my uncle Ron worked behind one for the best part of 30 years.
He was a settler in the days before everything went digital, the fellow who worked out what every winning slip was owed in his head, at a speed that made your eyes water.
Hand him a Yankee with 4 winners at awkward prices and a dead heat in the third leg, and he'd have the answer before the customer had reached the counter.
He didn't own a calculator, and the area manager once timed him against one on a complicated bet and lost
The busiest day of Ron's year was Grand National Saturday, when the queue ran out of the door and halfway to the chemist.
People who hardly bet at any other time put £2 each-way on a name they liked, and Ron would settle every slip by teatime.
He told me once that the National was the day the shop made friends, and the other 364 days were the days it made money.
On a Saturday my mum would drop me at the shop while she did the market, and Ron would perch me on a stool in the corner with a lemonade and a stack of old race cards.
I learnt to read a race card before I could read a newspaper, and Ron would test me on the runners while the commentary crackled over the speaker.
He'd cover the jockeys' names with his thumb and ask which horse had won over the trip before, and I only got a fruit pastille if I answered right.
By the age of 9 I could tell you which runners were carrying a penalty, without the faintest idea what a penalty was.
I watched that shop from the inside for years, and it taught me something it then took me decades to understand.
The house was never just lucky
The house had the numbers, and the numbers never once blinked.
Ron used to say the punters walked in chasing a feeling, while the shop stood there armed with arithmetic, and the arithmetic usually won.
Each Monday the manager would pin up a sheet showing what the shop had taken and what it had paid out the week before.
Ron once pointed out that across a whole year, the difference between those two columns barely moved, however wild the individual Saturdays had been.
Big winners came and went, a few punters walked out with fat envelopes, and still the shop kept roughly the same slice of every pound it took.
That was my first lesson in what happens when you stop judging each result on its own and look at a thousand of them together.
He also taught me that the punters who did well were never the loudest men in the shop.
A couple of regulars came in with a page of their own figures, placed their bets without a word, and left before the first race.
Ron watched them for years and never once saw either of them chase a loss
One of them, a retired surveyor called Mr Dellow, told me years later that he never backed a horse he couldn't explain in one sentence.
The other was a bus driver who only ever bet on Wednesdays, as that was his day off, and he'd spent the other 6 days studying.
I didn't understand what he meant until I'd lost a fair sum of my own money finding out.
I loved the buzz of the place, and the roar from the television as the field swung for home.
What I hated was watching decent people trail back out onto the pavement with their pockets turned inside out, weekend after weekend.
Ron retired in 2004, and the shop he'd worked in became a mobile phone shop the following spring.
For Years, I Walked That Very Same Path
I chased tips that weren't worth the newsprint they were smudged onto.
I paid for tipping services that started like a house on fire and fell to pieces inside a month
I threw good money after bad, forever telling myself I was one decent week away from finally turning the corner.
Looking back, the pattern was obvious, as I changed my approach every time it hurt and never stayed with anything long enough to find out if it worked.
I went through a phase of backing anything trained in Lambourn, and one ridiculous autumn of following a man on a forum who called himself The Professor.
The Professor vanished that November with a good chunk of his members' subscriptions, and I didn't hear from him again.
My worst year, 2009, cost me a little over £4,000, and if you'd asked me at the time I'd have sworn I was breaking even.
I only found out the real figure when I printed off a year of bank statements and went through every transfer to the bookmakers with a highlighter pen.
It's a strange feeling to discover you've been fooling yourself, and I suspect a lot of punters would recognise it if they did the same sum.
By Trade, I'm A Data Analyst
I spent the better part of 15 years building predictive models for a large insurer in the City of London.
I'd come to the insurer from a market research firm, where I'd spent 4 years counting things for other people.
The insurer was the first place that let me predict things instead of just counting them, and I loved it from the first week.
My working days were spent taking enormous, filthy datasets and dragging the patterns buried inside them out into the light.
My speciality was motor claims, working out which drivers were likely to have an accident in the next 12 months before anything had happened.
We never knew which driver would crash, but across 2 million policies we could say how many would, and get it right to within a whisker.
