Tools being used by your trading competition

[relval]

With regard to ML/quantitative modelling, your output is only as good as your input.

1. Using simple inputs like technical indicators absolutely will not work. Rest assured your data has been analysed by thousands of others and any obvious inefficiency exploited. If you build enough models, some will pass validation purely by chance. You will think you have something, but you don't.

2. Getting more creative and using inputs like intermarket relationships, you will find some correlations that enable you to build some models with an apparent decent edge. Only problem is that these kinds of relationships aren't stationary. Best case scenario is the model works for awhile in production before decaying, you switch it off then recalibrate/build a new one. I believe this is how a lot of quant shops operate.

3. Finding inputs that have a stable predictive correlation over time is really difficult. But here's the thing - if you find something like this you don't need a neural net or anything fancy to implement it.

 

[Trendnomics]

Wow this thread is terrible!

Still stand to be corrected. Very entertaining, but adding no value at all!

Go for the ball not the man.

It has become quite apparent what real profitable traders/investors think of Howard's ball(s).

 

[Modest]

Quants and trading

 

[OmegaTrader]

Still stand to be corrected. Very entertaining, but adding no value at all!



It has become quite apparent what real profitable traders/investors think of Howard's ball(s).

The only reason you are saying that is because he commented on your thread.


When I asked you on your thread what your alpha was you still haven't replied.

That was dec 03-2016


Who are the real profitable traders?

 

[DeepState]

Thanks for your perspectives.

With regard to ML/quantitative modelling, your output is only as good as your input.


2. Getting more creative and using inputs like intermarket relationships, you will find some correlations that enable you to build some models with an apparent decent edge. Only problem is that these kinds of relationships aren't stationary. Best case scenario is the model works for awhile in production before decaying, you switch it off then recalibrate/build a new one. I believe this is how a lot of quant shops operate.

Correct on how quant shops operate. Utilisation of factors is not unlike picking stocks. Stocks you like will not be cheap forever. Factors you like are unlikely to be permanent unless of a type which is compensation for risk (eg. equities is expected to outperform cash because it is riskier to hold an equity than cash).

3. Finding inputs that have a stable predictive correlation over time is really difficult. But here's the thing - if you find something like this you don't need a neural net or anything fancy to implement it.

If your signal required a neural net to identify and construct, you are going to need a neural net to develop the signals. A neural net is not readily approximated by simple linear relationships, for example.

---

Howard has an axe to grind. If that's not welcome here, fair enough. This community does not react well to sales motivated posts. I understand that this forum is meant to be more of an exchange of ideas rather than a means to generate sales.

I don't begrudge the man his desire to commercialise his knowledge or experience. Perhaps this isn't the place for that.

His tool set is solid. The arguments have rationale that you can concretely debate. It is useful for some - maybe not you. Nevertheless, you may not like what he does with the tools or the rationale used to jump from point to point. I acknowledge that the technical matter is beyond the maths paygrade of many here, but his stuff is ABC to finance students and not an effort to obfuscate.

This is not unlike a situation where someone thinks P/B is the ant's pants and another thinks the Montgomery formula is the ant's pants. Both read the same balance sheet and prices. Perhaps treatment of debate participants should proceed accordingly...which is probably to heap scorn on the opposing idea relentlessly...haha...sigh

 

[Trendnomics]

The only reason you are saying that is because he commented on your thread.


When I asked you on your thread what your alpha was you still haven't replied.

That was dec 03-2016


Who are the real profitable traders?

Howard did more than comment - he stated that all the back-testing my system was based on is flawed. He also added his usual doom-and-gloom spiel.

I'm no longer able to provide posts on my system - I'm potentially entering an arrangement which would apply a non-disclosure agreement.

Minwa is one of the profitable traders (many more lurking and posting).

 

[OmegaTrader]

Howard did more than comment - he stated that all the back-testing my system was based on is flawed. He also added his usual doom-and-gloom spiel.

I'm no longer able to provide posts on my system - I'm potentially entering an arrangement which would apply a non-disclosure agreement.

Minwa is one of the profitable traders (many more lurking and posting).

That is not about you it is about your strategy. don't attack the man.
Ok, so are you making risk adjusted returns?

 

[minwa]

Ok, so are you making risk adjusted returns?

Well we know for sure Howard isn't making any - he stated he doesn't currently trade. And from the tone he hasn't for a while. Everything is theory and back tested stuff which just contradicts everything he is saying on using the latest buzz to level the playing field in the markets against your competition.

When you make big controversial claims in profiting in the markets, especially as an educator/author, people like results to back them up. All we really got was - "buy the book".

