Index | Recent Threads | Who's Online | Search

Posts: 87    Pages: 9    1 2 3 4 5 6 7 8 9 Next
Last Post
New Thread
This topic has been viewed 3824 times and has 86 replies
mgerstein
To Quant Or Not To Quant, That Is The Question

In another thread, the one on optimization, Jin stated as follows (the all-caps and bold are his):

* * *

P123 HAS CLEARLY STOPPED BEING A QUANT SITE. P123 IS A SITE FOR PEOPLE WHO HAVE MATH PHOBIAS. I BELIEVE P123 IS MISSING OUT ON A LARGE MARKET. P123 DOES NOT ATTRACT PEOPLE WHO ALREADY KNOW THE ANSWERS TO THE QUESTIONS YUVAL KEEPS ASKING HERE ON THE FORUM. P123 DOES NOT ATTRACT PEOPLE WHO DO NOT WANT TO WAIT ON US DO GET THE ANSWERS--PEOPLE WHO KNOW THAT EVEN IF WE HAVE THE ANSWERS THE METHODS WILL NOT BE IMPLEMENTED.

P123 could attract new members if they could use basic, proven methods without having to get multiple features approved to be able to use these methods. Features—that in truth—will never be approved.

Finally, QUANTITATIVE HEDGE FUNDS OUTPERFORM HEDGE FUNDS THAT USE FUNDAMENTAL ANALYSIS. There can be no question that the methods work.

For an anecdotal example, WARREN BUFFETT IS NOT TOP DOG! Jim Simons beats him handsomely: see image.

BUFFETT'S RETURNS ARE DECLINING AS ARE THE RETURNS AT P123 I THINK. For Warren Buffett, it is a reality that he admits to frequently. Some of us are in denial.

* * *

Let’s talk about this.

There has been considerable talk within p123 lately about where the company should turn for growth. My position is that p123 is missing a huge opportunity by not speaking to people who just want to pick stocks, amd don’t necessarily want to automate everything they do. I know with 100% certainty that the market is big and that it exists (because I’ve seen it first hand for decades), and that it is not necessarily tied to “performance.” Back when I was at Value Line, I used to say that we were not in the financial services business but rather in the entertainment business. Stock picking is FUN and people love doing it, arguing about it, experimenting with it, etc., which why Wall Street Week was a hit program in the old days and CNBC, Bloomberg TV, Fox Business took over in this era and why Seeking Alpha, which played well to it, was able to reduce Yahoo! Finance, the once dominant web outlet, ti virtual irrelevance. But this is a hard market to expand in because getting attention requires gobs and gobs of outreach, marketing, promotion, etc. and whether p123 will have the stamina to stay with it is unknown at this time.

The quant market as p123 now defines probably has some juice left in the professional area once S&P is a memory after mid-2020 (S&P has done much to damage p123 in this area) and once global gets going. That’s good and could be impactful in a positive way.

But among US individuals, I think the quant market, as p123 serves it, may have peaked.

So the quant question is the nature of the quant opportunity as Jim defines it, a more intense quant market. I doubt p123 can serve hard-core quant pros on the platform. These folks are very demanding and want exactly what they want the way they want it. No “platform” can feasibly serve this much variety of demand. To serve this market, p123 will, I think, have to do so on a custom contractor type basis and some (many?) might even want to contractually bind an outside vendor to confidentiality. And that could, I think, be a strong opportunity for p123.

Custom one-on-one relationships aside, bear in mind that there are other vendors who provide high-end specialty quant capabilities. So its not as if p123 is needed to serve the Jim market. The question is what is the need to have it all on a single-source platform, the costs of meeting the need (which may include costs associated with being able to handle many different kinds of new/alternative/unstructured data sets), what the potential demand is for the general-purpose platform, and what people would pay. The market, I think, will be small since it would require a high degree of quant skills/education. (The more a seller demands of a customer, the smaller the custoner base will be)

My question, therefore, what, exactly, is the business opportunity for p123 in the high-quant area. This is not about capabilities Jim, Steve, Yuval or any other person might want. It’s about the business (profit) prospects for p123 in supplying it.

When it comes to this market segment/niche, I’m not a believer. But if you want to advocate for it, you need, ultimately to talk to Marco; he may follow this thread or if not, if credible points are made I and/or Yuval can be expected to call his attention to it.

