Exposing stock ID numbers was so that Yuval could do bootstrapping as Yuval says here:
Yuval does bootstrapping and can now upload that data back into P123 for more manipulation there and to look at the equity curve.
Fine.
What I do not understand is why he trashes Sharpe Ratio and the few methods that we can use on P123.
HE ALSO STOPS US FROM GETTING THINGS LIKE THE INFORMATION RATIO AND THINGS LIKE KURTIS HEMMERLING’s REQUEST REGARDING THE TOP QUINTILE MINUS THE BOTTOM QUINTILE. AT ONE POINT MARC INTERVENED AND SAID HE WOULD PROVIDE THIS.
[b]P123 is not supplying what pros are requesting. [/b] Maybe Yuval is too busy with his bootstrapping to do what Kurtis has requested and Marc has promised.
Yuval also trashes machine learning, only occasionally admitting some of the things he is using are okay with all other methods being “flawed” which is his usual adjective. Yuval will trash machine learning too if it is your idea.
Marco and Marc,
Yuval should be allowed to use data as he wishes.
But P123 is not just for Yuval. We do pay for a service.
[b]And you have the worst marketing strategy one could imagine which seems to be: [/b]
Have the product manager trash the few statistics that are available to P123 members like the Sharpe Ratio
Trash any machine learning ideas while he is at it.
Provide no new statistics or mainstream ideas (like Kurits requested).
Develop machine learning techniques that are not generally useful to members at the cost of some more useful methods that professionals are requesting.
Lecture us about Theil-Sen (an esoteric little-used machine learning technique) ad nauseam as an excuse to not provide what we are requesting.
But make no mistake, P123 is about machine learning now. Not that you will be able to use it. The P123 developments in that regard are not generally usable by the average member.
This is not remotely the case. Exposing IDs was very easy to do, and a lot of users wanted to be able to create randomly divided universes. Providing information ratios is a bit harder but it’ll get done quite soon. The custom factor and data series tools are major improvements that enable our users to import their own data. We are planning to provide ETF hedging soon, not because I use it but because our users want it. I am constantly listening to what the pros want and putting that stuff on our to-do list. Because the list of improvements is so long and because of the FactSet transition we have to prioritize our improvements, in part by how easily and quickly they can get done.
But Marc got it exactly right: We are treated like children when we ask for anything. Marc could not have come up with a better description of how it is at P123.
I still, for some reason, believe that is not good marketing to consider—at one time—forcing randomization into the data, then at a later date likening us to children at the kiddie table for the same idea. And just for the option of randomization at that.
I think Yuval has stopped and is a force for good now.
But I did not like having to listen about Theil-Sen (and more) whether I was recommending the information ratio or bootstrapping. Yuval is a fan of bootstrapping now. But it is a tool he related to astrology, somehow, when it was first presented in the forum.
I stand by my claim that P123 has had the worst marketing in the world. Ready fo a case study in business school about self-inflicted injury.
I will stop and indeed work with Yuval (if he ever wants my advice) because I think he gets it. How could he not.
And Marc has always been a force for good despite the poor marketing (I mean that). After all he is sending the Information Ratio to the kiddie table. And I do get the occasional pat on the head from him.
ok, lets just agree to disagree that you are not happy about p123 and I am very happy about p123.
P123 (and the comunity of P123) delivered > 750% (and the hint from P123 “Ideas first, Math second” was very valuable to me!!!) since 2011 to me, I am 100% fine with this, the platform is great, international stocks will be added, I could not be happier and I am very, very thankfull to the whole team and the whole comunity!
If you are not happy with it, why not try something else and keep us posted?
P123 could do quite a bit more: if it survives the terrible marketing.
I am actually very happy with what P123 provides. But I do not appreciate being treat like a child “at the kiddie table” even when I have the most mainstream of ideas like the information ratio. Or when I post about bootstrapping which Yuval has now adopted in spades. He was not too happy with the idea when I first posted about it, to say the least. There is no more mainstream idea within the field of statistical learning than bootstrapping.
Marc got that exactly right, IMHO. We will be treated like children even for the most mainstream of ideas from finance or for the most accepted ideas in statistical learning. It does not matter.
Maybe the forum isn’t for people with their own personal ideas of how to invest (including me). But I do not think you are the first to notice. Many have noticed and just gone away.
Andreas. Maybe the forum is just for people who want to market something with overfit, cherry-picked backtests. BTW, good luck with that. Let us know how your new career works out.
Most people on the forum now have a financial interest that seems to coincide with the ideas presented in their posts.
Jeff (the person to whom Marc refers to in his post) was just talking about a good idea. I good idea that could be expanded upon. The term for this is regularization and should be expanded upon at some point in the future to prevent overfitting. Walter uses it already.
Jeff, do you have a financial interest or where you just asking a question of the P123 staff? Not even a feature request or suggestion, I think.
I was simply asking the question about noise injection. If Portfolio123 doesn’t support it at this time that’s all I needed to know. I understand this is a very complicated but also capable platform. Are there more tools we could use - absolutely, but we can’t get too hung up on those. I think we have to be very measured in our expectations of what can be implemented over time as Yuval stated. As a software writer myself I am sympathetic to to how long the process takes to figure out how to implement, actually implement, and validate software changes to ensure they work and don’t create adverse affects elsewhere. Platforms like Portfolio123 take years to build up and can’t just be rebuilt on a whim.
