Trading What I see !?

[goat68]

Rubbish week...this method is pointless!

PnLBot.jpg

[beermonsterman]

Rubbish week...this method is pointless!
PnLBot.jpg

What method are you using goat see if I can give you a few pointers
Can you trade manually ?
do you have an edge or method

[goat68]

Rubbish week...this method is pointless!
PnLBot.jpg

What method are you using goat see if I can give you a few pointers
Can you trade manually ?
do you have an edge or method

Thanks beer, so my day job now means I can't really trade manually any more, hence automation..
Edge is hiding!
My current bot is based on what I posted the other day, trigger based on selection volume surge, I then either bet in same direction or reverse. Current bot this week is reverse, with no stop loss, and a fixed take profit order of about 5%.
As above, i'm now going to go down the API route, collect some data, and create a simulator, so I can back test strategies quickly at minimal cost, other than a API key.

[speedyhamster]

Are you saying there are 80 to 90 people, who can write a high number of programs or bots to consistently make money on fundamentally random markets?

do you think the most successful traders might learn to earn compound interest instead of paying it?

They do seem random markets to me, they were totally different today for my bot, another -£6 today...
So that's it, im not going to keep guessing with this bot.
I am going to capture market data and develop a simulator, to back test theories...
So capture data every 2seconds or so..runners prices volume matched vol queued orders...
Please suggest anything you think might be useful?
Thanks

Use the API and record the raw streaming data so you can parse when required and you don’t need to make those sort of decisions, 10 iterations down the line you will start kicking yourself for not having recorded something.

So you can parse data that you don't know what it means, write 10 versions of a code and make a fortune?
are your perceptions of the market the same as the perceptions of the others in the market currently?

[vhdgkl]

Rubbish week...this method is pointless!
PnLBot.jpg

What method are you using goat see if I can give you a few pointers
Can you trade manually ?
do you have an edge or method

Thanks beer, so my day job now means I can't really trade manually any more, hence automation..
Edge is hiding!
My current bot is based on what I posted the other day, trigger based on selection volume surge, I then either bet in same direction or reverse. Current bot this week is reverse, with no stop loss, and a fixed take profit order of about 5%.
As above, i'm now going to go down the API route, collect some data, and create a simulator, so I can back test strategies quickly at minimal cost, other than a API key.

before I venture into research...tell me if it's something like this...the lower the volumes, the more just before the start the favourite goes down?

[goat68]

They do seem random markets to me, they were totally different today for my bot, another -£6 today...
So that's it, im not going to keep guessing with this bot.
I am going to capture market data and develop a simulator, to back test theories...
So capture data every 2seconds or so..runners prices volume matched vol queued orders...
Please suggest anything you think might be useful?
Thanks

Use the API and record the raw streaming data so you can parse when required and you don’t need to make those sort of decisions, 10 iterations down the line you will start kicking yourself for not having recorded something.

So you can parse data that you don't know what it means, write 10 versions of a code and make a fortune?

So I see what you're saying...
So my assumption is there are automatable "patterns" that are long term profitable, it's just a matter of modelling/guessing them at minimal cost. Using my current method of try something run it for a week and lose £10 could take years to find something...If I collect data and create a simulator, I can come up with a "model" put it through the simulator and it tells me how profitable or not it might be based on historic collected data, in a matter of minutes...

[beermonsterman]

Rubbish week...this method is pointless!
PnLBot.jpg

What method are you using goat see if I can give you a few pointers
Can you trade manually ?
do you have an edge or method

Thanks beer, so my day job now means I can't really trade manually any more, hence automation..
Edge is hiding!
My current bot is based on what I posted the other day, trigger based on selection volume surge, I then either bet in same direction or reverse. Current bot this week is reverse, with no stop loss, and a fixed take profit order of about 5%.
As above, i'm now going to go down the API route, collect some data, and create a simulator, so I can back test strategies quickly at minimal cost, other than a API key.

Cool mate that is a good move its an avenue I'm going to explore also I wish you well
I can manually trade pretty well but trying to automate it is another story my rules are that strict it never triggers often probably 3 to 6 times a week which leaves a lot of ticks on the table

[beermonsterman]

What method are you using goat see if I can give you a few pointers
Can you trade manually ?
do you have an edge or method

Not sure what you mean their but low volume to me means volatile

Thanks beer, so my day job now means I can't really trade manually any more, hence automation..
Edge is hiding!
My current bot is based on what I posted the other day, trigger based on selection volume surge, I then either bet in same direction or reverse. Current bot this week is reverse, with no stop loss, and a fixed take profit order of about 5%.
As above, i'm now going to go down the API route, collect some data, and create a simulator, so I can back test strategies quickly at minimal cost, other than a API key.

before I venture into research...tell me if it's something like this...the lower the volumes, the more just before the start the favourite goes down?

[speedyhamster]

Rubbish week...this method is pointless!
PnLBot.jpg

What method are you using goat see if I can give you a few pointers
Can you trade manually ?
do you have an edge or method

Thanks beer, so my day job now means I can't really trade manually any more, hence automation..
Edge is hiding!
My current bot is based on what I posted the other day, trigger based on selection volume surge, I then either bet in same direction or reverse. Current bot this week is reverse, with no stop loss, and a fixed take profit order of about 5%.
As above, i'm now going to go down the API route, collect some data, and create a simulator, so I can back test strategies quickly at minimal cost, other than a API key.

If you was there trading it you could see it is not reversing so get out of the trade (same direction)
risk reward ratio, risk stake to earn 5%?
could you not use stored values and hedge out for small loss instead of a stop loss?

