Backfitting – Our Friend
Hi. Clive Jones here. I am the guy behind Racecracker, a systems based tipping service run through Bet Chat. I have been asked to write an article for you about Backfitting, so I’ll start with what backfitting actually is and having checked the Internet definitions there is plenty to choose from and none of them right!
Essentially Backfitting is creating predictions for the future from information gained from the past. Backfitting is our friend – with pitfalls!
Every time you predict an outcome for a horse race (although this applies equally to other sports), you are probably backfitting to either a larger or smaller degree.
Do you bet horses that won last time out? That’s backfitting.
Using knowledge from the past to predict the future.
So, what are the pitfalls?
1. Past performance is no guarantee of future performance.
You’ve probably seen this mantra on all manner of financial advice, often in very small writing because if they are trying to sell you something, they don’t want you to think about this for very long.
That’s because it is true.
If you really do back horses just because they won last time out you will quickly come to realise this. In fact, if you had placed £1 on every, last time out winner in the last 5 and a half years you would have placed 65779 bets and lost £10901!
2. It must have some logical thought to it.
Here’s a system for you
Bets Wins Win% P/L(SP)
368 34 9.24 641.5
What could go wrong with that?
A system that has found 368 bets and won 641.5 points profit.
Looks as solid as a solid thing from the planet solid. Away you go and make your fortune.
What is it?
Well, unfortunately its five trainers’ runners in non-handicaps run on Mondays, Thursdays, and Fridays. So, the thought process must be “I’m not running my good thing on Tuesday, Wednesday, Saturday or Sunday, I’ll save him for my good days.”
No trainer thinks like that, I’m sure.
In fact, between 2011 and 2015 this system had 68 qualifiers for 2 winners!
This anomaly comes about due to a 125/1 winner in 2021, 100/1 winners in 2019 and 2018, a 200/1 winner in 2016 and a 100/1 in 2011 but that’s another story altogether
3. In order to produce profit it must provide value
This means that it must, at least in part, be unknown to the odds compilers.
Odds compilers are often very skilled at their job. Frighteningly so. Today, a Monday, there are 6 meetings with 42 races and 393 runners, and they have to come up with odds for every single runner.
To do so accurately they have to consider the going, the race distance, the current form of the trainer, the form of the horse including its backform, the run style of the horse, the jockey, the weight being carried, whether each horse prefers going left or right handed, etc, etc. There are probably hundreds of factors to give a full assessment of each race.
It used to be the case, before computers, that we, the punters, had access to very little of this. Remember the old sporting life newspaper on the bookie walls?
Well, that was pretty much it unless you could afford Timeform who, for a not insubstantial sum would send you indexed sheets to keep in files covering every race that you could use to trace back every runners past races.
And of course, you were only getting information the odds compilers already had.
You, at best, could only hope to know what they already knew.
This is why the great gamblers of the day like Barney Curley would own their own racehorses. It was a route to knowing something the odds compilers didn’t know.
Nowadays we have access to everything at our fingertips through excellent websites like geegeez.co.uk and horseracebase.com.
If you can type and read, you can have it all for a very reasonable cost. That gives you a chance.
So, you should be simply looking for horses that are priced incorrectly!
That’s where the number of meetings is to our advantage.
Those 393 horses running today can hardly be all priced correctly. Let the odds compiler have 390. I’ll take three a day and wait for my winners.
4. Know when to stop
If you use horseracebase.com as many of us do, the key to finding successful systems is knowing when to stop. You don’t want a system to throw up 3000 bets a year. And at each stage of refinement, you are making the number of selections smaller. But if you go too far, you will be in danger of creating the backfitting that others usually mean when they say beware of backfitting.
Take a Kevin Ryan system I have in HRB.
In the past five years (that’s the point at which I measure all systems) for sprint races.
If you start with Kevin Ryan, 5-7 furlong handicaps you will find 1173 runners for a loss of 261 points. Not very promising!
But you don’t want 1173 bets anyway so let’s see how much we can clear.
Restricting it to just 6 furlongs it becomes -65 points from 559 runners.
Then adding straight 6 furlongs i.e., no bends you get 434 runners and +5 points.
Progress!
Does the class of race make a difference? Yes.
Now we are up to 229 runners for 69 points profit.
If you want 50 runners a year this would appear to be a good stopping point but at Racecracker we are looking for an average 5 a day across all systems so I’ll continue.
Limiting the horses age to 3-5 makes it 72 points profit from 156 runners and who he puts on top makes a difference too.
With Kevin Stott on board its now 39 runners for 64 points profit.
Ok we’ve lost a small bit of profit, but we’ve massively reduced the number of runners.
That’s the point that I stop, and I’ve run this system for Racecracker customers (and myself!) since August 2020 across 27 runners for 18 points of profit.
Is it all still logical? Well, yes.
Some trainers often either favour bends or straights. I suspect this is something to do with training facilities. The class is often important. They can’t all be the best trainers with the best horses and like the horses, trainers often find their own level. Age is important for sprinters. Just like humans they often want further as they get older. (There are exceptions – Copper Knight anyone?) and the most logical of all, the jockey. Clearly Kevin Stott is Ryan’s go to man when he wants to win.
It’s often a bit trickier when considering handicaps because the art of winning handicaps includes losing to get your handicap down and that clearly muddles the stats. Hopefully, Ryan knows the principles and runs at different distances, around bends or without Stott on board when he wants to lose.
Such shenanigans abound in horse racing!
Take the principles any further and you are in the realms of chance where one horse can win four times and skew the trainer’s performance or one big, priced winner can do the same.
Ryan has only had one dual winner in this system in five years and no winners above 12/1.
Is this beyond the odds compiler’s radar? I think so.
These horses have probably lost multiple times on tracks with bends and that is hard to spot.
Hopefully, it is just the sort of thing to keep this one ahead of the bookies for a while longer! Clive Jones – Racecracker