As for Portfolio123 I think it would be wise to keep gathering up these ideas. Absolutely grab the low hanging fruit. However there may come a time when confronting the hard changes that make sense and improve the modeling process make sense to implement as a block improvement. I suspect a lot of these other features that are being asked for would fall under that and have a consistent theme.
Hi Dan, do not have a link, its my interpretation of Marcs way on how to come up with a trading strat.
I talked to other systematic traders and listen to a lot of podcasts. The ones that say “ideas first, math second” are usually
in the business very long.
Wayne Himelsein is a good example.
Though there others I talked to, that use brute math and computational force first and then use (non OLS) robustness tests to come up with stuff that works.
Not for me, I rather stick to stuff that can be explained by an idea. Even if the idea is weak (e.g. emotional (hard to buy a 52 Week high → momentum) and structural factors (big guys can not invest in small caps) and gives not a 100% proof.
But that is kind of the thing, I do not believe in 100% things when it comes to complex systems (e.g. non-linear, cause and effects constantly changing).
ggee, maybe get a thicker skin and live with the fact that other feature requests are higher prioritised as yours?
I am fine with the example of Marc that we start as kids in the markets and then grow up. Marc put a ton of work in
his courses and they helped me a lot, clear words from Marc helped me too!
This is a good board, so lets not get into the trolling game.
Marc’s contributions are greatly appreciated and no doubt form a bedrock but it seems as if what he teaches is and shall be the last word. Maybe it is, but that’s not how a productive exchange of ideas works. In my employment those who are new often while not experienced come in with new ideas and perspectives. Some of them unworkable and others innovative. The point being we don’t improve without new ideas.
Maybe Marc is just frustrated. I think Jim would be wise to limit his complaints on these boards. Suggest features and let them be. Like has been said we can go other avenues if the platform is not serving our needs. With import and export features we can potential do more processing outside of the system as well.
I was going to code up my own system but then I found InvestorsEdge then moved to here. These platforms are very powerful and would take a lot of work to code up. If I had the time maybe I would code up my own system but I have a job and kid soon to be kids so I don’t have the time to invest.
I have to agree that this forum is not the place to bicker, blame or troll. Its a place to exchange wisdom, ideas, systems, etc.
I have to imagine that someone new looking to subscribe, or simply looking for clues how this platform works, will most certainly be turned off by some of the poorly written back and forth that has gone on in recent weeks. If you want to help market the platform, then your best effort is to make this forum more valuable, and not a soapbox for your personal ambitions.
I’ve been on here 12 years and have witnessed first hand how much P123 has grown exponentially in its capabilities. I do not expect to be able to have the capability to write code, or train machines, etc. This is a screening & ranking tool first, everything else is simply a perk. I have found a multitude of ways to do some very advanced work on here within the present parameters.
There are new investors just discovering the markets everyday. They want to learn, and screening is an obvious first step. If they start to screen, then they will learn to rank. These are the future subscribers that will help keep P123 thriving. We don’t want to scare them off because of infighting, right?
So, as a community, why don’t we pledge to tone down the rhetoric and get back to helping each other thrive on here.
If the bickering keeps up, I’ll be submitting a “feature request” that will limit one particularly noisy subscriber to only being able to make one post a week.
That noisy subscriber talks a lot about how P123’s marketing could be improved. The best improvement I can think of would be for him to limit himself to one (1), yes just one, post a week for the next month (till February 9). It would be nice if he could demonstrate the self control to do this self limiting for the sake of the P123 community.
I am sure there is a lot that Portfolio123 can learn from financial platforms like Reuters Eikon/Bloomberg in terms of marketing/customer services. (In fact, both platforms are competing with each other for market share.)
My personal experience when dealing with their staffs is they are always helpful and never argues with you or question you (or lecture you on OLS regression for instance). Maybe because most of their users are working in investment banks/hedge funds, they never think they know more than their users about the financial markets.
I understand it is a bit different for Porfolio123 since the majority of Portfolio123 members are not full time investment professionals. Therfore, I think it is a good idea for people like Marc (with years of experience in the market) to help bring users up to speed. However, that doesnt mean you should be limited to what you can ask and when you can make suggestions. Since you are a paying customer on Portfolio123 and as long as you have a point, I think you should continue to express you views (maybe tone down your language a bit) as there is no gag order here.
I will say I am glad you are posting and that your are willing to stand up for the mainstream financial statistics and methods you use. I probably have it wrong (or less than perfect) but you got it exactly right, IMHO.
Note that my liking the statistics you use does not mean that I do not think most can be useful. I only advocate for expanding the options. I like bootstrapping and Theil-Sen also. As well as the regularization (adding randomization) that Jeff was asking about. I have never said anything different.
Anyway, including your paper, RT RIEF data and insight into hedging (reducing variance being part of that discussion) the best discussion on P123 in years. Hands down.
I will be happy to do so myself if Jim will too. I promise not to criticize anything Jim says in the public forums anymore if he’ll promise to do the same to me. We can e-mail each other. My e-mail is yuval@portfolio123.com.