A bit of reassurance.... please

[ruthlessimon]

to win long term you need to understand the reason why something should be winning.

Got any tips for this SB? I personally think this is one of the most difficult aspects.

For example, sometimes, I will "make an educated guess" where I expect something to outperform (i.e. here's an edge, here's where it should be even better) - but I wouldn't call that a "why" (& it certainly isn't sustainable). But usually, I'll "make a story", for why something is working/should work (i.e. people getting anchored) - but "whys" like that aren't really provable

[spreadbetting]

to win long term you need to understand the reason why something should be winning.

Got any tips for this SB? I personally think this is one of the most difficult aspects.

For me that's the easy bit. I always start any 'bot' with a reason of why I think the market will move/react, then I simply set about trying to disprove or prove I can make money from it. I might use stats to try to disprove it but usually I'll throw money into the market to find if my idea is workable or not but always have an initial idea of why it should work.

I couldn't approach the markets in the way you seem to, data mining 'winning' strategies as it'd most likely be far too hard for me to drill down to reason why something worked during that period. Once you start with stats you can never have enough, a 10 tick move at 3 minutes out will happen for reasons other than a 10 tick move 10 seconds before the off etc. Experience has shown me the simplest strategies are usually the biggest earners and it's a lot easier to find something that returns 1% rather than 20%, plus they're usually long lasting as everyone else wants the golden goose not the little eggs it's laying.

[northbound]

it's a lot easier to find something that returns 1% rather than 20%

If your trade risks £1 to make £1, you're basically saying that out of 101 trades, you won 56 and lost 55. Correct?

It's gonna take a massive sample to ascertain whether you really do have a 1% edge. Or is it me looking at this from the wrong angle?

[ruthlessimon]

For me that's the easy bit. I always start any 'bot' with a reason of why I think the market will move/react, then I simply set about trying to disprove or prove I can make money from it

Ty sb, a really interesting viewpoint

[ShaunWhite]

it's a lot easier to find something that returns 1% rather than 20%

If your trade risks £1 to make £1, you're basically saying that out of 101 trades, you won 56 and lost 55. Correct?

It's gonna take a massive sample to ascertain whether you really do have a 1% edge. Or is it me looking at this from the wrong angle?

The incorrect assumption is "risks £1 to make £1".
You risk £1 to make £x.
x can be a penny or a tenner.

The same sample size would prove a 0.1% edge as accurately as a 90% edge

[spreadbetting]

it's a lot easier to find something that returns 1% rather than 20%

If your trade risks £1 to make £1, you're basically saying that out of 101 trades, you won 56 and lost 55. Correct?

It's gonna take a massive sample to ascertain whether you really do have a 1% edge. Or is it me looking at this from the wrong angle?

I'm simply saying it's easier to find something that will return me £101 for every £100 I bet not necessarily 56/101 rather than something that will return me £120. If I can churn over enough 1%'s over time I'm happy enough with the crumbs everyone leaves behind.

I'm not sure why the forum seems obsessed with sample sizes, if I know something wins and think I have a good idea why it works I just bet on them I don't need to wait for 10,000 events to confirm it for me.

You're currently looking at how far the market moves just before the off to see if you can lay an overround book to get a profit. If say x-matching is turned off and the book is bouncing around 99% I know I can back all runners and guarantee a profit of 1%, I don't need a large sample to prove that to me, all I need to know is if I'm able to get in fast enough to profit from it by actually trying it.

[jimibt]

it's a lot easier to find something that returns 1% rather than 20%

If your trade risks £1 to make £1, you're basically saying that out of 101 trades, you won 56 and lost 55. Correct?

It's gonna take a massive sample to ascertain whether you really do have a 1% edge. Or is it me looking at this from the wrong angle?

I'm simply saying it's easier to find something that will return me £101 for every £100 I bet not necessarily 56/101 rather than something that will return me 20%. If I can churn over enough 1%'s over time I'm happy enough with the crumbs everyone leaves behind.

