greyabbey
Colt
Hi all,
I've been away from the forum for a while, and it's great to see the trainer change system thread still going!
Among a lot of other interesting posts and threads was a very interesting challenge from
pete
I've tested a lot of systems and found a way of working that suits me, so I thought I'd share some of my thoughts here (and try and answer the question too!).
CONTEXT
A brief context about my context as a system bettor. I started to get interested in system bets as a way of reducing human error - specifically, my errors. I started testing some ideas here and on one or two other forums, particularly during the COVID era when I had a bit more time to while away on Horseracebase.
After a while, I had a lot of ideas built up and I decided to test them publicly elsewhere, which I'm still doing in another online space. I'm happy to discuss that privately but that's not why I'm writing this thread.
MY PREVIOUS DEFINITION OF "WORKED"
It's obviously important for Pete's question to have an idea of what we mean by "worked". I don't think there is a single good answer as everyone has their own versions of "what does good look like" when it comes to betting. For some people, it's a quick win, for others it's whatever looks good on a spreadsheet, for others it's a bunch of different things.
For me, I settled first on numerical/financial ideas of "worked". Good ROI, good P/L, good strike rate.
In other words: past £££ performance as an indicator of future £££ performance.
I've drifted away from that, which I'll come back to later.
But throughout I'll use "for me" because I think you need to find what works and what is comfortable for you.
WHAT I DID AND WHY THAT DIDN'T WORK FOR ME
When I got into betting using more than a copy of the Racing Post and a cuppa, I did a bit of digging and eventually settled on Horseracebase.
The thread I stole Pete's question from has a decent discussion of the various options. Again, what works is up to you. I use Horseracebase. It works for me. Geegeez, Proform, others work for others.
I had a few ideas that I started to run with. Looking back, some still suit me well (although largely with some tweaking), some I've dropped entirely, some I've kept an eye on without betting just to see how they do.
I genuinely can't remember all the ideas. Initially, I started using Horseracebase during a lockdown so it was generally sat with a beer with nothing to do of an evening browsing through the 100s of criteria/categories. I summarised a few of them here back in 2021.
I have one or two versions of those ideas that I still use but for the most part I don't use them anymore.
The main reason is that most of these are what I would regard as "backfitted": I was looking back at previous performance of systems and using that previous performance as my main indicator of whether it would work in the future.
And that still probably has to play a part. You probably want some evidence that your idea has worked previously/recently. But for me it was too dominant in my initial ideas and (again, for me personally) is too dominant in lots of people's thinking when it comes to systems.
There's an excellent series of blogs on system betting by Dave Renham on Geegeez. I'd warmly recommend them. I use some of the ideas as prompts for my own systems. But I think most of them are driven first by backfitted past data rather than an idea.
And that's what I realised - for me - didn't work. I found on most of the systems I tested that I'd have 10 years of previously positive financial results then I'd test it for 6 months and it would be badly loss making. And I realised that either I had to keep on testing it because I had confidence in the system as an idea (1) or I should ditch it (2).
For most of them I was being led too much - for me - by past data than the idea that is the actual system. I realised that - rightly or wrongly - I doubted whether the systems were previously profitable because they were good systems or because of coincidence.
And whether that was right or wrong, once you doubt your betting, you'll do the worst thing and stop backing it until you miss a big winner and then hop on again.
So i just abandoned them entirely.
And to me, the idea just matters. One big defining feature of system betting is that you back qualifier horses whether or not you like their chances. If you don't actually like the idea that's generated the qualifiers, for me it's a bit of a death sentence.
To come back to
pete 's specific questions, one generalisable example category which I largely moved away from that didn't work for me but I think still does work for lots of others is systems revolving around individuals. I.e. certain trainers, or certain jockeys, or certain T/J combinations, or trainers at certain tracks.
I have a tiny handful of these either because there is an underlying idea behind why it might "work" or just because the numbers keep doing well and I'll ride that rollercoaster until everyone else breaks it.
Some people love them, and that's great for them. Just generally not for me. For me, they just don't work well enough as a "system". My rationale for this is:
- The individuals involved (i.e. the trainer or whomever is your system idea) did well coincidentally and probably won't repeat it again; or
- They did very well and there was a good reason for it, but everybody else has noticed and it's probably not going to be profitable in the future as a result; or
- They did very well, there was a good reason for it, and nobody else has noticed the thing you've noticed. But you don't know what that reason is, so as an idea to keep faith in for the future, it's sketchy.
For others, their rationale will be different. And that's fine. The joys of the market.
