Imagine it's just past seven in the morning.
Your inbox pings.
A single email from Bet Maximiser App.
Inside it: three, maybe four selections, each one flagged by an AI model that processed over 4.6 million data points overnight while you were asleep.
You log into your betting account.
The selections are on in under 8 minutes.
You close the laptop and get on with your day.
By mid-afternoon, your balance is up.
That's the entire system.
No form guides. No newspapers. No sweating over each-way margins.
Let me tell you how this came to be, and why I firmly believe it's the most unfair advantage available to any punter in Britain right now.
My Name Is Simon Westman. I've Been On Both Sides Of The Counter.I'm 51 years old, I live in Worcestershire with my wife and four kids, and for the past couple of years I've been running the most reliable AI-driven horse racing selection system I've ever encountered.
I'd like to tell you the full story, because it matters.
It matters not just because the results speak for themselves, but because of where I started.
I know what losing feels like. I know what it's like to have a system you're sure of, only to watch it fall apart over six weeks.
I know what it's like to be up £1,200 in March and down £900 by April.
I spent nearly a decade going round in circles.
And the reason I'm writing this today is because I eventually found a way to break out of that cycle entirely.
Not through harder study. Not through a better tipster. Not through following the money in the market.
But by building something that operates completely outside the way the average punter thinks.
Before I Was A Punter, I Was On The Other Side Of The Counter.Not many people who use this service know this about me, but I spent the better part of eight years working in the back office of a regional bookmakers chain.
The role was called a settler.
When a customer came in with a complex accumulator, a fold bet, or a Lucky 63, it landed on my desk to work out and pay out.
I saw everything. Every payout. Every losing return. Every clever accumulator that nearly came off.
And crucially, I sat next to the traders.
The people who decide where the prices go.
I watched how they moved the market. I watched how they reacted to money coming in from certain directions. I watched how they used early morning data to set their tissue prices before racing even started.
I absorbed all of that information without fully realising what I was learning.
When I eventually left that role and got properly into betting as a punter, I was carrying eight years of inside knowledge about how bookmakers operate.
The trouble was, knowing how they worked didn't automatically mean I could beat them.
I still had to find an edge.
I spent years searching. Followed tipsters who made strong starts and petered out. Built my own spreadsheet models that worked beautifully on paper and fell apart in the real world.
Attended seminars, read books, had long conversations on forums with people who swore they had it cracked.
I Did Make Money, But Not Consistently Enough To Change Anything.I'd have a solid three-month spell, then a rough patch that gave back half of what I'd made.
It was progress, just not the kind you can build a life around.
The breakthrough came from an unexpected direction.
I'd been at a trade event in Birmingham, one of those industry gatherings where developers and analysts talk about machine learning applications.
I wasn't there for betting.
I was there because a friend had dragged me along to look at software solutions for a small property business he was running.
But I wandered into the wrong seminar room and ended up sitting in on a 40-minute talk about predictive modelling in competitive sports.
The speaker was a data scientist called Ryan. Mid-to-late twenties, presenting to a room of maybe 15 people.
He Was Talking About AI Finding Patterns In Datasets That No Human Could Reasonably Identify.Not because humans aren't smart enough.
But because humans can't hold enough variables in their head at once to spot the relationship between them.
He gave an example from football analytics.
A model had found that a certain combination of six factors, none of which looked individually significant, predicted the outcome of corner kicks with a success rate that was 31% above random.
No human analyst had spotted that relationship. Not because it was hidden. But because you'd have to cross-reference six different data streams simultaneously, across thousands of events, to see it emerge.
That example sat with me for the entire drive home.
The traders I'd worked alongside didn't rely on gut instinct.
They used morning data feeds, liability models, and weighted probability tools to set their markets.
The bigger firms had been using algorithmic pricing systems for years.
They weren't guessing. They were running calculations I couldn't see.
And as a punter, I'd been essentially bringing a notepad to a gunfight.
That evening, I dug out Ryan's details from the event programme and sent him an email.
