Last updated: 5 June 2026
The most common question I get sent goes something like this. “I built a system. I tested it. It made money on paper for weeks. So I started betting it for real, and it lost almost straight away. What happened?” If you have ever asked some version of that, you are in very good company, because it has happened to nearly every punter who has ever built a method and then watched it fall apart the moment real money went on it.
The frustrating part is the obvious conclusion you reach next: the system is broken. So you go back in and start fiddling with it, and somehow that makes things worse. Here is the uncomfortable truth, and it is worth sitting with for a second. In the overwhelming majority of these cases, the system did not break. The reason your selections lose has almost nothing to do with the rules you wrote and almost everything to do with two mistakes in how you went from paper to live.
Your system did not break, you did
Let me be clear about what I am assuming here. I am talking about a system you have genuinely paper traded on real, live selections going forward, not one you only ever ran across past results. If all you did was point your rules at historical data and admire the profit, that is a separate problem, and a bigger one, because you may simply have built something that fits the past and predicts nothing. Set that case aside. I am talking about the punter who did the honest thing, watched live selections for a stretch, saw a profit, and then got burned anyway.
For that punter, the losses are not evidence of a faulty system. They are evidence of two specific, avoidable errors made at the worst possible moment. Once you see them, you cannot unsee them, and you stop throwing away methods that were working all along.
The winning streak that pulled you off paper trading
Picture the typical sequence. You have spent weeks building something you believe in, and to be sensible you decide to paper trade it on live selections before risking a penny. Good. Then the first few days go brilliantly. The system hits a winning run, and every day you watch profit you are not collecting because you are only on paper. After three or four days of that, sitting on your hands becomes almost unbearable. You know it works. You have just proved it works. Carrying on paper trading feels like deliberately setting fire to money.
So you stop. You tell yourself you will start betting it live tomorrow, and you do. That decision, the one that feels like the smart and obvious move, is the first mistake. It feels like seizing an opportunity. What it actually does is set you up to fail, and the reason why is pure timing.
You started betting right at the top
Every profitable system in existence still goes through good runs and bad runs. The profit does not arrive smoothly, one winner after another. It arrives in waves, with winning streaks and losing streaks stacked on top of a long-run edge. That is not a flaw. That is simply what a real edge looks like when you watch it day to day.
Now follow the timing through. If you only started betting live because you had just seen a winning streak, then by definition you started at or near the top of an upswing. And the one thing that reliably follows an upswing is a downswing. So you walked through the door of your live account holding real money, just in time to catch a losing run. Worse, because this is a brand new system, you have never lived through one of its losing runs before. You have no idea how deep it goes or how long it lasts. You are not braced for it, and a few days in you start to panic, convinced the whole thing has fallen apart, when in reality a losing run of several days is completely normal and could comfortably run longer.
Put a real number on it. Race Advisor’s analysis of more than 150,000 qualifying runners between 2013 and 2026 shows just how routine these runs are. Take the top-rated horse in every race, a signal that wins a healthy 33.4% of the time. Backed flat stakes in date order, that signal still went ten or more bets without a single winner on more than 900 separate occasions, and once ran 27 bets deep before a winner landed. So a losing run of several days is not a warning light. It is the normal weather of a method that genuinely works.
This is the part most punters never grasp. The downswing did not appear because you started betting real money. It was always coming. You just timed your entry so that you bought in at the peak and met the dip with cash on the table and no experience of how it feels.
Changing the system after a bad week is just backfitting
Here comes the second mistake, and it is the one that does the real, lasting damage. A week of losses lands, you decide the system is broken, and you go back in to fix it. You look at the last week of results, find the tweak that would have avoided those losers, and apply it. The next day you bet the adjusted rules and lose again. Worse, the original rules would have won. So you switch back, and lose. Whatever you change, the losing continues, and now you are genuinely lost.
What you are actually doing has a name. It is backfitting, and you are doing it to yourself in real time. You are building a brand new, completely untested system fitted to one week of data, then throwing away the carefully developed method you spent weeks on. A system shaped to a handful of recent results predicts nothing about tomorrow, so of course it keeps losing. Then you reshape it again to the newest losers, and again, chasing your own tail one day at a time. The single most useful instinct you can develop here is to stop changing it. A process you built with care should only ever be altered on a large body of results and a great deal of thought. Reacting to a week is not adjustment, it is panic with extra steps.
How to paper trade so going live does not blindside you
The fix for both mistakes is the same, and it is almost annoyingly simple. Before you start, pick a paper-trading window and commit to it. One month, two months, whatever you are comfortable with, but decide the length in advance and then hold to it no matter what the results do. If it wins for the first week, you keep paper trading. If it loses for the first week, you keep paper trading. The results during the window do not shorten it or extend it.
The reason this works is that paper trading is not really about confirming the system makes money. You did that when you built it. The window is there to train you. It puts you through at least one full cycle of upswing and downswing while there is no money at stake, so that by the time you go live you already know in your bones what a losing run feels like and roughly how long this system tends to stay underwater. You go in braced, not blindsided, and a bad week becomes a shrug rather than a crisis. That single shift in realistic expectations is the difference between a punter who survives long enough to collect the edge and one who quits or tinkers it to death first.
Picking winners was never really the problem
There is a deeper reason these losing runs hit harder than punters expect, and it sits underneath everything above. Landing winners is not the same as making a profit. The market is not stupid. The runners that look most likely to win are priced accordingly, so backing obvious winners at short prices can leave you going backwards even when you are right about the result more often than not. A real edge does not come from being right more often. It comes from being paid more than your selections are truly worth, often enough, over a large enough sample, that the long-run numbers tip in your favour.
The numbers behind that are stark. Across more than a million and a half runners between 2013 and 2026, the top-rated horse by Race Advisor’s PR rating wins about a third of its races, yet its A/E, the measure of whether it wins more often than its price implies, sits at just 1.02 to Betfair SP. Run down the ranks from first to fifth and below and that figure barely moves, holding between 1.02 and 1.07 the whole way. The VDW ratings tell the same flat story. The most-backed horse in the country is no different: the SP favourite wins more than a third of the time across nearly 146,000 races, but its A/E is only 1.03, and on Flat turf it slips to 0.99, below the break-even line. Picking the likely winner, even picking the favourite, hands you no automatic edge once the price is sitting in front of you.
That is exactly why a genuinely profitable system still spends chunks of its life losing. The edge on any single bet is thin, and it only shows up across a large number of bets, which is precisely why a week of results tells you almost nothing and why reacting to that week is so destructive. Hold that idea and the losing run stops feeling like proof of failure. It is just the shape of a real edge being collected the only way it can be: slowly, and over far more bets than one nervous week.
Keep the system, change the punter
So the next time your selections start losing the moment you bet them, run the checklist before you touch a single rule. Did you go live off the back of a winning streak, which means you almost certainly started at the top? Are you reacting to a few days of data, which means you are backfitting? If the answer to either is yes, the problem is the timing and the temperament, not the method. Choose a paper-trading window, sit through the full cycle, and leave the rules alone unless you have a mountain of results telling you otherwise.
All of that assumes you have built something worth keeping in the first place, a method developed and tested properly rather than fitted to the past. That is where a system that actually works begins. The custom race cards and the Race Predictor let you score every runner against the real conditions of a race using thousands of proprietary factors, and test how a set of rules would have performed before you risk anything, so that when you do go live, the only thing left to manage is yourself. Open today’s cards and build a method you will not be tempted to throw away on a bad week.