[speedyhamster]

What method are you using goat see if I can give you a few pointers
Can you trade manually ?
do you have an edge or method

Thanks beer, so my day job now means I can't really trade manually any more, hence automation..
Edge is hiding!
My current bot is based on what I posted the other day, trigger based on selection volume surge, I then either bet in same direction or reverse. Current bot this week is reverse, with no stop loss, and a fixed take profit order of about 5%.
As above, i'm now going to go down the API route, collect some data, and create a simulator, so I can back test strategies quickly at minimal cost, other than a API key.

Cool mate that is a good move its an avenue I'm going to explore also I wish you well
I can manually trade pretty well but trying to automate it is another story my rules are that strict it never triggers often probably 3 to 6 times a week which leaves a lot of ticks on the table

You can see when a variation is happening in the market and take action to correct or get out of the trade, you could use a servant to take action when u see what the market is doing, I cant see how you can fully automate that in pre play

[ShaunWhite]

So I see what you're saying...
So my assumption is there are automatable "patterns" that are long term profitable, it's just a matter of modelling/guessing them at minimal cost. Using my current method of try something run it for a week and lose £10 could take years to find something...If I collect data and create a simulator, I can come up with a "model" put it through the simulator and it tells me how profitable or not it might be based on historic collected data, in a matter of minutes...

There's plenty of winning strategies that make a loss for a week or more. It's not that it'll take years to find something it's that you won't even know if what you're looking at is profitable.

There's no guarantee that anything you find in the past will continue in the future, but as long as you don't fall into the trap of backfitting, and you use in and out of sample data correctly, there's a pretty good chance it will.

You asked about what data to collect. Anything and everything...what you think you don't need today is exactly what you'll need in 6 months and you'll wish you had it. You might also want to consider collecting supplementary non Betfair data scraped from eleswhere too. eg you'd been talking about spikes, you can't define a spike' without a good estimate of the expected market vol at any given point in time, and that's largely governed by race class.....but race class isn't in Betfair data.

But your data will reflect your approach, will you be looking at 100+ parameters to determin a fair price (aka a mini Bill Benter) or will you be looking at market micro mechanics that exist in every market everywhere from dogs to pork bellies. It's about finding your own niche because it's easier to find an edge where the snow is less yellow.

At this rate goat you'll be getting into the gps api.

[goat68]

So I see what you're saying...
So my assumption is there are automatable "patterns" that are long term profitable, it's just a matter of modelling/guessing them at minimal cost. Using my current method of try something run it for a week and lose £10 could take years to find something...If I collect data and create a simulator, I can come up with a "model" put it through the simulator and it tells me how profitable or not it might be based on historic collected data, in a matter of minutes...

There's plenty of winning strategies that make a loss for a week or more. It's not that it'll take years to find something it's that you won't even know if what you're looking at is profitable.

There's no guarantee that anything you find in the past will continue in the future, but as long as you don't fall into the trap of backfitting, and you use in and out of sample data correctly, there's a pretty good chance it will.

You asked about what data to collect. Anything and everything...what you think you don't need today is exactly what you'll need in 6 months and you'll wish you had it. You might also want to consider collecting supplementary non Betfair data scraped from eleswhere too. eg you'd been talking about spikes, you can't define a spike' without a good estimate of the expected market vol at any given point in time, and that's largely governed by race class.....but race class isn't in Betfair data.

But your data will reflect your approach, will you be looking at 100+ parameters to determin a fair price (aka a mini Bill Benter) or will you be looking at market micro mechanics that exist in every market everywhere from dogs to pork bellies. It's about finding your own niche because it's easier to find an edge where the snow is less yellow.

At this rate goat you'll be getting into the gps api.

Don't tell me each horse has a garmin on it?!

[goat68]

So I see what you're saying...
So my assumption is there are automatable "patterns" that are long term profitable, it's just a matter of modelling/guessing them at minimal cost. Using my current method of try something run it for a week and lose £10 could take years to find something...If I collect data and create a simulator, I can come up with a "model" put it through the simulator and it tells me how profitable or not it might be based on historic collected data, in a matter of minutes...

There's plenty of winning strategies that make a loss for a week or more. It's not that it'll take years to find something it's that you won't even know if what you're looking at is profitable.

There's no guarantee that anything you find in the past will continue in the future, but as long as you don't fall into the trap of backfitting, and you use in and out of sample data correctly, there's a pretty good chance it will.

You asked about what data to collect. Anything and everything...what you think you don't need today is exactly what you'll need in 6 months and you'll wish you had it. You might also want to consider collecting supplementary non Betfair data scraped from eleswhere too. eg you'd been talking about spikes, you can't define a spike' without a good estimate of the expected market vol at any given point in time, and that's largely governed by race class.....but race class isn't in Betfair data.

But your data will reflect your approach, will you be looking at 100+ parameters to determin a fair price (aka a mini Bill Benter) or will you be looking at market micro mechanics that exist in every market everywhere from dogs to pork bellies. It's about finding your own niche because it's easier to find an edge where the snow is less yellow.

At this rate goat you'll be getting into the gps api.

Yes, "my data" approach makes sense... I actually quite like BA Guardian, so for the moment as it will take a while to get up and going with external API, I am going to run a Guardian rule to collect 1second data of all BA metrics guardian gives me, then create a simulator for that data, that should be relatively easy and quick for me to do. It's a half decent first approach I think for me.

[speedyhamster]

I see lots of spikes in the markets and see repeating patterns over and over again at a high percentage rate.
If your subconscious mind does not agree with your conscious mind about being a successful trader will lead to self sabotage
It is very easy to assume your tactics are wrong and adjust your tactics than considering your mindset might have issues

[goat68]

Data Collection rule ran well today, so happy with that.
Just knocked up a program to chart a market dataset from the database to verify the data looks good, 30mins-off, favs1-4, 1second intervals:

datachart.jpg

Happy with that