I explained my background. The eight years in the back office. The data I'd accumulated over years of serious betting. The problem I was trying to solve.
Instead of a polite brush-off, he called me back within 48 hours.
He Already Had A Machine Learning Framework Built For Sports Analytics.It had been designed initially for tennis, of all things, but the underlying architecture was sport-agnostic.
Feed it structured data, define the target outcome, set the learning parameters, and it would find the patterns.
More importantly, it would tell you which patterns held over time and which ones were noise.
I had 9 years of detailed race-by-race records sitting on my hard drive.
Every horse. Every course. Every going condition. Every class of race. Every trainer, jockey, draw bias, and price movement I'd tracked.
Most of it was in spreadsheets, meticulously categorised. Some of it was in notebooks I'd scanned and digitised.
It was the most thorough personal racing database I'd ever heard of anyone keeping outside of a professional operation.
Ryan agreed to take it on. We settled on terms over a call that lasted three hours.
It took about 6 weeks from that first call to having something worth testing.
Those 6 Weeks Were The Most Excited I'd Been About Betting In Years.The build phase took six weeks from that first call.
Six weeks during which I fed Ryan every spreadsheet, cleaned up the data formatting at his instruction, and learned more about feature engineering than I'd ever expected to.
Feature engineering is the process of taking raw data and converting it into the inputs a machine learning model can actually learn from.
It sounds technical, and it is. But the principle is simple.
Raw data tells you what happened. Feature engineering helps the model understand why it happened.
Ryan spent the first fortnight just working on the features. Not building the model. Just deciding how to represent the data so the model could find signal in it rather than noise.
That process alone was an education.
How The Bet Maximiser App AI Actually Works.Horse racing generates an enormous amount of data.
Every race. Every trainer's season statistics. Every jockey's performance across different track types and going conditions. Every horse's full history, broken down not just by wins and losses but by how it ran, where it was in the race, how it finished, how its rivals subsequently performed.
Course biases. Draw effects. Class changes. Weight shifts. Market movements from tissue price to off. Weather data. Going reports. Going stick readings where available.
A serious human analyst can look at maybe 8 to 12 variables simultaneously before cognitive load kicks in and performance starts to drop.
Simultaneously. Without fatigue. Without the distortion of having watched the race last week and formed an opinion about a particular horse.
Without the subtle pressure that comes from having told your mates to back something and then seeing it drift in the market.
It runs the same calculation every single time. No emotion. No momentum. No favouritism.
The key to where it really earns its keep is in what Ryan calls interaction effects.
This is where two or three or four variables combine to produce an outcome that none of them would predict individually.
A Horse That Looks Unremarkable On Its Own Record.But pair it with a specific trainer who has a 74% strike rate when moving horses up a class after a layoff, at a course with a pronounced draw bias that favours its racing style, on ground that's been independently tested at good-to-firm when the official going says good...
Suddenly that 20/1 shot looks very different.
A human tipster would have to hold all of that in their head simultaneously. Process it. Weight it. Arrive at a conclusion. And do that for every runner in every race they're covering.
It's impossible to do consistently. The model doesn't even break a sweat.
It runs every combination of factors in milliseconds, flags the selections where the evidence is strong enough, and ignores the rest.
Most punters look at form and see a sequence of results. Won, fell, fourth, second, won. They read that and form an opinion.
The model reads the same data completely differently. It's not interested in the headline results. It's interested in what the results reveal about the horse's underlying performance.
Was it gaining or losing ground in the final furlong? Was it making up lengths or giving them away? Did it travel well through the race or did it spend energy fighting the rider before the turn?
These details exist in the race replay data that gets fed into the system. And they paint a very different picture from the raw finishing position.
The Second Layer Is Conditions Matching.Every horse has a profile of conditions it performs best in. Going. Course layout. Distance. Class of race. Draw.
Most of these are well known to experienced punters. But the model goes further. It looks at the interaction between conditions.