Any “sales” process involves overcoming objections, so I’ll share with you some of my concerns and give you an opportunity to overcome them (and I’m open minded, if you persuade, I’ll change my mind).

1) To what extent are hi-quant (my label) customers driven by the desire to simply do it versus the desire to perform (however you define it)? This concerns me deeply because I fear much success is attributable, not to quant skill, but to the good fortune of having been able to successfully exploit as 35+ year period of loos (often wildly loose) money creation and pro-investor political tendencies both of which are under serious threat. I’m always motivated to serve a customer that cherishes the joy of experimenting and use of use of quant data/tools to expand understanding. But selling to someone who is going to pissed if they can’t earn at least 25%-30% a year, oh no. I think that would be commercial suicide.

2) How effective are hi-quant customers in generating performance — attributable to use of tools, etc., as opposed to unique genius that leads to success in any endeavor? This for me, is a major concern and I’d want to see much more evidence than the brief and flawed anecdotal table Jim reproduced in the other thread. I suspect that VT>LO*NAMC (Volume of Talk > Level of Outcomes Not Attributable to Market Conditions). In other words, does this stuff REALLY work. We’ve seen over many years how p123, as it is, can make people better than they otherwise would be. I am yet to see where the fancy stuff can make a large number of investors better. Show me I’m wrong — or show me that desire to do high quant is attributable to the process of doing it, apart from performance.

In business, nobody can be all things to all customers and I don’t see a cost-benefit case for p123 straying into high quant. My fear is that p123 would spend a ton of money only to wind up with few new customers most, if not all of them would, put forth a never-ending series of demands for more and more things (most with substantial cost implications and questionable at best incremental revenue prospects). If p123 is to become the kind of hi-quant platform some dream of, show us its a bona fide business. Show me I’m wrong to believe, as I now do, that it would just cause p123 to wind up perennially digging into a rabbit hole the bottom of which keeps sinking deeper and deeper and deeper.

Seriously, I’ll advocate for it if you show me its a business, not just a feature wish list or an anecdote. As Michael Bloomberg liked to say when he was Mayor of NYC, “In God We Trust. All Others Bring Data.”

Marc Gerstein
Director of Research, Chaikin Analytics
Blogs: https://actiquant.com, https://portfoliowise.com/portfoliowise-blog/ , https://www.chaikinanalytics.com/blog/
Twitter: @MHGerstein
I predict the future, as soon as it becomes the past

Dec 3, 2019 6:30:21 PM       
Edit 2 times, last edit by mgerstein at Dec 3, 2019 6:31:21 PM
Jrinne
Re: To Quant Or Not To Quant, That Is The Question

I doubt p123 can serve hard-core quant pros on the platform. These folks are very demanding and want exactly what they want the way they want it.


Are you kidding? The aggressive quant is the one that looks at your shoes instead of their own when they make a request. Case in point: the only real quant we have at P123 now is Supirate whom we never hear from. There were a few others, but they are gone from the forum at least. Primus, CyberJo, others. Primus did post a lot but he was very kind in his posts. We never heard much from CyberJo.

Granted, I want to be a quant but calling me that would definitely be generous. You could count 5 quants, at most, here on the site. I hesitate to name them.

Quants like CyberJo tended to make a few quant comments (like this thread with CyberJo) and get comments back like this:
I'm wondering if any of you can explain this to someone whose knowledge of statistics doesn't go beyond what's available on Excel. I've read this thread several times and I can't figure out what bootstrapping is. Or what an equity curve is. Or what p-value and t-test mean. Or what an I.I.D. is. I take it R means correlation (as in r-squared)? When I look these up, I get just as confused. What does randomness have to do with P123 results? Do I need to take a stats course, or can someone explain in simple terms what you're doing? Are you manipulating screens and/or simulations in Excel (which is what I do)? Does this have anything to do with alpha and standard deviation or is it something altogether different? And isn't applying statistics to technical analysis like applying Newton's laws of motion to astrology?

Yuval


"Newton’s laws of motion to astrology?" Certainly does not qualify as appreciation of quants in the forum. Definitely one would get the idea that, at best, your requests will not be understood.