 

[barney]

Quants and trading

LOL ....
I was suspicious at the Unilateral phase detractors (23 sec), but automatically synchronizing the Cardinal Grand Meters at 28 seconds really had me super skeptical given that it was obviously a rip off of the much more reliable "turbo-encabulator" which has proven itself over time to be much more reliable ...

I kid you not!! ...Created by Engineers at General Electric back in the 60's it was the most advanced piece of 'non-engineering' ever! Superseded by the advanced model, the "Retro-Encabulator: in the late 80's

PS All the above meant in good fun and totally factual so no hate mail please

 

[howardbandy]

If growing up means conforming and not questioning, then better you don't grow up I reckon.

Disrespect is in the eye of the beholder. Its the "my way, nothing else will work" attitude expressed in the first post and follow up posts that I find disrespectful. And the calling the thread to an end when challenged that I find childish.

There's nothing wrong with the comprehension and understanding of those that have a different view.

I'll like thingy's post because I can't give tech's a dislike and I really dislike it.

Hi Craft --

I began to thread. With the simple intent of pointing out techniques that some of our competition in trading is using. The ensuing discussion went well for many postings. Every civil question addressed to me, or answerable by me, was answered. Then began both going considerably off-topic, and involving personal attacks. The purpose I had was complete.

I had no idea the first post would be interpreted as "my way, nothing else will work." My apologies. How should that material have been presented?

I am open to additional civil on-topic discussion. What more, related to the first post of the thread, should be discussed?

Best, Howard

 

[howardbandy]

Greetings --

I have read through the thread again. Please don't shoot the messenger.

But please do allow for an analysis of techniques being used.

On risk --

If you have not read through my postings on risk, or watched the presentation "The Four Faces of Risk" on YouTube, please do. The math starts there.

Any set of trades can be analyzed to determine the risk of drawdown in them, and if, as we all rely, the future resembles the past, the risk of future trading that same system. Traders using algorithms have an easier time doing the analysis, but discretionary traders can apply exactly the same procedures.

There are two fundamental facts about risk:
1. Risk increases with holding period.
2. Risk increases with trading inaccuracy.

Each trader has his or her own risk tolerance. It can be quantified. The risk associated with any set of trades can be compared with the trader's risk tolerance. If the risk of the trades is in excess of the trader's tolerance, position size is too high, there is a high probability of a drawdown that will cause either serious account loss or to stop trading. Lower the position size until the risk of the trades is within the risk tolerance of the trader. If the resulting position size is too low, do not trade that system.

There is no "my way or the highway" from me here. There is nothing here but the math of modeling, simulation, statistics, and the scientific method applied to a set of trades. The mathematical analysis does, however, establish some boundaries for what works and what does not. I hear you -- the boundaries are beyond what many of you are now doing. Wishing it were different does not make the math wrong. Neither does making personal criticisms or unreasonable demands of me.

On machine learning and artificial intelligence --

My original post quoted a statement made in a presentation that Morgan Stanley is employing several thousand developers working with Python. This was Stephen Simmons statement, not mine. My comments are intended to point out some ways us ordinary retail traders can still compete, even though it is becoming more difficult.

The responses in this thread include several that dismiss the competition from Morgan Stanley and those similar.

I understand the personal psychological investment we have in our creations and faith in our abilities. Some 90% of Americans believe they are among the top half of drivers in terms of safety. An item we already own is more valuable to us than one available for sale. One we have made ourselves is more valuable still.

Ignore the messages about artificial intelligence projects and accomplishments we are hearing if you wish. But realize this is not a single voice from science fiction, or from me. This is a tidal wave coming from giant players -- Morgan Stanley, Goldman Sachs, James Simons, David Shaw, Apple, Google, Microsoft, Facebook, Carnegie Mellon University, Stanford, ... The list goes on and on. It should be evident without my saying so directly, but retail traders must be certain they are in the top half -- the half that gets a portion of the profit.

All presented with the best intentions. Please read them as such.

Best regards, Howard

 

[howardbandy]

Well we know for sure Howard isn't making any - he stated he doesn't currently trade. And from the tone he hasn't for a while. Everything is theory and back tested stuff which just contradicts everything he is saying on using the latest buzz to level the playing field in the markets against your competition.

When you make big controversial claims in profiting in the markets, especially as an educator/author, people like results to back them up. All we really got was - "buy the book".

Greetings --

Reread the reasons why traders stop trading. One is they have enough and no longer need to put funds at risk.

Theory, yes. And research. But not backtested. Backtested is necessary, but not sufficient. Way beyond backtested.

Where is the contradiction?

The product I am selling is professional advice. The part you can see is books. If you have not yet read them, please do. If they are not valuable to you, send them back and ignore them.