MY CURRENT DEFINITION OF "WORKED"
So once I'd figured all that out and wasted many an evening and bottle of lockdown beer studying Horseracebase, I decided to try and think of ideas that I *would* want to have faith in.
I started from my working definition of value: a bet where your interpretation of its chance of happening is higher than the market's interpretation.
And then extending that to a definition of a successful system: a set of criteria that I think will win more often than the market thinks it will.
I wrote on this forum a long time ago a post in jest but I can't find it. It was a set of humorous things you'd love to know to make up a system: whether the trainer is travelling 500 miles to run one horse because they really fancy it and think it's worth the drive, or because they're on their way to visit their Auntie Margaret and she fancied a dayout.
In jest! But it does neatly summarise what for me is key to a system: you think you know something better than the market.
That then led me to some basic criteria for any system I might want to use:
- At least one defining feature that I think materially affects a horse's chance in a race (e.g. a trainer change, to return to an old favourite)
- A rationale I could justify to myself that the market might underplay that defining feature(s)
Then - and only then - did I then look at whether or not those systems had performed well previously. And I've conditioned myself nowadays to not add or remove other categories solely because the data says it's worked before. The idea is the system, not the coincidental data.
MY DEFINITION(S) OF "WORKED" NOW
Those little snippets above are my foundation of system betting now. For me, what works - translated as what I personally choose to put money on - is an idea I have faith in, with some previous numerical performance to suggest it might work.
Those numbers are for me generally a mix of return on investment, strike rate, place strike rate, and Actual vs Expected.
Then I measure them on those same criteria going forward. Some systems I measure for a while without betting. Some I dive straight in with a betting bank.
Another final metric of "worked" for me, which again might not be for everyone, is whether that defining feature(s) that makes up your system was noticeable in the horse's performance or in the market's interpretation. Put crudely, actually watch the race. Does the system do what you thought it would?
Over time, I've started to stumble upon a nice mix: the systems that are my favourite ideas are also the ones that tend to perform well in future testing. Not always. But I think I'm getting there.
THE KINDS OF SYSTEMS I USE NOW
I have a load of different ideas that form the basis of systems. Some I hope would make sense if we were in a pub and I was defending them to you over a pint. Others work for me but would seem nonsensical to others. Some are the foundation of some deep thinking. Others are based literally on a throwaway remark in a blog or an interview. But at the heart of them is that I can justify the idea.
Hope that's somewhat useful to someone...!
I've been away from the forum for a while, and it's great to see the trainer change system thread still going!
Among a lot of other interesting posts and threads was a very interesting challenge from
peteI've tested a lot of systems and found a way of working that suits me, so I thought I'd share some of my thoughts here (and try and answer the question too!).
CONTEXT
A brief context about my context as a system bettor. I started to get interested in system bets as a way of reducing human error - specifically, my errors. I started testing some ideas here and on one or two other forums, particularly during the COVID era when I had a bit more time to while away on Horseracebase.
After a while, I had a lot of ideas built up and I decided to test them publicly elsewhere, which I'm still doing in another online space. I'm happy to discuss that privately but that's not why I'm writing this thread.
MY PREVIOUS DEFINITION OF "WORKED"
It's obviously important for Pete's question to have an idea of what we mean by "worked". I don't think there is a single good answer as everyone has their own versions of "what does good look like" when it comes to betting. For some people, it's a quick win, for others it's whatever looks good on a spreadsheet, for others it's a bunch of different things.
For me, I settled first on numerical/financial ideas of "worked". Good ROI, good P/L, good strike rate.
In other words: past £££ performance as an indicator of future £££ performance.
I've drifted away from that, which I'll come back to later.
But throughout I'll use "for me" because I think you need to find what works and what is comfortable for you.
WHAT I DID AND WHY THAT DIDN'T WORK FOR ME
When I got into betting using more than a copy of the Racing Post and a cuppa, I did a bit of digging and eventually settled on Horseracebase.
The thread I stole Pete's question from has a decent discussion of the various options. Again, what works is up to you. I use Horseracebase. It works for me. Geegeez, Proform, others work for others.
I had a few ideas that I started to run with. Looking back, some still suit me well (although largely with some tweaking), some I've dropped entirely, some I've kept an eye on without betting just to see how they do.
I genuinely can't remember all the ideas. Initially, I started using Horseracebase during a lockdown so it was generally sat with a beer with nothing to do of an evening browsing through the 100s of criteria/categories. I summarised a few of them here back in 2021.
I have one or two versions of those ideas that I still use but for the most part I don't use them anymore.
The main reason is that most of these are what I would regard as "backfitted": I was looking back at previous performance of systems and using that previous performance as my main indicator of whether it would work in the future.