CyberJo quietly went away after this. I will admit that things have improved a little bit. But quants understandably do not think P123 does modern techniques or cares much about them.

Like it or not, P123 is a quant site. Yet you have very few real quants here now. The quant market may have peaked (I doubt this) but you are attracting almost none of them. There are plenty that could be attracted no matter what methods become popular in the future.

You do not attract quants because you are not using any modern techniques. It is all manual optimization. Also there are problems with the method such as the linearity assumption (the ranking system is a linear equation that is normalized if you want to debate this). This and other basic things that a true quant would immediately recognize are reasons true quants are not here. You know, the ones who have not signed up because they understand this.

If you think making no advances is good for some reason then do not make any advances. Personally, I do not think this is something that is noble or good in any way. I cannot see how this is a good business decision either.

If you have a plan to attract Fundamental Analysts you should definitely implement that plan. Why have you waited? Not because of me I hope.

I do think you should see how well this plan works out before you make of point of telling quants you are not going to be doing anything to keep them at P123. But you are the one who understand businesses.

I think you could attract a lot of quants if you actually made any effort. Good luck with whatever you decide.

BTW, I was minding my own business when Yuval asked about techniques that did not require optimization. Sorry, if the simple techniques I presented caused any problems.

-Jim

From time to time you will encounter Luddites, who are beyond redemption.
--de Prado, Marcos López on the topic of machine learning for financial applications

Dec 3, 2019 8:17:13 PM       
Edit 18 times, last edit by Jrinne at Dec 3, 2019 9:13:44 PM
yuvaltaylor
Re: To Quant Or Not To Quant, That Is The Question

Jim -

While I agree with 99% of what Marc wrote here, I would certainly be open to some very PRACTICAL suggestions of concrete quant tools that could be integrated into our current functionality. These suggestions should include some kind of specification as to how the tool would be applied. "Integrate Python" doesn't tell me what's involved or how it would be used or how one could accomplish it. These suggestions should be comprehensible to wannabe quants like me. There are dozens of things I'm working on to improve Portfolio123 and its capabilities because I really care about its users. But they're all in the nature of "add a universe parameter to the Rating command" or "enable buy-driven simulations" or "add buy and watchlist buttons to the screener" or "enable hedging in ETF simulations." We're willing to entertain larger ideas too, like "in addition to designer models, let's have stock-picker models" or "allow users to import their own factors with an API." So if you want to make a concrete suggestion as to how P123 could implement random forests, I'd be happy to listen. But you would need to walk me through the process so that I understood exactly how it could be done. And only you could do that, since you're the only one here who has actually used random forests to optimize ranking systems.

Yuval Taylor
Product Manager, Portfolio123
invest(igations)
Any opinions or recommendations in this message are not opinions or recommendations of Portfolio123 Securities LLC.

Dec 3, 2019 9:18:11 PM       
Jrinne
Re: To Quant Or Not To Quant, That Is The Question

Jim -

While I agree with 99% of what Marc wrote here, I would certainly be open to some very PRACTICAL suggestions of concrete quant tools that could be integrated into our current functionality. These suggestions should include some kind of specification as to how the tool would be applied. "Integrate Python" doesn't tell me what's involved or how it would be used or how one could accomplish it. These suggestions should be comprehensible to wannabe quants like me. There are dozens of things I'm working on to improve Portfolio123 and its capabilities because I really care about its users. But they're all in the nature of "add a universe parameter to the Rating command" or "enable buy-driven simulations" or "add buy and watchlist buttons to the screener" or "enable hedging in ETF simulations." We're willing to entertain larger ideas too, like "in addition to designer models, let's have stock-picker models" or "allow users to import their own factors with an API." So if you want to make a concrete suggestion as to how P123 could implement random forests, I'd be happy to listen. But you would need to walk me through the process so that I understood exactly how it could be done. And only you could do that, since you're the only one here who has actually used random forests to optimize ranking systems.

I have posted about this before.

I think people will agree I post a lot.

Perhaps, it is not possible to explain this to you. Certainly it is questionable as to how much more time I should spend here.

I would be happy to talk to a programmer or Marco at this point if P123 is serious. I would even be happy to travel to Chicago and talk with Marco and the programmers.