A horse that goes well on good ground might perform entirely differently on good-to-firm ground when a particular jockey is on board.
A course that favours a high draw in small fields might favour a low draw in fields of 12 or more.
These micro-adjustments matter. And most humans, however experienced, aren't applying all of them simultaneously to every runner in every race.
The Third Layer Is Market Intelligence.This is where my years in the back office become directly useful.
The model doesn't just look at starting prices. It tracks price movement from the morning tissue through to the off.
It's looking for specific patterns. A horse that shortens significantly in the early market and then drifts out before racing. A horse that's been ignored by the market all week and then comes in sharply in the last hour.
These aren't just random fluctuations. They reflect information flow. Money moving in specific ways at specific times tells you something about what informed parties know.
The model has been trained to recognise these patterns. And when the market movement aligns with the form profile and the conditions match, the confidence score goes up significantly.
That's when you get the high-value selections. The ones priced at 6/1 or 8/1 or longer that the model identifies as having a real probability of winning at 2/1 or 3/1.
This is the part most systems ignore completely.
As much as the model is looking for reasons to back a horse, it's equally looking for reasons not to.
Certain combinations of factors produce a consistent pattern of underperformance. The model has learned which those are.
And it will exclude a horse that looks compelling to a human eye if the underlying data contains one of those negative patterns.
Trainers who consistently underperform with certain jockey combinations. Horses that show a specific type of flat finish in their recent runs. Market patterns that historically precede a poor run.
All of it gets factored in. That's why the losers are smaller in number than you'd expect. It's not just about finding winners. It's about reducing the rate of avoidable losers.
And Here's The Part That Really Matters.The bookmakers are doing exactly the same thing on their side.
The major firms have had algorithmic pricing tools since the early 2000s. The biggest names in the industry now run fully automated pricing models for their most competitive markets.
They're not setting prices based on what their traders think. They're running calculations across millions of data points and letting the model determine where the price should be.
They're using AI to stay ahead. And for years, punters have been fighting that with notepads and newspaper columns.
It's not trying to outguess the bookmaker's trader. It's competing with the bookmaker's algorithm.
And because I built mine on my own data, tuned to the specific patterns in British horse racing over nearly a decade, it finds gaps that the bookmaker's generic model leaves open.
That's the edge. Not luck. Not inside information. Not a magical formula.
It's a machine finding mathematical discrepancies between the price on offer and the true probability. And betting them when the margin is wide enough to be worth backing.
How It Improved, Month By Month.The first version of the model was finding winners, but it was also finding too many losers. The strike rate was sitting around 48% to 51%.
Ryan explained that this was expected. The model needed to process enough outcomes to learn which of its early assumptions were wrong.
Every losing bet fed information back into the system. Every winner reinforced the patterns that were actually predictive.
Week One: Around £350 In Profit.Not life-changing, but a positive start.
Week two, Ryan made some adjustments to how the model weighted going conditions. The strike rate crept up to 53%. The profit that week was £570.
The week after that, 7 winners from 13. A balance of £724.50 across the week.
Something was working.
I Started Paying Closer Attention To What The Model Was Selecting.Not because I was going to second-guess it. But because I wanted to understand the logic.
And what I noticed was striking.
The model wasn't selecting horses that looked like obvious winners. It was finding horses where the price the bookmaker was offering was significantly out of step with the horse's actual probability of winning.
That's the entire game, distilled. Find the gap between the price and the probability. Bet when the gap is big enough.
That's what professional traders do. That's what the sharps do. And my model was doing it automatically, every single morning, before I'd finished my first coffee.
Month 1: 32 Successful Bets From 49Profit sitting at just over £3,000 across my accounts. I opened two new accounts and spread the risk.
Month two: 35 winners from 53. Over £7,200 banked.
I started withdrawing slowly, keeping the account balances at levels that wouldn't attract automated scrutiny.
This Is Something I'd Learned From My Years In The Back Office.The major bookmakers run flagging algorithms on their own customers. If your account shows a consistently positive return over a 90-day window, it gets reviewed.
