Looking for an experienced trading bot builder/mentor

[Real Money]

I was looking for edge/quant models that can predict/estimate risk/reward and demand and/or positioning for size players in futures markets and synthetics.

It took a long time and ultimately involves fairly sophisticated statistics models.

Differential and integral functions of various regression model errors/residuals. This is similar to how you would design autopilot/robot/AI or automated control systems etc.

Futures markets are controlled by algos that use liquidity models based on things like volume, OI, vol, and resting/filled orders. IMO, its similar to this picture where the blue line represents volume buying and selling and the orange line represents the price or price vs something like VWAP or related assets/instruments.

IMO, full automation is an entirely different animal. Building useful models takes heavy math skills, but an actual managed robot takes very heavy programming chops and will entail significant risks. I wish I could tackle full automation but I'm much more of a mathematician than a programmer.

 

[Sprout]

I was looking for edge/quant models that can predict/estimate risk/reward and demand and/or positioning for size players in futures markets and synthetics.

It took a long time and ultimately involves fairly sophisticated statistics models.

Differential and integral functions of various regression model errors/residuals. This is similar to how you would design autopilot/robot/AI or automated control systems etc.

Futures markets are controlled by algos that use liquidity models based on things like volume, OI, vol, and resting/filled orders. IMO, its similar to this picture where the blue line represents volume buying and selling and the orange line represents the price or price vs something like VWAP or related assets/instruments.

View attachment 292801

IMO, full automation is an entirely different animal. Building useful models takes heavy math skills, but an actual managed robot takes very heavy programming chops and will entail significant risks. I wish I could tackle full automation but I'm much more of a mathematician than a programmer.

Luv the predator/prey analogy, although lambs might be more apropos when describing liquidity.

 

[themickey]

I was looking for edge/quant models that can predict/estimate risk/reward and demand and/or positioning for size players in futures markets and synthetics.

It took a long time and ultimately involves fairly sophisticated statistics models.

Differential and integral functions of various regression model errors/residuals. This is similar to how you would design autopilot/robot/AI or automated control systems etc.

Futures markets are controlled by algos that use liquidity models based on things like volume, OI, vol, and resting/filled orders. IMO, its similar to this picture where the blue line represents volume buying and selling and the orange line represents the price or price vs something like VWAP or related assets/instruments.

View attachment 292801

IMO, full automation is an entirely different animal. Building useful models takes heavy math skills, but an actual managed robot takes very heavy programming chops and will entail significant risks. I wish I could tackle full automation but I'm much more of a mathematician than a programmer.

I think for a sole trader, full automation or even partial automation is an impossible task.
There is just too much complexity involved with its creation for one person to manage.
I enjoy writing code but just algos not involved with automation, and some trading days so busy coding that I'm too distracted with coding I can't find time to trade.

 

[easymon1]

I enjoy writing code but just algos not involved with automation, and some trading days so busy coding that I'm too distracted with coding I can't find time to trade.

The tool-builders eternal dilemma. task, time, resources. Tool - not good enough - good enough - optimal - guilding the lilly.

 

[themickey]

The tool-builders eternal dilemma. task, time, resources. Tool - not good enough - good enough - optimal - guilding the lilly.

There's a saying, "everyone gets from the markets what they want".
Money obviously the prime reason you would think, but other reasons; entertainment, occupation, satisfaction, distraction, understanding, education, ego validation, improvement.
Building trading tools is hugely fun but you've gotta enjoy maths and challenges.
The more you do, the easier it becomes, but the road is long and winding.
It suits introverts who live in a cave.

 

[M.W.]

Why do a few people always pollute the very few high information and value content threads with nonsense? It seems to be an addiction. Can't wait for Overnight to chime in as well to post some stupid music clips. Sigh...

Damn skippy, it IS fun, lol. Doing it for non comercial, eg for yourself ya go no monthly reviews, updates, schedules, milestones, boss but the ones you set for yourself. In business as well as in trader's time, it becomes wise to 'fly it while you build it', an art in itself, lol. The toolmaker's delimma comes with them opposable thumbs I guess.
It's all good. Building beats the shit out of tearing down. Hate to see that waste of time, effort, resources, productivity.

Mongo likes tools. Tools Good!
View attachment 292814

Whaddaya think, too much? Needs wheels bad. Close though...
Maybe I'll get a driver's license next year...
View attachment 292815

 

[easymon1]

Squeeze me M.W.
Duly noted.

 

[easymon1]

A back test and live algorithmic trading architecture can be as simplistic or complicated as desired. It really is the choice of the user.

It's easy to throw a few python packages together and load some data, iterate over them and call a strategy function that feeds data to the actual algorithm. Every now and then the algorithm places orders that are filled in a simulated environment. Then at the end performance calculations are output.

I have worked years on my trading framework and architecture, probably now close to a decade. Of course did I trade on it in its earlier stages but I added and changed a lot of code and features over time as I saw fit.

One example is how I represent orders. Orders are represented by a ParentOrder class that can split off Child Orders and hence allows for fairly involved and non trivial order logic. For example I have a Basket Order type that can split off fractions of the original order size and hence submit parts of the total order size to get the basket legs submitted at a lower risk of unbalanced fills. It also allows for bracket order logic where target and stop loss levels are already baked into the Basket Order logic. Hence, orders become small algorithms themselves that are updated by backtest or live data that affect any of the basket legs. This is just one of many examples where the architecture logic can get fairly complex in order to support more advanced features.

I agree with some here who belittle those book writing clowns. They usually present the absolute basics which guarantee to not work or at the very least won't take a beginner very far. I recommend the best combination is professional experience in the trading space coupled with coding skills. Most developers who approach algorithmic trading would not even know about parent/child order logic unless they read or learned about it from someone who is involved in this space. The same goes with other features. Jus knowing how to code means nothing. You need to know what you want to achieve and why. Often times the biggest challenge is not knowing something exists that makes life much less complicated.

Do you allow your creations to trigger entries and exits?
How much of the average week do these operate in a more or less 'hands-off' fashion?

 

[M.W.]

5 out of 7 days. How do you define hands off? Either a framework is fully algorithmic or it's not.

Do you allow your creations to trigger entries and exits?
How much of the average week do these operate in a more or less 'hands-off' fashion?

 

[globalarbtrader]

be careful, Andreas is a book salesman, not a trader. He does not have a trade room, does not have a track record and has no recorded profits. He is all vendor/no trader and there 2000 more just like him.

Not true.

He is a fund manager and so has an audited track record.

WTF is a 'trade room'?

Disclosure: I wrote a chapter in that book.

Rob