I do think P123 could attract a lot of quants and that it would be a good business decision. Right now P123 has next to none.

If you are not serious that is fine too.

-Jim

From time to time you will encounter Luddites, who are beyond redemption.
--de Prado, Marcos López on the topic of machine learning for financial applications

Dec 3, 2019 9:32:58 PM       
Edit 5 times, last edit by Jrinne at Dec 3, 2019 9:58:49 PM
yuvaltaylor
Re: To Quant Or Not To Quant, That Is The Question


I have posted about this before.

I think people will agree I post a lot.

Perhaps, it is not possible to explain this to you.

If you are not serious that is fine too.

-Jim


I'm absolutely serious. Dead serious. I want to know how to use random forests. I want to understand kernel regression. I want to know if P123 can help its users with PCA. I have read your posts, many times, and I have seen a lot of interesting ideas in them. Because of you, I have read about random forests on a number of different websites. Because of you, I came across the term "bootstrap aggregating." But I couldn't understand how any of this could work with what P123 offers. I want to learn more. I want to believe that there are capabilities here. But you have to keep in mind that I'm not very sophisticated in the use of machine learning tools. If implementing all this in P123 is something you want done, you have to explain to me, the product manager, how to do so. You're the only user I know who has used all these sophisticated (or, as you put it, "simple") tools. And I don't understand why you're advocating these tools but won't explain how they can be used here. I really don't mean to be "rude." I'm just asking for help.

- Yuval

Yuval Taylor
Product Manager, Portfolio123
invest(igations)
Any opinions or recommendations in this message are not opinions or recommendations of Portfolio123 Securities LLC.

Dec 3, 2019 10:01:03 PM       
Jrinne
Re: To Quant Or Not To Quant, That Is The Question


I have posted about this before.

I think people will agree I post a lot.

Perhaps, it is not possible to explain this to you.

If you are not serious that is fine too.

-Jim


I'm absolutely serious. Dead serious. I want to know how to use random forests. I want to understand kernel regression. I want to know if P123 can help its users with PCA. I have read your posts, many times, and I have seen a lot of interesting ideas in them. Because of you, I have read about random forests on a number of different websites. Because of you, I came across the term "bootstrap aggregating." But I couldn't understand how any of this could work with what P123 offers. I want to learn more. I want to believe that there are capabilities here. But you have to keep in mind that I'm not very sophisticated in the use of machine learning tools. If implementing all this in P123 is something you want done, you have to explain to me, the product manager, how to do so. You're the only user I know who has used all these sophisticated (or, as you put it, "simple") tools. And I don't understand why you're advocating these tools but won't explain how they can be used here. I really don't mean to be "rude." I'm just asking for help.

- Yuval

I would be happy to talk to Marco and Marc. I have tried.

Actually for a few years I have been trying. This was your response when I first opened a thread about bootstrapping in 2017:

I'm wondering if any of you can explain this to someone whose knowledge of statistics doesn't go beyond what's available on Excel. I've read this thread several times and I can't figure out what bootstrapping is. Or what an equity curve is. Or what p-value and t-test mean. Or what an I.I.D. is. I take it R means correlation (as in r-squared)? When I look these up, I get just as confused. What does randomness have to do with P123 results? Do I need to take a stats course, or can someone explain in simple terms what you're doing? Are you manipulating screens and/or simulations in Excel (which is what I do)? Does this have anything to do with alpha and standard deviation or is it something altogether different? And isn't applying statistics to technical analysis like applying Newton's laws of motion to astrology?

-Yuval


But now you propose bootstrapping in the forum. Understand it well. Understand a lot, actually, when you propose it. E.G., unsupervised learning (K-means published at SeekingAlpha), Quantile regression, Omega etc.

Anyway, you will not get it now. That is just a fact. We can wait 'till when you propose Random Forests over at Seeking Alpha in a little bit, however.

Even after we read about Random Forests at SeekingAlpha we still will not see it here at P123. That also is just a fact.