The real craft lay in knowing how far to trust a small amount of evidence
A 19-year-old with a brand new licence has no claims history at all, so you lean on what thousands of drivers like him have done before.
A 55-year-old with 30 years behind the wheel has earned a record of his own, and that record deserves more weight than any average.
Get that balance wrong in either direction and you end up overcharging the careful drivers or undercharging the reckless ones.
Actuaries call this credibility, and it would turn out to be one of the most useful ideas I ever carried out of that office.
One of my proudest pieces of work found that drivers who renewed their policy late at night claimed more often than those who renewed in the morning.
Nobody had thought to look, as the time of renewal sat in a column everybody else ignored.
That's the habit I'd later bring to racing, since the useful clue is often hiding in the column nobody reads.
And it took me an embarrassingly long time to notice I'd never once pointed those same skills at the thing I loved most outside of my own family.
The Bookmakers Aren't Guessing
They stopped guessing an awfully long time ago.
They employ people with backgrounds just like mine, and they've done so for decades.
And my own skills were sitting idle at the end of every working day, making somebody else rich.
One Ordinary Sunday, I Decided To Change That
The moment it all fell into place arrived in the winter of 2023, and it wasn't at a racecourse or over a betting slip.
My uncle Ron had passed away that autumn, and I'd spent a grey Sunday helping to clear out the back room of his old flat.
He was 81, and he'd still been checking the evening results on his tablet until a few weeks before the end.
Tucked inside a box of old photographs I found a birthday card he'd written for me years earlier and never got round to posting.
Inside, in his shaky capitals, was a single line that read: "You've got the head for numbers, son, so make them work for you and not for the shop."
I sat down on the floor and read that one line over and over for the best part of an hour.
The card was dated the year I'd started at the insurer, and he'd kept it in a drawer for well over a decade.
I've no idea why he didn't post it, and there's nobody left to ask
Perhaps he worried it would read like an order, or perhaps he just forgot, which would have been very like him.
And the thought landed on me, plain as day
For 30 years my uncle had settled slips in his head, and I worked through far knottier sums than that on a computer every day of my working life.
I'd just never thought to turn that computer loose on the horses.
That Was The Night The Building Started
I went home, cleared the dining table, and opened a blank project on my laptop with nothing on it but a cursor blinking back at me.
For the next 18 months I worked on it most evenings between 8 and midnight, and for a good part of most Sundays as well.
My wife took to calling the laptop my other job, and she wasn't far wrong.
I wrote everything in Python, the language I'd used at work for years, and kept the races in a database on my own laptop until the files grew too big and had to move onto a rented server.
The first thing I ran headlong into is that horse racing data is an absolute nightmare to work with.
Insurance data, which I'd wrangled for years, is messy in ways you learn to see coming.
Racing data is messy in ways I hadn't braced for at all
Going descriptions shift from one track to the next.
"Good to firm" at Newmarket is a materially different surface to "good to firm" at Carlisle, yet most databases treat the two as identical twins.
Distances are rounded in some records and measured to the yard in others, and a rail movement can add 30 yards to a trip without the race title changing at all.
Rail movements alone took me 2 weeks to untangle, since each course reports them in its own way.
Trainers' licences can pass from father to son halfway through a season, and the records don't always make it clear who was really in charge of a horse.
One early mistake taught me more than any success did
My first back-tests used the official going recorded after each race, and the results looked fabulous.
The trouble was that nobody knows that figure until the race is over, so the model had been peeking at information no punter could have had on the morning.
Once I limited it to what was known by 7am on race day, the fabulous results shrank, and I was glad they did.
From then on, every test I ran obeyed one rule, which is that the model only sees what you or I could have seen when the email goes out.
Before I wrote a single line of predictive code, I spent 3 months just cleaning the data until it could be trusted.
Once The Data Was Right, The Building Could Start
I pulled in race history stretching back five full seasons.
On top of that I layered going data from every major track in the land, alongside trainer and jockey form trends.
Then came split timings from recent runs, stall position records, breeding influences and the betting history of each race.
