Trading What I see !?

[ShaunWhite]

To solve my hedging problems and grab every last ounce of "value" I changed my dog bot to minimum possible stake and no-hedge

#1 Everyone recommended that a beginner shouldn't do that.
#2 You don't know if your hedge bet is value or not because you don't have enough data to test that.

Trying to look on the positive side of things, 0.6% is about 1 in 167, so I shouldn't see such a losing streak again for another 166days! phew thank god for that! I should expect a day like this twice a year, there you go I feel better now

Twice a year? Your maths is waaay off, you've used basic counting theory but to work that one out is harder than it seems, you need to use a discrete time Markov chain.

Presumably you're backing iro 250 dogs a day, with a losing probability of 87.5%, so your chance of a 38 losing streak on any one day is 16.2%, during the week it's 75% (74.99%).

An event that occured twice a year would need a daily probability of 0.55% which, for 250 selections a day, is a streak of 63.

Scale accordingly if it's not 250/day

which probably means my strat is not anywhere near as good as I hoped backtesting was showing...Crap!
Ok, i'm going to keep running it, the only way to know is forward testing....

#1 You didn't test with enough data
#2 You didn't bother paper trading it which would have pointed to the likely slippage from test to live

Backtesting is not the way to find a profitable quant strategy, linear regression of data and derived data seems to be the way.

I don't know how you can make an assertion like that, analysis like you describe is step #1, backtesting is step #2, paper trading is step #3, live is step #4

[ShaunWhite]

I understand your frustation Goat,

I had an idea for a new dogbot this morning, I knocked up a rules file for it and ran it in practice mode, I let it run on the next six races and yippee!! all six gave very good returns, not a sniff of a loser. I then let it off the leash, very small stakes, live mode...............lost on every one of the next six races

dogbot disabled!

goodluck all

12 races or 120 races will tell you nothing. And practice mode only tells you the bot is functioning as expected, it's not a test that will give you anything like the necessary degree of certainty.

[goat68]

Just checked today's AUS horse bot live trades vs backtest of same stream data, and difference of 2 extra live trades.
I did a bit of log checking, and the reason unsurprisingly is in the "matching", the bot has a short 10second fill or kill. One of the trades the bet was taken after 6 seconds live, but failed to match after 10 in backtest. Given queue matching simulation is super difficult to emulate and it is bound to differ to live, and combine with the short 10second kill period, i'm bound to end up with apples vs oranges...!
This begs the question of the usefulness of backtesting, paper trading,...
The only way is to have some "market knowledge" argument as to why your strategy might be profitable, and then throw a small pile of money at it!

[decomez6]

I am starting to think anyone thinking of trying to make any money at all from sports trading starting from little experience of sports markets or how they work... STOP NOW!

🖐 …” hold for one minute please”
Before you throw the little angel away with its bath water !
How would you rate the bots scattered around in the forum .?
For those of us with no backtesting options ,we are left wondering , …are the bots worth a dime/time ?
Have you gone through the lot , Can you recommend any of them .
…May be it’s time to have aStar rating system for them

[goat68]

I am starting to think anyone thinking of trying to make any money at all from sports trading starting from little experience of sports markets or how they work... STOP NOW!

🖐 …” hold for one minute please”
Before you throw the little angel away with its bath water !
How would you rate the bots scattered around in the forum .?
For those of us with no backtesting options ,we are left wondering , …are the bots worth a dime/time ?
Have you gone through the lot , Can you recommend any of them .
…May be it’s time to have aStar rating system for them

I'm not using BA at the moment I am using BetfairLightWeight and Flumine framework.
However, when I was a few months ago, I did try some bots, in particular the VWAP crossover bot, in fact I took that quite a way for some considerable time. Those BAF examples don't stand up on their own, you need to apply them, which is where the skill is, as they are they are just "tools". The VWAP one definitely has some legs, and i've used bits of it in my AUS horse value strategy in Flumine, so from that perspective of a "toolkit" for bot makers, i'd give them all 5*****'s.