And that still probably has to play a part. You probably want some evidence that your idea has worked previously/recently. But for me it was too dominant in my initial ideas and (again, for me personally) is too dominant in lots of people's thinking when it comes to systems.
There's an excellent series of blogs on system betting by Dave Renham on Geegeez. I'd warmly recommend them. I use some of the ideas as prompts for my own systems. But I think most of them are driven first by backfitted past data rather than an idea.
And that's what I realised - for me - didn't work. I found on most of the systems I tested that I'd have 10 years of previously positive financial results then I'd test it for 6 months and it would be badly loss making. And I realised that either I had to keep on testing it because I had confidence in the system as an idea (1) or I should ditch it (2).
For most of them I was being led too much - for me - by past data than the idea that is the actual system. I realised that - rightly or wrongly - I doubted whether the systems were previously profitable because they were good systems or because of coincidence.
And whether that was right or wrong, once you doubt your betting, you'll do the worst thing and stop backing it until you miss a big winner and then hop on again.
So i just abandoned them entirely.
And to me, the idea just matters. One big defining feature of system betting is that you back qualifier horses whether or not you like their chances. If you don't actually like the idea that's generated the qualifiers, for me it's a bit of a death sentence.
To come back to
pete 's specific questions, one generalisable example category which I largely moved away from that didn't work for me but I think still does work for lots of others is systems revolving around individuals. I.e. certain trainers, or certain jockeys, or certain T/J combinations, or trainers at certain tracks.I have a tiny handful of these either because there is an underlying idea behind why it might "work" or just because the numbers keep doing well and I'll ride that rollercoaster until everyone else breaks it.
Some people love them, and that's great for them. Just generally not for me. For me, they just don't work well enough as a "system". My rationale for this is:
- The individuals involved (i.e. the trainer or whomever is your system idea) did well coincidentally and probably won't repeat it again; or
- They did very well and there was a good reason for it, but everybody else has noticed and it's probably not going to be profitable in the future as a result; or
- They did very well, there was a good reason for it, and nobody else has noticed the thing you've noticed. But you don't know what that reason is, so as an idea to keep faith in for the future, it's sketchy.
For others, their rationale will be different. And that's fine. The joys of the market.
MY CURRENT DEFINITION OF "WORKED"
So once I'd figured all that out and wasted many an evening and bottle of lockdown beer studying Horseracebase, I decided to try and think of ideas that I *would* want to have faith in.
I started from my working definition of value: a bet where your interpretation of its chance of happening is higher than the market's interpretation.
And then extending that to a definition of a successful system: a set of criteria that I think will win more often than the market thinks it will.
I wrote on this forum a long time ago a post in jest but I can't find it. It was a set of humorous things you'd love to know to make up a system: whether the trainer is travelling 500 miles to run one horse because they really fancy it and think it's worth the drive, or because they're on their way to visit their Auntie Margaret and she fancied a dayout.
In jest! But it does neatly summarise what for me is key to a system: you think you know something better than the market.
That then led me to some basic criteria for any system I might want to use:
- At least one defining feature that I think materially affects a horse's chance in a race (e.g. a trainer change, to return to an old favourite)
- A rationale I could justify to myself that the market might underplay that defining feature(s)
Then - and only then - did I then look at whether or not those systems had performed well previously. And I've conditioned myself nowadays to not add or remove other categories solely because the data says it's worked before. The idea is the system, not the coincidental data.
MY DEFINITION(S) OF "WORKED" NOW
Those little snippets above are my foundation of system betting now. For me, what works - translated as what I personally choose to put money on - is an idea I have faith in, with some previous numerical performance to suggest it might work.
Those numbers are for me generally a mix of return on investment, strike rate, place strike rate, and Actual vs Expected.
Then I measure them on those same criteria going forward. Some systems I measure for a while without betting. Some I dive straight in with a betting bank.
Another final metric of "worked" for me, which again might not be for everyone, is whether that defining feature(s) that makes up your system was noticeable in the horse's performance or in the market's interpretation. Put crudely, actually watch the race. Does the system do what you thought it would?
Over time, I've started to stumble upon a nice mix: the systems that are my favourite ideas are also the ones that tend to perform well in future testing. Not always. But I think I'm getting there.
THE KINDS OF SYSTEMS I USE NOW
I have a load of different ideas that form the basis of systems. Some I hope would make sense if we were in a pub and I was defending them to you over a pint. Others work for me but would seem nonsensical to others. Some are the foundation of some deep thinking. Others are based literally on a throwaway remark in a blog or an interview. But at the heart of them is that I can justify the idea.
Hope that's somewhat useful to someone...!