By the spring of 2024 the database held around 450,000 individual runs, enough to fill most of an external hard drive.
It was thankless, unglamorous work, and it was the foundation that everything after it would come to rest upon.
A machine learning model is only ever as reliable as the data you feed into it.
Give it muddled records and it learns muddled patterns
Give it clean, well-ordered data and it starts turning up things that actually matter.
The earliest versions of the model were crude, and I've no shame in saying so.
It was plain logistic regression on the obvious variables, such as recent form and the weight each horse carried.
Logistic regression is the workhorse of the insurance world, a tidy formula that gives each factor a fixed weight and adds them up into a probability.
It's quick and easy to explain to a manager, but it assumes every factor counts the same amount in every race, and racing doesn't work like that.
The results were thoroughly underwhelming
Across 2 seasons of back-testing, it turned a £1,000 bank into £640 on paper.
The obvious variables were never going to be the advantage, because they were already priced into the market down to the penny.
The Advantage Had To Come From Where They Weren't Looking Closely Enough
So I started digging into the interactions between the variables.
Not single factors standing on their own, but combinations of conditions rare enough that the market had never learned to price them.
A horse dropping in class after a run on ground it plainly hated, trained by a yard in red-hot form over this exact trip in the last 28 days, and yet drifting in the betting because its profile doesn't catch the eye.
To catch combinations like that, I swapped the old formula for a method called gradient boosting, which builds hundreds of small decision trees one after another.
Each tree is a short string of questions, such as whether the trip is longer than last time and whether the ground has softened since the horse's last run.
Every new tree concentrates on the races the earlier trees got wrong, so the model gradually learns the awkward cases that a single formula would miss.
By the end there were 800 trees in the model, and between them they could spot patterns involving 4 or 5 conditions at once.
Training it on 5 seasons of racing took about 3 hours, so I'd usually set it running before bed and read the results over breakfast.
I knew what I was looking at the moment it began showing up
Here's how the app takes on a full day's card in practice, step by step, while you're still asleep.
Every morning, before the first race is anywhere near off, it pulls a complete data feed for every declared runner across the whole day.
The work starts at 5am on a rented server, and the full run through a big Saturday takes around 40 minutes.
It reads each horse's past form broken down by trip and by ground, rather than as one lazy lump figure.
It tracks trainer form across rolling 28-day and 60-day windows, so a yard coming into form is caught early and a yard going off the boil is spotted just as quickly.
It weighs jockey performance at this specific type of track, because a rider who thrives round a tight turning circuit isn't always the one you'd want on a long galloping straight.
It reads the split timings from recent runs to understand how a horse truly gets the trip, not merely where it happened to finish.
It reads class movement, whether a horse is stepping up in grade or slipping down a level to pinch a race
It maps the likely pace of the race, working out which horses want to lead and whether the shape of the contest will suit a hold-up horse or a bold front-runner.
Pace matters more than most punters think, as a horse that needs to settle in behind can be ruined by a slowly run race.
It watches for the small clues a seasoned judge lives for, such as a first-time headgear change or a yard suddenly booking a top jockey for an ordinary-looking handicapper.
It checks the weight each horse carries today against the weight it carried when it last ran well, since a few pounds can be the difference between a winning mark and a hopeless one.
It notes when a horse has moved yards, since some trainers improve new arrivals within a run or two while others take months to get to know them.
And it tracks live market movement overnight, following how each price has drifted or shortened from the early show right up to the moment my email lands with you.
That's Several Hundred Data Points Per Horse, Weighed For Every Runner At Every Meeting
A motivated human analyst could maybe work through 10 or 12 of those factors for one single horse in half an hour, with a strong coffee and a following wind.
My App Works Through Every Last One Of Them In A Matter Of Seconds
But raw speed was never the real advantage, and I want to be clear on that point.
The real advantage lies in how the model weighs each factor depending on the exact context of the race sitting in front of it.
Draw position matters enormously in a big-field sprint handicap round the tight bends of Chester.