I knew how to stay under the radar. Multiple accounts. Regular small withdrawals. Varying stake levels so no pattern emerged.
And when I wanted to get bets on at higher stakes, I'd walk into town and use the shops. A different one each time. Cash in hand. No account trail.
The model kept improving. What had started as a 51% strike rate climbed to 58%, then 61%, then plateaued at around 63% by the end of month three.
Ryan told me at that point that the plateau was actually a good sign.
Models that keep climbing indefinitely are usually overfitting. They're learning the noise as well as the signal. A model that finds its level and holds it is a model you can trust.
The Improvements Ryan Made Along The Way.The first six months were entirely about learning. Learning where the model was making errors. Learning which feature weights were out of line. Learning where my own data had gaps that needed filling.
Ryan would send me weekly summaries of where the model's confidence scores had been highest, cross-referenced against actual results.
Where it was most confident and most wrong, that's where we focused.
We rebuilt the going condition weighting three times in the first year.
We added a draw bias correction module after month four when I noticed the model was underperforming on straight-mile courses at specific tracks.
We brought in jockey booking data as an additional feature after month seven, because the pattern of jockey switches in the 48 hours before a race turned out to be significantly predictive.
Every adjustment made it sharper.
And now, it's the most reliable thing I've ever worked with. Not perfect. No model is perfect. But profitable, consistently, month after month, in a way that no tipster I followed in my first decade of betting ever came close to matching.
I Called It 'Bet Maximiser App'.Ryan continues to update the model with every new race result. The learning doesn't stop.
Each month it gets a fraction sharper. Each quarter the interaction effects become a little more refined.
That's the advantage of a machine learning system over a fixed methodology. It doesn't stagnate. It evolves.
It Only Responds To What The Data Confirms.It doesn't back a horse because it won last time out and looks like it should win again.
It doesn't avoid a horse because it drifted in the market or because a columnist gave it a poor write-up in the morning paper.
It looks at the raw numbers and asks one question: is the price available on this horse meaningfully higher than its actual probability of winning?
If the answer is yes by enough of a margin, it flags the bet. If the answer is no, it ignores the horse entirely regardless of how obvious a winner it might look to a human eye.
That discipline is impossible to maintain manually over months and years.
The AI never gets overconfident after a winning streak. It never chases losses after a bad week. It just runs the numbers and reports the output.
That's Precisely Why Bet Maximiser App Has Been In Profit Every Single Month Since It Launched.Over the summer of 2025, things really went into overdrive. Winning months of £6,430.55, £7,749.20 and even £8,150.75.
I had more free cash than I'd ever had before. My friends took notice.
My mate Dave was first in. He'd watched me turning a consistent profit for months and finally asked if he could copy my bets.
I said yes on the condition he kept a proper record and gave me his real assessment after 30 days.
His wife sent me a bottle of whisky.
By the end of that summer, Dave had made consistent profit for four consecutive months.
He wasn't following every selection. He was cautious, the way most people are when they're watching someone else's system work before going all in.
But even with his conservative approach, he was banking between £800 and £1,400 most months.
His wife told me later he'd started talking about it at family dinners. His brother-in-law asked Dave to send his bets too.
I let a couple more people in on the same terms. Follow the selections, keep your own record, give me your feedback at the end of 30 days.
All 6 of them finished in profit. The lowest was £1,200. The highest was over £3,000. Nobody lost.
That's when I began to think seriously about opening this up more formally.
Not as a large-scale tipster service. I didn't want that.
The moment you scale to hundreds of members, the market impact becomes a real problem. Your selections shift the odds before half your members have even logged in.
I've watched it happen to other services. Strong early results, expanding membership, and then the edge gradually erodes because too many people are following the same output.
30 Members Was Always The Number.Manageable. Spread across enough different accounts and bookmakers that the market doesn't feel it.
Small enough that the odds stay intact when the email goes out each morning. But large enough that I know I'm actually helping people.