-Jim

From time to time you will encounter Luddites, who are beyond redemption.
--de Prado, Marcos López on the topic of machine learning for financial applications

Dec 3, 2019 10:05:48 PM       
Edit 13 times, last edit by Jrinne at Dec 3, 2019 10:52:41 PM
SnortingElk
Re: To Quant Or Not To Quant, That Is The Question

I’m not a P123 power user, but I was involved in the development of several software products for data processing (billions of transactions a year) that were used by a range of users between “analyst” and “C++ Coder”, and I think some of the frameworks we implemented may apply:

1: Custom Data Series: I’ve tried using this feature in P123 and was unable to do what I wanted. Custom Data Series was a basic feature needed at the “analyst” level on the products I managed. Even StockCharts.com has this feature. I believe this is an “analyst” level feature but would also be used by developers.

2: Extended Custom Formulas: P123 has this feature, but it is severely limited in my opinion because of field length restrictions (I had to break my single formula into 30 separate formulas). These should also have the ability to access the custom data above for reads and possibly writes. I believe this is an “analyst” level feature but would also be used by developers.

3: Simple Automation: I think sometimes it’s useful for “analysts” to automate the export of screen results.

4: Read Webservice API: Anything that the user sees on the screen should theoretically be accessible through a web service call. For example, it should be possible to access the results of a simulation, or a screen, by making a web service call and getting back JSON data. I believe this is a developer level feature, which would then allow developers to write custom code in the language of their choice on their own remote platform.

5: Write Webservice API: It should be possible to create a new simulation, or ranking system, etc. by making web service calls. I believe this is a developer level feature, which would then allow developers to write custom code in the language of their choice on their own remote platform.

6: Embedded Authoring and Runtime Engine: In one of my old products, we also integrated an open-source Java Scripting engine, so that analysts could write code within the product. As our runtime engine processed data, there were callout points to the custom code. One difference between this and the existing P123 “custom functions” is that users could define user variables within different scopes. So, for example, start of simulation, call function x, which might initialize some custom variables that could be accessed by other custom functions. Our software ran locally, so custom code could also make callouts to the local system, databases, etc. How much additional functionality is needed is not clear.

In the products I worked on, we also had a custom reporting framework, an audit framework, a validation framework, a mapping framework, and a private label framework, but these are I believe less relevant.

I believe #1 to #3 above are basic features P123 should have. A limited set of APIs (#4 and #5) seem like they wouldn’t be too hard. For example, if the custom data series above had a Write Webservice API, a developer could write some Java/VBA/... code on their own machine to extra data from source A and insert it into a P123 custom data series, which custom functions could then access when a simulation was started through a WebService Call, and the simulation results processed through yet another web service call. I believe #6 above could also be reasonably done but would have to be very well defined through use cases.

My expectation would be that anything that Jim or other Quants wanted to do would then be implementable through their own custom code using the standard frameworks above, but that the frameworks would have a lot of other uses as well. But I'm not a Quant, so perhaps I'm offbase.

Anyway, I think P123 is a great product.

Dec 3, 2019 11:33:22 PM       
judgetrade
Re: To Quant Or Not To Quant, That Is The Question

I am with Marc Gerstein here, 100%, 100%!

Its very well possible to make >30% with p123 with a smaller portfolio. With ETF Modells you can capture a lot of liquidity and
Georg has shown that good performance can be made here too!

Sorry to be blunt: P123 has everything to be successfull. So ask your self, is it the platform or is it yourself?
There will be always something missing, if you blame it to the platform.

The most important factor of a platform is clean PIT Data and I would say P123 does a fine job here!

Of cause you can improve it (in both directions: stock picking and quant direction) and the direction into a different data provider
and internationalisation is the right one bc it scales the business on existing features that makes perfect economical sense!

I was in Software Product Management for 25 Years, the question is always the same: "Just this one feature and its good.
And then the next feature, and then the next feature." What the customer is missing is to implement and use the
value of the software that is in front of them. You can reach with 20% of the work, 80% of the success for 90% of the customer base and
every software provider is in the 20/80/90 game, everything else makes no economical sense.

If you want more, then raise the capital and give it to the p123 Team to implement your stuff, then you got some skin in the game!
Until then you demand something that is easy to demand bc. your risk is 0.

I made north of 600k with P123, I scaled up a small portfolio from 50k to 850k (with some regular cash attributions to the port) and I will scale it up to 3 Million within the next 5 Years, that is life changing and I do not know any other platform where I could have done this at least not for the
price p123 is offered!