Over a mile and a half at Goodwood it barely registers at all
Course shape plays its own tricks, with a stiff uphill finish like Sandown's suiting a horse that stays every yard of the trip.
Going preference can be the entire story for certain breeding lines, and almost meaningless for others.
The model learned every one of those distinctions from the historical data itself.
Not because I sat down and hard-coded them in as rigid rules.
But because it chewed through enough races, under enough different conditions, to work out for itself which factors matter and precisely when.
Hard-coded rules snap the instant they meet a situation nobody wrote them for
A trained model bends and adapts, because it's already seen enough variety to know where to lean its weight.
But the most important shift of all was a change in thinking, not a change in code.
The Real Breakthrough Came When I Stopped Comparing Horses With Each Other
Instead of asking which horse was going to win, I rebuilt the whole model around an entirely different question.
Under what conditions does each horse run its very best race, and are those conditions in place today?
That idea came straight from my years in motor insurance.
We never rated a driver against the national average, we rated him against his own record, his own car and his own daily journey.
A horse is no different, because each one carries its own fingerprint of trip, ground, track shape, pace and time between runs.
Some horses run their best race fresh after a 10-week break, while others need a couple of runs in quick succession before they come to the boil.
Some only show their true ability going left-handed, and some won't settle unless there's a strong pace for them to chase.
There's a crucial distinction hiding in there
The race card shows you how a horse has run, but it doesn't tell you which of those runs came under conditions it actually liked.
A horse with 3 dismal runs on fast summer ground can look hopeless on paper, and yet be a different animal when autumn rain softens the track.
So the app builds a personal profile for every horse in its database, and there are now more than 48,000 of them.
Each profile is a set of preferences, scored for trip, ground, course direction, track shape and the time since its last run.
When the app reads a race, it measures how closely today's conditions match those preferences, and it calls that measure the fit.
A horse with a high fit today, whose recent poor runs all came with a low fit, is the pattern it hunts for above all others.
This is also where that old insurance idea of credibility earned its keep
A 3-year-old with only 2 runs to its name doesn't have enough history to build a profile on its own, so the app borrows from horses with the same sire and a similar racing pattern until its own record fills in.
A 7-year-old chaser with 25 runs has said a lot about itself already, so its own record does nearly all the talking.
Before I added that weighting, the model kept falling for lightly raced horses on the strength of one eye-catching run, and it cost me on paper week after week.
Each morning it checks every declared runner against its own profile, rather than against the rest of the field.
When a horse's ideal conditions line up on the day, and its recent form was posted under conditions it disliked, the app flags it.
Those are the runners the betting public writes off, because they're reading the bare results and not the story behind them.
Not The Horse That Looks Most Likely To Win, But The Horse Whose Day Has Finally Come
Once the app has its shortlist, it runs one last check against the field.
It asks whether anything else in the race is just as well suited, because a horse on its perfect day can still bump into a rival on an even better one.
Only the runners that come through both checks make it into the morning email
On some mornings that leaves 2 bets, and on a packed Saturday it can leave 7 or 8.
The threshold is the minimum fit a horse needs before it's even considered, and I've raised it twice since 2024 as the profiles have grown richer.
Once I'd rebuilt the model around that single principle, the back-test results shifted dramatically.
The same 2 seasons that had shrunk my £1,000 to £640 on paper now grew it to a little over £3,900.
By the summer of 2024 it was flagging patterns I'd never once spotted in years of manual study.
Horses whose last few runs had hidden their true ability from almost everyone.
Hidden Ability, Waiting For The Right Afternoon
The first time I trusted it with real money, I fully expected to be disappointed.
I staked £10 a bet through October 2024 and logged every selection alongside what the back-test said I should expect.
The first winning bet came on the third morning, a 9/2 chance in a mile handicap at Kempton, and I remember staring at the screen as it crossed the line.
It was only £45 of profit, and the screenshot is still saved on my phone, next to a photo of Ron at the shop counter in 1986.
By the end of that month the 2 lines matched almost to the pound, which told me more than any lucky winner ever could.
In insurance we'd call that a model that holds up in the real world, and it's a far rarer thing than you'd imagine.