[decomez6]

Those BAF examples don't stand up on their own, you need to apply them

finding the right markets sucks ,its un-thankfull and costly ,takes a while to identify donkeys while flogging dead horses . furthermore ,practice mode wont cut it.... (fillrate , liquidity and active response nightmare)

The VWAP one definitely has some legs

how about BSP /Vwap mapping .any good?
-Or
average true range (ATR)

a "toolkit" for bot makers, i'd give them all 5*****'s.

keeps one bussy with plenty of half finished projects .

[ShaunWhite]

finding the right markets sucks ,its un-thankfull and costly ,takes a while to identify donkeys while flogging dead horses . furthermore ,practice mode wont cut it.... (fillrate , liquidity and active response nightmare) .

You'll never do it by trial and error. Think about the maths and how long you'd have to run each variation and with each being run on a different set of races how will you compare results?

Manual traders take a year to learn their craft and if you want to automate then expect the same, or longer. Get serious, create a plan to get to what you need and execute it. Data being #1, decent excel or coding skills, reasonable understanding of probability and stats and some database/data analysis techniques. There's no shortcuts, all jobs need some sort of skills especially ones that have the potential to pay well. You need more than a pencil to become an architect so why do you think you only need BA to be a trader? It's not amateur hour, do it right or don't bother.

[goat68]

finding the right markets sucks ,its un-thankfull and costly ,takes a while to identify donkeys while flogging dead horses . furthermore ,practice mode wont cut it.... (fillrate , liquidity and active response nightmare) .

You'll never do it by trial and error. Think about the maths and how long you'd have to run each variation and with each being run on a different set of races how will you compare results?

Manual traders take a year to learn their craft and if you want to automate then expect the same, or longer. Get serious, create a plan to get to what you need and execute it. Data being #1, decent excel or coding skills, reasonable understanding of probability and stats and some database/data analysis techniques. There's no shortcuts, all jobs need some sort of skills especially ones that have the potential to pay well. You need more than a pencil to become an architect so why do you think you only need BA to be a trader? It's not amateur hour, do it right or don't bother.

words not minced there

[goat68]

finding the right markets sucks ,its un-thankfull and costly ,takes a while to identify donkeys while flogging dead horses . furthermore ,practice mode wont cut it.... (fillrate , liquidity and active response nightmare) .

You'll never do it by trial and error. Think about the maths and how long you'd have to run each variation and with each being run on a different set of races how will you compare results?

Manual traders take a year to learn their craft and if you want to automate then expect the same, or longer. Get serious, create a plan to get to what you need and execute it. Data being #1, decent excel or coding skills, reasonable understanding of probability and stats and some database/data analysis techniques. There's no shortcuts, all jobs need some sort of skills especially ones that have the potential to pay well. You need more than a pencil to become an architect so why do you think you only need BA to be a trader? It's not amateur hour, do it right or don't bother.

Although, Shaun's obviously right... i'm just mustering energy for my next move, it's hard when i've got a busy day job and a house to look after as well, but i've mentally researched my next steps, just preparing the gusto to launch!

[goat68]

finding the right markets sucks ,its un-thankfull and costly ,takes a while to identify donkeys while flogging dead horses . furthermore ,practice mode wont cut it.... (fillrate , liquidity and active response nightmare) .

You'll never do it by trial and error. Think about the maths and how long you'd have to run each variation and with each being run on a different set of races how will you compare results?