I offered and sent out hundreds of free copies of the latest book before the official publication date. The current price is as close to free as Amazon will allow me to make it. My profit is at most a few cents per copy, depending on where it is ordered and shipped. Some copies are net negative profit.
https://www.amazon.com/Foundations-Trading-Developing-Profitable-Scientific/dp/0979183863

Or go to the book's webpage and read almost all of it free:
http://www.blueowlpress.com/123-2/foundations-of-trading

Best, Howard

 

[howardbandy]

Howard did more than comment - he stated that all the back-testing my system was based on is flawed. He also added his usual doom-and-gloom spiel.

I'm no longer able to provide posts on my system - I'm potentially entering an arrangement which would apply a non-disclosure agreement.

Minwa is one of the profitable traders (many more lurking and posting).

Hi Trendnomics --

I do not recall evaluating a trading system for you. Please remind me of the details.

What I do say regularly is that backtesting is necessary, but not sufficient. Backtested results always underestimate risk and always overestimate profit. They have no value in estimating future performance.

That's what I say to Everyone about All backtesting. That is what every professional practitioner in the entire modeling and learning community says. Except trading system developers, who, for some reason I do not understand, ignore many of the basic tenets of the scientific process. And too often seem to be proud of it.

Backtesting without independent validation is necessary for us to notice that we might have found a good model. But it has no value beyond that. Some models are guaranteed to be perfect in-sample and give absolutely no guidance about out-of-sample. Others have reasonable agreement between in-sample and out-of-sample providing some criteria have been met. It depends on the model, on the data, and on the technique the practitioner uses.

The decision trees model type used by traditional trading system development platforms are notorious for overfitting in-sample. Very few pass independent rigorous validation. Consequently, most are unsafe to trade. Everyone who has seen a walk forward fail will understand.

Best, Howard

 

[howardbandy]

Howard has an axe to grind.

Greetings --

What is my axe?

Best, Howard

 

[Ves]

I had no idea the first post would be interpreted as "my way, nothing else will work." My apologies. How should that material have been presented?

In my opinion, when dealing in the public sphere no matter what content you present, and how you present this content, different people will always have different interpretations of both the content and your original intentions. Once it has been published you have really lost most of your own control over it - people will do with it as they see fit. You can disclaim it as much as you like, "this doesn't apply to..." "this is correct in cases where..." etc. but that in itself can also be misinterpreted or missed.

I don't think there is really anything wrong with this, it just is. It is human nature. Conflict around ideas and personalities is unavoidable.

Howard, you are not generally rude and I think you remain pretty civil and you certainly don't hide the fact that you are an academic with publications available for purchase.

You have started an interesting conversation, and there have been some fantastic responses for and against.

Cheers

 

[DeepState]

Greetings --

What is my axe?

Best, Howard

Here it is. And it's a beauty.



and also here:

Greetings --

The product I am selling is professional advice. The part you can see is books.

If you are readily identifiable, make posts which bring direct invitation to visit a portal through which you can generate revenue (profit is another thing) and where such activities increase your marketability (eg. Author of twenty books, web site visited by 200k...) then this looks like commercial activity.

If you offered your opinion in a way that is largely anonymous and could not reasonably be seen to benefit anyone other than the free exchange of information, I might not perceive an axe. Perhaps my perception of an axe is a Type II error.

In any case, respect to you.

 

[minwa]

Hi Craft --

I began to thread. With the simple intent of pointing out techniques that some of our competition in trading is using. The ensuing discussion went well for many postings. Every civil question addressed to me, or answerable by me, was answered. Then began both going considerably off-topic, and involving personal attacks. The purpose I had was complete.

I had no idea the first post would be interpreted as "my way, nothing else will work." My apologies. How should that material have been presented?

I am open to additional civil on-topic discussion. What more, related to the first post of the thread, should be discussed?

Best, Howard

As I write in my "Foundations" book, the days of chart reading, long term holding, and simple trading algorithms are over. The business of trading is changing with astonishing speed. It is now about applied mathematics, machine learning, Bayesian statistics. Traders without skills in math, programming, statistical analysis, and scientifically developed trading techniques are at a severe disadvantage. Stephen and his colleagues will "eat the lunch" of unprepared traders.

This is obviously the part from first post most of us had problems with. I've bolded the parts where the wording made it pretty clear to us you are trying to stat facts, not opinion. Facts with nothing backing them, I've already posted 2 indexes (1 is from Fundseeder other from Barclays) of accumulation chart of real world traders where discretionary outperformed.

Most of the fireworks then started when Tech started hammering away that it's gona take over the trading and use it or become obsolete (fossil was the exact world he used).