Most models that shine in testing fall over the moment real money and real prices get involved.
And I felt something I hadn't felt in all my years of betting.
Not Luck, But Calculated Certainty
From that point, things moved quickly.
Through the back end of 2024 I refined the model without an ounce of mercy.
The first improvement came from my insurance days, where old claims count for less than new ones because drivers change over time.
Horses change too, so I taught the app to give each run less weight as it gets older, with a 6-year-old's juvenile form now counting for almost nothing.
The next improvement was a rating for the strength of every field a horse had run in.
Finishing 4th behind 3 future winners can be worth more than winning a weak race, and the app now knows the difference.
Then came a check I borrowed directly from the claims department
Each month I grouped every horse the app had rated as a 1-in-4 chance and counted how many of them went on to win.
If the answer drifted away from 1 in 4, I knew the model had grown too bold or too timid, and I adjusted it before it cost me a penny.
In practice that means sorting every rated horse into bands, from 1 in 2 down to 1 in 20, and comparing each band's predicted winners with the actual number.
It's the same test we ran on motor policies every quarter, and it's the most useful habit I brought with me from the City.
I want to be specific about what that journey looked like, because it wasn't an overnight success and it matters that you understand that.
The first couple of months of live betting were modest, profitable but only just.
A few hundred pounds here, a slightly better run there
Enough to tell me the model was working, and not yet enough to get excited about.
By the closing weeks of 2024 I'd raised my stakes, and the monthly profit had climbed past £1,500.
Raising the stakes wasn't a leap of faith, it was arithmetic
I followed the same rule the app now uses for members, staking a fixed share of the bank, so the stakes only rose as fast as the profits allowed.
By the time 2025 was in full swing, £3,000 months had become the norm rather than the exception.
My first £5,000 month arrived in November 2025, and I took my wife out for dinner at the hotel where we'd held our wedding reception.
In the spring of 2025 I handed in my notice at the insurer, after 15 years and one very long conversation with my wife.
My manager offered me a 4-day week to stay, and I lay awake for 2 nights before I turned him down
The timing could hardly have looked worse, because barely 2 months after I'd cleared my desk, the app hit its roughest patch to date.
Across 3 weeks of that June the bank slid back by close to £700, and the old punter in me was desperate to start tearing things apart.
Instead I ran the monthly calibration check, saw the 1-in-4 horses were still winning about 1 time in 4, and left the model alone.
By the middle of July every penny was back, and the bank went on to fresh highs before the month was out.
On my last day at the insurer I'd bought cakes for the team, and I was very glad none of them could see my bank balance that June.
Those 3 weeks taught me that the hardest part of following a model isn't building it, it's trusting it through a rough patch.
Through the rest of 2025 I made 2 more changes that are worth telling you about
The first taught the app how racecourses compare with one another, so a horse that loves the tight turns of Chester is recognised as likely to enjoy Catterick, even on its first visit.
The second came from a rule that's particular to British racing, where trainers must declare when a horse has had a wind operation.
The app learned which yards' horses tend to improve for that first run after surgery, and which ones rarely do.
And by the time this year came around, my AI betting app was a completely different animal to the clumsy thing I'd first cobbled together.
The wins came regularly, and then they came in real volume
Let me walk you through this year in particular, because 2026 is where it all truly came good.
January£6,340.75
February£4,920.20
March£6,910.55
April£5,480.30
May£3,760.80
June£6,590.40
July£5,924.30
August£5,215.65
September£5,875.35
October£3,562.15
January opened the year at £6,340.75.
February was a shade quieter at £4,920.20, and it still never once looked like finishing in the red.
March turned into the best month of the year at £6,910.55.
The Cheltenham Festival helped, with big handicap fields full of horses whose best form had been posted on very different ground.
April held its nerve at £5,480.30
May was a softer stretch at £3,760.80, the sort of month that would have had an ordinary tipster reaching for the panic button.
June bounced straight back with £6,590.40.
Royal Ascot's packed handicaps did much the same job that Cheltenham had done in March.
July kept the run going with another £5,924.30.