I spent a couple of weeks putting this page together. Not to sell something I wasn't sure about. But because I was completely certain about what I had, and I wanted to share it with the right people.
People who've been around betting long enough to know the difference between a reliable system and a lucky streak.
People who are ready to act on what the data shows rather than second-guess it.
What Early Members Said After Their First 30 Days...I keep records of everything. Every bet. Every result. Every month's performance.
Not because someone told me to. But because when I was back in that bookmaker's office, the traders who were any good kept meticulous records. You can't improve what you can't measure.
Here's the full 18-month record for Bet Maximiser App. Every figure is from my actual accounts.
| Month | Bets Placed | Winners | Strike Rate | Monthly Profit |
|---|---|---|---|---|
| January 2025 | 46 | 27 | 58.7% | £3,140.25 |
| February 2025 | 48 | 28 | 58.3% | £3,470.80 |
| March 2025 | 52 | 31 | 59.6% | £4,280.50 |
| April 2025 | 54 | 33 | 61.1% | £5,060.30 |
| May 2025 | 58 | 35 | 60.3% | £5,490.75 |
| June 2025 | 56 | 35 | 62.5% | £6,040.55 |
| July 2025 | 60 | 38 | 63.3% | £6,820.20 |
| August 2025 | 62 | 39 | 62.9% | £7,150.75 |
| September 2025 | 55 | 34 | 61.8% | £5,970.40 |
| October 2025 | 58 | 36 | 62.1% | £6,280.90 |
| November 2025 | 54 | 32 | 59.3% | £4,620.15 |
| December 2025 | 52 | 31 | 59.6% | £4,050.60 |
| January 2026 | 48 | 29 | 60.4% | £5,310.40 |
| February 2026 | 50 | 31 | 62.0% | £5,680.55 |
| March 2026 | 56 | 35 | 62.5% | £6,700.55 |
| April 2026 | 58 | 37 | 63.8% | £7,240.80 |
| May 2026 | 60 | 38 | 63.3% | £6,890.30 |
| June 2026 | 54 | 34 | 63.0% | £5,420.60 |
| Total | 961 | 592 | 61.6% | £101,419.35 |
18 consecutive months in profit.
A strike rate of 61.6% across 961 bets.
Over £101,000 withdrawn in that period.
Early Members Have Averaged £595 Per Week So Far.They are not professional gamblers. They didn't come from a betting background.
One member is a retired postman from Staffordshire who had never placed a bet online before his son-in-law walked him through it.
Another is a secondary school teacher who places her bets on the school bus on the way in.
If the model's output is clear enough for them to act on, it's clear enough for anyone.
This Isn't About Studying Form Or Reading Newspapers.It's about following a structured output from a machine that does not guess.
The AI analyses every race each morning and produces a selection list based purely on where the mathematical value sits.
There's no opinion involved. No preference for famous trainers or fashionable yards. No tendency to back the horse that won well last time and looks certain to win again.
Those are the kinds of human biases that cost punters money year after year. My model has none of them.
It processes the data, identifies the discrepancy between price and probability, and flags the bet. If there's no discrepancy worth betting, it flags nothing.
Some Mornings There Are Four Selections. Some Mornings There's One.Occasionally, when conditions don't produce clear value, the email will tell you to sit it out.
That discipline is something most human tipsters simply cannot maintain. The temptation to send something out, to give members their money's worth for that day, is enormous.
I've seen it ruin plenty of good tipsters. They have a quiet week and start forcing selections to show activity.
My model has no ego. It doesn't feel the need to justify its existence by producing bets. It produces bets when the evidence warrants them.
And that's exactly why the strike rate has held above 60% for 18 consecutive months.
More Members, More Results...The reason entry is controlled is practical.
When the same selections go out to too many accounts, the market reacts. Early movers can shift the price. If my members all pile into the same horse within 20 minutes of each other, the odds shorten before everyone gets on.
The edge gets eroded. 30 memberships keeps the impact on the market manageable.