What some have a problem is - lack of real world results. Sorry but "I have reached my desired financial target and don't know what to do with more money", while commendable in other context, is a pretty lame excuse when you are a trading educator actively talking and not trading. People will be more receptive to someone that actually do, in the live markets - until then, it's all theory. Don't see how trading algos can't be run with minimal maintenance so time is not a problem. Money is obviously not a problem - in a fact it's a good solution for your extra money - trade to enhance education for your listeners/readers. There are so many third party tracking methods this days no one will dispute it's authenticity. It will talk louder than anything anyone can write on this thread.

 

[ChaosHedge]

I've found this thread really interesting and have learnt a lot from many of the people posting.

Not sure if my understanding is correct but its seems that short term discretionary traders have a set of rules regarding their strategy and then use instinct or gut-feel to confirm. Is the instinct part driven by subconscious pattern recognition, or feeling positive based on previous experience? Would a ML algo basically be doing the same thing? Just having some defined rules and matching patterns with historic ones, however complex they may be.

I believe there are many profitable discretionary traders here and it's seems logical that there should be a similar number of quant traders. The time required to become successful discretionary would be enough time to learn ML methods? Are quant traders less willing to share their systems because they can be more easily taken compared to the skills of discretionary trader?

Still learning, keen to hear thoughts. Cheers

 

[minwa]

I've found this thread really interesting and have learnt a lot from many of the people posting.

Not sure if my understanding is correct but its seems that short term discretionary traders have a set of rules regarding their strategy and then use instinct or gut-feel to confirm. Is the instinct part driven by subconscious pattern recognition, or feeling positive based on previous experience? Would a ML algo basically be doing the same thing? Just having some defined rules and matching patterns with historic ones, however complex they may be.

I believe there are many profitable discretionary traders here and it's seems logical that there should be a similar number of quant traders. The time required to become successful discretionary would be enough time to learn ML methods? Are quant traders less willing to share their systems because they can be more easily taken compared to the skills of discretionary trader?

Still learning, keen to hear thoughts. Cheers

Discretionary is quite broad, I think a lot have general rules and others simply use gut. Having defined rules and matching patterns makes it more systematic than discretionary. The discretionary part is the freedom/judgement to take that trade or not/bend rules a bit or not..something the machine cannot do.

Time required to be successful in ANY discipline is different for everybody..probably never for a good amount of people. Yes quant traders system would be more easily replicated if shared, like Howard's book's RSI system - however no one is asking for any systems to be shared..just the results.

 

[tech/a]

Most of the fireworks then started when Tech started hammering away that it's gona take over the trading and use it or become obsolete (fossil was the exact world he used).

If taken in the context it was meant----Not including sound data analysis in my view will mean in the future you will be regarded as a fossil when trading by those who have at their disposal the knowledge and the ability to apply quant methods and testing to their trading arsenal even if your a discretionary trader many questions can be answered with data analysis.(Well answered with more certainty than gut feel).

What some have a problem is - lack of real world results. Sorry but "I have reached my desired financial target and don't know what to do with more money", while commendable in other context, is a pretty lame excuse when you are a trading educator actively talking and not trading. People will be more receptive to someone that actually do, in the live markets - until then, it's all theory. Don't see how trading algos can't be run with minimal maintenance so time is not a problem. Money is obviously not a problem - in a fact it's a good solution for your extra money - trade to enhance education for your listeners/readers. There are so many third party tracking methods this days no one will dispute it's authenticity. It will talk louder than anything anyone can write on this thread.

I'm no different I am one of those.
2 things
(1) Having worked with these types of people for so long I understand that its about study of data. Application is up to the user. 20 different users of the same data analysis using different inputs and or money risk inputs. There are certain things we want to know about the data, profitability is just one.

So a Quant could post up any data set with any number of inputs and show those with both profit and loss.
Just a slight digression and to make a point.

In 2000 I started trading live on Radges site Tech Trader. A very rudimentary average trading method designed by me---a duck.
We traded 30K with another 60k on margin using the BT margin stocks available to trade on margin at the time. Traded it for 7 yrs. Our 30k went to a high of $438k and I closed it all out at around $350K prior to the GFC.

At the time the system was picked to death over the whole period---fine that's what I wanted as I was trading it.
Then I got ---but you traded it in a screaming bull market---yup but I had no idea that it was going to be a bull market when I started.
Then ---but you traded on margin then compounded your profits---urr yes.

The end result was despite protests as loud as a jumbo jet it did well.

To my point.
people can post systems traded/paper traded/ test results only and you'll have a great number of other people who wont be happy with one condition or another.
But I agree results are great to see---but they are the results of the user not you

See I've done it myself!!!