August is the busiest month of the Flat season, with big fields at Goodwood and York filling the screens.
Big fields put most punters off, but they hand the app more runners to profile, and August closed at £5,215.65.
September brought the St Leger meeting at Doncaster and the Ayr Gold Cup, and the app kept going with £5,875.35.
That took the profit for the first 9 months of 2026 to £51,018.30
October is a changeover month, with the last of the turf Flat overlapping the first big jumps cards of the season.
Many of the jumpers coming back are having their first run since the spring, and their profiles from last season can be months out of date.
So the app applies a rule I built in last autumn, which trims its confidence in any horse returning from a long summer break until it has run once.
The Cesarewitch at Newmarket and Champions Day at Ascot both fell in October, and the app found a few bets on each, but the weeks either side were thin on selections by design.
That means fewer bets during the changeover, and fewer bets means a smaller monthly figure, even when the selections are landing well.
October 2026 closed at £3,562.15, the lightest month of the year and still comfortably in profit
I'm happy with that trade, as the horses it held back on in October have now had their first run and their profiles are up to date again.
Members who joined over the summer saw it too, with fewer emails in October and busier ones again by the start of November.
And in the first 7 days of November it added another £1,487.60, as the jumps cards filled up and those returning horses came back into play.
The first week of November is when the big jumps yards start running their better horses, and the app has been busy ever since.
There Wasn't A Single Losing Month In The Last Year
Not one month, from last November right the way through to this October, where the model finished in the red.
That kind of reliability, month after month, is what separates a systematic advantage from a lucky heater that burns out by autumn.
For the record, the last 2 months of 2025 brought in £5,760.30 for November and £3,180.50 for December.
The full 12-month total came to £63,521.25, drawn out into my account, tax-free, across the last year.
I'd Finally Solved The Puzzle My Uncle Never Could
I'd beaten the bookmakers at their own game, using the same kind of arithmetic that kept Ron's shop open for 30 years.
The evening it truly sank in, I didn't celebrate the way you might picture.
I took Ron's unposted birthday card out of the drawer and propped it up beside my laptop.
His one line of advice sat on one side, and 12 months of profit sat on the screen on the other.
My wife found me sitting there and asked what on earth I was grinning at.
I told her I'd finally done what Ron asked of me, and made the numbers work for me instead of for the shop.
The card still sits on the shelf above my desk, and it's the first thing I see each morning when the app finishes its run.
There's no Sunday evening dread in this house any more
I've turned my oldest passion into something that hands me real, lasting financial freedom.
And now I'd like you to feel the same way.
Every One Of Those 12 Months Finished In Profit
Remember those 8 ordinary people I let follow my bets at the start of the year?
The results they've pulled in are very nearly the mirror image of my own.
Between them they've followed well over 1,500 of my selections this year, at stakes ranging from £2 to £50 a bet.
The most common thing they tell me isn't about the money at all, it's how much less of their week racing now takes up.
This Model Isn't Only Working For Me
It works for ordinary people willing to follow the bets as they're sent, without picking every one to pieces.
Here are four more of their accounts.
They come from all corners of the country, and from very different walks of life
"You asked what the first £1,000 went on, so here's my answer: a rebuilt garden wall and a long weekend in Tenby. It took me 9 weeks to get there, and the account has since gone past £1,700 without me raising my stakes."
Terry Pettifer, Cirencester
"I couldn't have told you the difference between a handicap and a maiden when I started, and I'm still a bit hazy now. None of that has stopped me making £1,094.60 since January, which tells you where the hard work is being done."
Sandra Ackroyd, Harrogate
"My first fortnight was as flat as a pancake, and I came very close to asking for my £20 back. I'm glad I held on, because the next 5 weeks took me to £1,438.70 and the guarantee is the last thing on my mind."
Kevin Mottram, Mansfield
"Last year I worked out I'd lost £1,860 across 12 months of picking my own horses. This year the app has turned that into £1,612.25 of profit, and I haven't picked a horse of my own since January."