As I'm writing this particular section, 21 of the 30 places have already been taken. There are 9 left. Provided you're still on this page, one of them is yours.
How Exactly Does This All Work?The morning of your first day as a member, you'll receive an email with the day's selections.
Each one will include the horse, the race, the course, the time, and the price you should be looking to get.
Placing the bets takes most members under 10 minutes. You can do it online with any major bookmaker. Or you can take it to the shop if you prefer.
Whatever suits your routine.
No watching the races unless you want to. No checking odds movements. No late-night form study.
The system does all of that between one selection email and the next. You receive the output and act on it.
The process resets every 24 hours.
I've had members tell me they didn't even realise they had winning bets running until they checked their account after dinner. That's how hands-off this is.
The morning email does the heavy lifting. You just place the bets.
I've got retirees in my group, students and busy professionals. They all find the time to follow along and they're extremely happy so far.
One member is a hospital consultant who places his bets between ward rounds. Another is a 71-year-old from Shropshire who'd never bet online before joining. If either of them can do it, so can you.
There's No Learning Curve Here.You don't need to understand how the model works any more than you need to understand how your car engine works to drive it.
The selections arrive in your inbox, you back them, and the balance grows.
It takes most members under 10 minutes each morning from opening the email to having their bets placed. After that, your day is completely your own.
What That Means For You As A Member Going Forward.The service you join today is not the same service it will be in six months. It'll be better.
The selections you receive in early 2027 will come from a model that has processed everything that happens between now and then. Every winner. Every loser. Every market movement. Every going report.
It's all feeding back in, and the model is adjusting. That's the compounding advantage of a system that learns. The longer it runs, the more refined the output becomes.
You're not just buying into today's performance. You're buying into a system that gets better over time.
Not £20 a month. Just £20. A single one-off payment. No subscription, no renewal, no hidden tiers.
I want to be direct about why it's priced this way.
My accounts are performing well, but like most people who've run a successful system for any length of time, I'm aware that spreading the bets across more accounts keeps things operating smoothly.
30 members placing these bets across 30 separate accounts, none of which are linked to mine, distributes the activity in a way that keeps the edge intact.
It's a practical arrangement that works for both of us.
I run multiple betting accounts to keep my winnings away from their automated flag systems. On weekends, I love to bet in my local town centre, choosing a new shop every time.
So far, so good. My AI has generated up to £8,150.75 in a single month.
30 new members placing these bets across 30 different accounts keeps everything spread out and under the radar. It benefits both of us.
Your membership comes with a full 30-day money-back guarantee that allows you to follow every selection.
Watch the results come in and decide for yourself whether Bet Maximiser App delivers the standard you expect.
You have the entire 30-day period to use the service exactly as intended, to see the bets in real time.
To judge the accuracy, to track the consistency, and to feel the confidence that comes from a clear and structured approach.
If at any moment within those 30 days you feel that the service isn't for you, you can request a refund and you'll receive 100% of your joining fee back without delay.
No questions asked. No pressure. No delay.
Nobody has taken me up on that refund yet. But it's there in black and white and I'll honour it every time.
If you leave this page now, I can't promise it will still be here when you return.
The remaining places will go when they go, and once they're gone the service will be closed to new members until further notice.
If you're ready to take this step, click the button below and complete the checkout for just £20.
You'll have access to today's selections right away.
P.S. My bets are live right now for today's members.
In a few minutes they'll have their bets on and their day will carry on as normal.
If you join now, you'll be ahead of the next full day's selections.
The 30-day guarantee means you're carrying no financial risk at all in that first month.
And the £20 joining fee can be recovered from a single winning bet on day one.
Most members do exactly that. A few cover it twice over before lunchtime.
But even if it takes you a couple of days, the guarantee means you're carrying absolutely no financial risk at any point in that first month.
Take the full 30 days. Follow every selection. Track every result.
If you're not satisfied, you get your £20 back and we part ways without any awkwardness.
It really is that easy.
Yours sincerely,
Simon Westman