Manual traders take a year to learn their craft and if you want to automate then expect the same, or longer. Get serious, create a plan to get to what you need and execute it. Data being #1, decent excel or coding skills, reasonable understanding of probability and stats and some database/data analysis techniques. There's no shortcuts, all jobs need some sort of skills especially ones that have the potential to pay well. You need more than a pencil to become an architect so why do you think you only need BA to be a trader? It's not amateur hour, do it right or don't bother.

words not minced there

and fyi, my next move relates to the data analysis techniques Shaun eludes, linear regression in my case

[ShaunWhite]

words not minced there

Yep. I don't subscribe to this "you can be anything if you dream hard enough" or "boo hoo it's too hard". Roll your sleeves up and get on with it. But that said I do think good traders are born not bred because you can't teach common sense and that's all trading is. So basically start savvy and graft. IQ doesn't help much, I've seen lots of smart people fail because as my old Ma says, they're too clever for their own good.

Keep that work/life/hobby stuff in balance goat, and don't expect to progress too quickly and feel rushed from one thing to another. You're probably sick of me saying the devil is in the detail, it's slowly slowly catchy monkey and shortcuts only catch up with you later. There's no Agile development here

I'm not sure why I keep thinking people want these streams of consciousness.

[goat68]

words not minced there

Yep. I don't subscribe to this "you can be anything if you dream hard enough" or "boo hoo it's too hard". Roll your sleeves up and get on with it. But that said I do think good traders are born not bred because you can't teach common sense and that's all trading is. So basically start savvy and graft. IQ doesn't help much, I've seen lots of smart people fail because as my old Ma says, they're too clever for their own good.

Keep that work/life/hobby stuff in balance goat, and don't expect to progress too quickly and feel rushed from one thing to another. You're probably sick of me saying the devil is in the detail, it's slowly slowly catchy monkey and shortcuts only catch up with you later. There's no Agile development here

I'm not sure why I keep thinking people want these streams of consciousness.

Hmm, spent a couple of hours producing some model test data for linear regression, and it's blatantly obvious there's nothing very linear about a market!! as expected there's a very random look about it...
money chunks coming in here and there
prices move one way, then another,...

Hmm, puzzling to think how quant guys create a fair price model...?
A few SMAs may help possibly!

[decomez6]

.

You'll never do it by trial and error. Think about the maths and how long you'd have to run each variation and with each being run on a different set of races how will you compare results?

Every race category has its own variables, every one of those variables can make the difference between a loosing and a winning .
Then comes the running costs of a bot and it’s consequent variations , which can turn out to be as good as random coin flip .
I am beginning to feel like a liquidity provider with not much to show for it , you say it will take , a year if not years!! All that liquidity I shall be recycling through the market just to feed my curiosity.

worse of all bots , is one that starts off nicely, with promising results, then tanks with a nose dive and a double dip in succession.
Most people will give up , but the likes of af a few will soldier on, hoping to be the shauns and the Peters in the making .
How many bullets can one dodge ?
So wish me luck ( I need it ) Hope to see you on the other side of the matrix !!

[decomez6]

You need more than a pencil to become an architect so why do you think you only need BA to be a trader? It's not amateur hour, do it right or don't bother.

You mean to say , BA is an entry level kind of success and one needs more than that to run a successful sports trading career?
How far can BA go?
May be it’s time to jot down a list of things that BA won’t do.
keep expectations in check.

[wearthefoxhat]

words not minced there

Yep. I don't subscribe to this "you can be anything if you dream hard enough" or "boo hoo it's too hard". Roll your sleeves up and get on with it. But that said I do think good traders are born not bred because you can't teach common sense and that's all trading is. So basically start savvy and graft. IQ doesn't help much, I've seen lots of smart people fail because as my old Ma says, they're too clever for their own good.

Keep that work/life/hobby stuff in balance goat, and don't expect to progress too quickly and feel rushed from one thing to another. You're probably sick of me saying the devil is in the detail, it's slowly slowly catchy monkey and shortcuts only catch up with you later. There's no Agile development here

I'm not sure why I keep thinking people want these streams of consciousness.

A few SMAs may help possibly!

I'd pitch in and say, yes to this. Certain racetypes/price ranges have less randomness than others, so the trend picking is more reliable.

If you could use your back-data and re construct a race over time periods pre-off using SMA'S, you may spot something.