Lynne Ashdown, Chichester
STEP 1
Before 8am, a single email lands in your inbox
STEP 2
Open it, and the selections, odds and bookmaker are all inside
STEP 3
Place the bets with your online bookmaker
STEP 4
Done, and you get on with your day
5 to 10 minutes, and that's the whole process on your side
Every morning, before 8am, a single email drops into your inbox.
That email holds the day's selections and the exact odds to aim for, along with which bookmaker to place each bet with.
Each email is short, usually fewer than 100 words, with the course, the race time, the horse's name, the price to take and the bookmaker.
There's no waffle and no long write-up about why a horse might win, since you don't need one to place the bet.
Each bet is marked as win or each-way, with a suggested stake shown as a share of your bank, so it works whether you bet £2 or £200.
The share is 1.5% of the bank for most bets, and the app never goes above 2% on any one horse.
It takes you somewhere between 5 and 10 minutes from start to finish
Most members place their bets on a phone app in under 5 minutes, often before they've left the house.
You don't need to know the first thing about machine learning or probability theory.
You just need a funded betting account and an email address.
You don't need to watch the races either, though most members tell me they can't resist.
By the time my email reaches you, the app has already worked through the full day's card and filtered it down to the bets that clear its threshold.
The Hard Work Is Finished While You're Still Asleep
Now let me bring you right up to date on where things stand as we move through November 2026.
The 8 trial members I mentioned at the start of this page are all still following the bets, and all still sitting nicely in profit.
Every One Of Them Passed The £1,000 Mark Months Ago
Which brings me neatly to why I'm writing to you today.
I've decided to open 50 memberships inside Race Ranker App to a hand-picked group of new members.
I need to keep the numbers tightly capped while I continue to build this thing.
I need enough time each morning to get the bets out and deal with any questions that land in my inbox.
And I want to read every one of the first 50 success stories myself, which I can't do if thousands of them arrive at once.
I could have kept this to myself, and for nearly 2 years I did
What made me open it up was those 8 trial members, as watching them do as well as I had told me the method would travel.
A method that only works for the person who built it is a hobby, but one that works for 8 strangers from all over the country is something else.
I answer every member email personally, and 50 new members' worth of questions is about as much as one man can handle in his first week.
If you're reading this right now, memberships are still open
I expect all 50 to be gone inside the next 24 hours.
You've been personally invited by one of my trusted partners to see this today.
They reckon you're just the sort of person who'd get real use out of this.
Someone who could quite easily be £1,000 in profit inside their first fortnight.
Now, you're almost certainly wondering about one thing above all else.
How much is this going to cost me?
A system that produced over £63,000 in profit across the last year carries serious, real-world value.
When I first talked about opening this up to more people, those closest to me suggested a three-figure entry fee without a second's hesitation.
Perhaps £500 up front, with a monthly profit-share arrangement layered on top of it.
And that's the pricing model I'll move to once this opens up to the wider public in the new year.
But right now, I've got one specific priority before any of that happens.
I Want To Build A Wall Of Rock-Solid Proof
I want 50 solid case studies from ordinary, everyday people across the country who've used Race Ranker App to turn their finances around.
Once those 50 stories are on record, the doors close and the price climbs to its full value.
But you're here today, while those 50 beta spots are still on the table.
I'm happy to let you follow my bets for a one-time fee of £20.
This modest fee covers little more than the server hosting and the automated email system I use to get the bets into your inbox each morning.
It costs less than a losing each-way double, and you only ever pay it once.
In return for such a heavy discount, all I ask is that once you've made your first £1,000, you send me a short email telling me what you spent it on.
30-Day Money-Back Guarantee
I'm backing every penny of your joining fee with a full 30-day money-back guarantee.
You can follow my bets for the next 30 days without putting a single penny of that £20 at any real risk.
Start with the smallest stakes you like while you build your confidence in the system.
Or go straight in and aim for that £250 daily target from day one.
If at any point in those 30 days you're not completely happy with the results, just send me a single line by email.
I'll Refund Your £20 Immediately
No questions asked, and no awkward attempt to talk you round.
That's how sure I am that your first month will speak for itself.