I'm not sure why the forum seems obsessed with sample sizes, if I know something wins and think I have a good idea why it works I just bet on them I don't need to wait for 10,000 events to confirm it for me.

SB, that's a breath of fresh air. i've always been a mechanics based *practitioner*, rather than a stats believer and think i get your approach quite well. i guess to acrue the profits, you need to be triggering out a fair number of those transactions. definitely the crumbs are there, as i'm sure your years of experience are telling you. as they say up here, MONY A MICKLE MAKS A MUCKLE!!

[northbound]

Fair enough and I'm sure the crumbs are there, otherwise all us relative newbies would have given up long ago

However, if we are not yet profiting consistently, it's because we haven't found angles that can reliably produce 1%. Or perhaps plenty of beginners find them, but they are the wrong angles: they produce 1% 100 times out of 101, then lose 100% 1 time out of 100.

Hence our questions, which I'm sure look really stupid when you are on the other side of the fence and have it all figured out.

[ShaunWhite]

There's more than one way to skin a cat. And all of those ways can be done well or done badly.
If everyone had the same technique then we'd all just be passing round the same mickle and never make a muckle.

[northbound]

There's more than one way to skin a cat. And all of those ways can be done well or done badly.

I agree

[ShaunWhite]

Hence our questions, which I'm sure look really stupid when you are on the other side of the fence and have it all figured out.

No no no no don't think that. It's NOT asking questions that makes people stupid.

Doesn't matter either if you need to ask once or ten times. More than 10 and yes, maybe they're stupid

[spreadbetting]

Fair enough and I'm sure the crumbs are there, otherwise all us relative newbies would have given up long ago

However, if we are not yet profiting consistently, it's because we haven't found angles that can reliably produce 1%. Or perhaps plenty of beginners find them, but they are the wrong angles: they produce 1% 100 times out of 101, then lose 100% 1 time out of 100.

Hence our questions, which I'm sure look really stupid when you are on the other side of the fence and have it all figured out.

Like Shaun says there aren't any stupid questions. If you don't understand something you ask til you do, the only stupid thing would be to worry about maintaining some kind of ego on an anonymous forum. It's easy enough to make another account if you've shot yourself in the foot on here.

[ruthlessimon]

because we haven't found angles that can reliably produce 1%.

I think SB's right about small edges, but mentally how many people could hack that, even with a bot?

Let's say I was averaging +1tick a race over 10,000 races - even with massive scale, that's still a crumby p&l imo.

Is a +1tick average really the limit of the opportunity available??

Which then opens a wider question which never really get's talked about - how does one improve a said edge (where are true limits to profitability)

[spreadbetting]

because we haven't found angles that can reliably produce 1%.

I think SB's right about small edges, but mentally how many people could hack that, even with a bot?

Let's say I was averaging +1tick a race over 10,000 races - even with massive scale, that's still a crumby p&l imo.

Is a +1tick average really the limit of the opportunity available??

Which then opens a wider question which never really get's talked about - how does one improve a said edge (where are true limits to profitability)

I'd have thought mentally that would be the whole point of running bots, so you can accept the small returns without feeling the need to up rates or stakes to unrealistic amounts.

I run a bot on the dogs that is averaging around 66p a race , a very simple strategy that I could easily increase the profitability of it if I did it manually. But the money just isn't there in the dog markets so even though I could increase it's profitability eventually it becomes a case of diminishing returns and my time is spent better elsewhere rather than spending all day on the dogs for maybe another 50% return. I've been running it for a long time and last 3 months it churned over 15000 markets so the pennies do add up.

[eightbo]

I'd like to share some extremely relevant words from Brent Penfold on the topic of defining your strategy, and 'mechanical' vs. 'discretionary' traders: https://www.youtube.com/watch?v=iAvBbBvth0Q&t=36m51s
36:51 - 42:55

改善