Sorry, to be blunt but I am big fan of p123 and I got everything with p123 to make very, very good money, thats more then enough for me!

Regards

Andreas

Dec 4, 2019 12:37:23 AM       
SpacemanJones
Re: To Quant Or Not To Quant, That Is The Question

fwiw, p123 is great for me. The data and backtesting is feature #1 that I find value in and is the key selling point for investors like me. The tools for building models are excellent for my use. But to be fair, I'm just a small retail investor and I know my needs are simple compared to what others may want. p123's offerings are probably at the top of my price point, so I'm really more on the "please keep my subscription costs within budget" side of things than the "add more expensive features" side of things.

For me, p123 is far better than my efforts working with quantopian code several years back, and even with considerable programming study and access to a data provider I doubt I could build anything as sophisticated, reliable, or as outright useful as what I get here at p123 straight out of the box. As to investing systems, it took me probably a year or year and a half to work through iterations and get good models together. And I _think_ I have models that will work going forward and suit my investing personality - but the past year or so has not seen results that I would expect over longer term - so I'm not sure what I have will work prospectively. I do think it should work and I've put my savings behind it, but at this point I'm just trying to give the systems time to prove themselves OOS. This past year or year and a half have been tough for many quant approaches, so I suspect I'm not alone with concern.

Functionally, p123 does what I value. I particularly value the ability to backtest and research/build models as p123 currently enables. For me it is, like Marc suggests, partially entertainment. I really do like trying to figure out good investing approaches. After reading so many investment books, it's nice to actually be able to study the data more systematically and scientifically - at least during a 15 yr window of time we can examine. But given the amount of time I've put into this and financial needs, I also hope I can get good forward returns out of it also. So while I admit the entertainment aspects are appealing, future investment success is also important, would make a meaningful financial impact, and make a significant positive impact on my lifestyle.

From a business standpoint I suspect expanding p123 functionality to international data/markets is a good move - so I hope that works out well for p123. I don't know alot about investing directly overseas, but look forward to learning. Given the valuation spread between US vs. many intl markets the addition seems timely.

Dec 4, 2019 2:35:32 AM       
Jrinne
Re: To Quant Or Not To Quant, That Is The Question

but I want to use some sort of objective/mechanical method to combine them into a ranking system.

-Yuval


My apologies.

My mistake that there might be a market for this. Don’t know where I could have gotten the idea.

There were a few other responses in the thread. None outlining any "mechanical" methods. I guess I am not going to apologize for answering the actual question to the best of my ability.

If you ask for a "mechanical" method one might expect a machine learning answer, I would think.

These folks are very demanding….
-Marc


But it is good to know that P123 is so financial sound that it can let quants know what P123 thinks of them and that they have no intention of catering to their demands. For the record none of what I said could be classified as a demand, I think. In fact, Yuval says he wants me to go into greater detail. Seems like Yuval can be a bit of a quant himself—depending on the situation. Guess Marc had not made it clear to everyone how he felt yet.

BTW, no matter what one thinks of my situation, I do not hear any true quants making any requests. None at all since Primus went dark. But stereotyping people seems to be okay in 2019. Even if you are: Director of Research, Chaikin Analytics which seems kind of quant-like to me. I am sure there are sound fundamental reasons for the the Technical Indicators in this case. Stereotypes are okay and some get a pass. Yep. All good in 2019.

I am sincerely glad that P123 is doing so well because I did not know how many Financial Analysts were going to want to purchase the Designer Models in the future as it does seem kind of mechanical to me. I would have though they would want to study the companies and screen them on their own. Plus, the Designer Models really are not doing too well. Then again, maybe only a quant would notice this fact. Maybe we are good.

I would like to see P123 stay in business. Again, sorry for my suggestion in that regard. Glad I had no need to worry.

Best,

-Jim

From time to time you will encounter Luddites, who are beyond redemption.
--de Prado, Marcos López on the topic of machine learning for financial applications

Dec 4, 2019 3:34:47 AM       
Edit 32 times, last edit by Jrinne at Dec 4, 2019 6:22:34 AM
Posts: 87    Pages: 9    1 2 3 4 5 6 7 8 9 Next
 Last Post