Your winnings stay yours either way, since they sit in your own bookmaker account and don't pass through me.
Joining takes about 2 minutes from start to finish.
Click any of the buttons on this page, enter your details on the secure ClickBank order form, and you'll be taken straight to the members' area.
Your first email arrives the next morning before 8am, along with a short welcome note explaining how the staking works.
The Odd Questions That Land In My Inbox
Why is it called Race Ranker App?
Because my uncle Ron used to say a good punter ranks the runners before he backs one, and the app does that for every race on the card.
It scores how well each runner suits the day's conditions, then ranks the whole field from top to bottom, which is a long way from glancing at the last 3 results.
Did you raise your stakes as the bank grew, or bet the same amount all year?
The app stakes a fixed slice of the current bank on each bet, so stakes rise after a good month and ease back after a quieter one.
That's how a £1,000 bank grew to over £63,500 without me ever risking more than a small portion of it on any one race.
Does the app ever send a correction after the morning email has gone out?
No, the email is final, and the app never revises its view once the selections have left the server.
I built it that way on purpose, because a tip that changes at 11 o'clock is a tip nobody can follow reliably.
Are Irish meetings included, or only British ones?
Both are included, as the database covers every race run in Britain and Ireland, and an evening card at Dundalk gets the same treatment as a Saturday at Sandown.
Most mornings the bets lean towards British tracks, as there are more of them, but Irish runners turn up most weeks.
What does it make of a horse that has never run before?
It leaves debutants out of the email altogether, since a horse with no runs has no profile, and borrowed evidence alone isn't enough to bet on.
Once a newcomer has run twice, the app starts building its profile, leaning on similar horses until its own record can take over.
Do I have to bet win only, or are each-way bets included?
The email marks each selection as win or each-way, depending on the price and the size of the field.
Bigger-priced runners in big handicaps tend to go each-way, while shorter prices in small fields are nearly always win only.
Can I keep placing a few bets of my own alongside yours?
You can, though members who keep their own bets in a separate account find it much easier to see what the app is doing for them.
Mixing the two is how a profitable month gets buried under a handful of hopeful punts.
How much of a betting bank do I need to start with?
Most members start with somewhere between £100 and £500, and the stakes in the email scale to whatever you choose.
Starting small costs you nothing over time, as share-of-bank staking lets a bank grow at the same rate whatever its size.
Does the app treat the big festivals any differently from an ordinary midweek card?
No, a handicap at the Cheltenham Festival goes through identical checks to a Wednesday afternoon at Fontwell.
The only difference is that bigger fields tend to produce more runners whose profiles fit, so festival weeks often carry a few more bets.
Do I need a smartphone to follow the bets?
No, the email reads just as well on a desktop computer or a tablet, and a few members print it off and take it to their local betting shop instead.
What happens on the handful of days with no racing at all?
There are only a few of these each year, mostly around Christmas, and on those mornings you'll get a short email telling you to enjoy the day off.
Why do you still run the 1-in-4 check every month when the app is already making money?
Because a model that's making money can still be drifting, and the drift shows up in that check weeks before it shows up in the bank.
It takes me about 20 minutes on the first Sunday of each month, and it's the one job I'll always do myself.
What do you do with the emails people send about their first £1,000?
I read every one of them, and with permission, the best of them will become the case studies that let me set the full price later on.
So far one member has paid for a new boiler, and another took his grandchildren to Blackpool for a week.
Why charge £20 at all, rather than letting people in for nothing?
A free service attracts people who don't place a bet, and I need 50 members who'll follow the email and report back.
£20 is enough to show you're serious, and the guarantee means it isn't really at risk.
What would Ron have made of it all?
I like to think he'd have laughed, because he spent 30 years settling other people's bets and hardly ever placed one himself.
He'd have asked to see the figures first, and then he'd have asked what took me so long.
His line on that birthday card is still the best advice about betting I've ever been given.
The only way you lose out today is by clicking away and letting someone else take your spot.
This Is About As Low-Risk As Any Opportunity Gets
Click the button below to lock in your place before all 50 memberships are claimed.