I’m writing to gauge interest in a feature we’re considering offering. You would be able to import your own data into Portfolio123 and use it in ranking systems, screens, and simulations. You would provide us with a file with the following information: tickers, dates, and values. We would integrate this into the rest of our data and you could use it as you wish.
This might be useful if, for instance, you have access to data about executive compensation (data we currently don’t offer) and want to see if using that data will provide you with an edge. Or, for instance, if you have access to the past rankings of a non-Portfolio123 system (e.g. Zacks ranks or ValueLine ranks) and want to integrate those ranks into your investing system.
If this interests you, would you be willing to pay extra for this service? Or would you only use it if it were offered as part of your existing subscription (for a certain number of resource units)? What kind of data might you be interested in importing?
Great question. This is not something I want to use now but I would love to have it as an option—in any form.
Great question because I have, in the past, imported Zack’s Rank (one or two) as an In-List and done some refinements on those stocks. Obviously, you are considering something even better than this.
But Zacks does not seem to have much more information useful data than what P123 already offers.
I do not have any new projects that would require importing data NOW. But again this is a great idea that I might want to use in the future.
You would be able to import data for backtesting and adding current values for rebalances. Initially it will be through specific pages on the website much like Custom Formulas. Eventually you will be able to manage your data via APIs.
I would also be interested in this. For example, I could provide the weekly Business Cycle Index series for recession forecasting. It is currently available free for downloading here: https://imarketsignals.com/bci/ and it could easily be integrated into P123.
Also, as previously requested many times we need more FRED data, the Shiller CAPE ratio is one of them I would like to use.
Finally. (Yes. This is a no-brainer way to attract and retain customers.)
An additional major leap would be offering an API towrite custom functions. What we define today as custom formulas only accept predefined values. I’d like to see actual parameters that can be explicitly passed through custom functions.
I know y’all are already using JavaScript/JSON. Maybe you could offer an API through Handlebars. I’ve had a lot of success with this tool at work.
Awesome! I love it. I have lots of macroeconomic data that I use (over half from FRED but the rest scattered). I wrote an application to parse the data in a format that I can use for backtesting etc.
But if I could paste it into P123 it would be great!!! Not only save me a lot of time but also make things much more accurate and I could blend fundamental P123 screens with macroeconomics.
What I, however, would be willing to pay for and like to have is automatically EXPORTING (user-generated) data such as holdings, trades, ranks, etc. to an FTP server. This should definitely be restricted to user-generated data to avoid conflicts with data providers, and should require at least a research license.
PLUS
I would be paying extra for an automated daily rebalancing of ports and books AND exporting this data daily to be used elsewhere
P123, could you please quickly comment on this one as well?
First question that I have… Is the data private or available to other users?
Second question … If I have unique data that I have gathered and processed myself, is there any possibility of offering my datasets to the P123 community in return for a subscription fee. P123 would obviously take a cut.
I currently have (in prototype) price data series for the 11 GICS sectors as they are defined now but I have extended the series back to 2001. I will soon have aggregate fundamentals associated with each of these sector-based price series.
NOTE: The datasets were generated using price and fundamental data external to P123. (No P123 Intellectual Property is involved with the generation of these series.) Stocks were sorted into the correct sector bin by hand, which required more than six months worth of work.
The price series can act as proxies for the SPDR ETFs with the obvious benefit that they can be backtested from 2001. The backtests can be supported by the aggregate sector fundamentals that I have created. This is an obvious advantage over what we have now, as many of the sector definitions have changed over the years and sector ETFs reflect the changes so that any backtests are suspect.
I view this as an absolutely vital step - particularly for researchers. We need to be able to integrate other data-sets. Even better (and of economical advantage to P123) would be to get trial data sets for all the major and minor providers in addition to importing our own. This should be free as we are just backtesting at this point. Anyone that wants the actual live data will incur a commission paid to P123. At this point, the data can either be imported or, in time, P123 can integrate the fees with the subscription.
Having access to alternative data sets and integrating them in P123 is vital for researchers. I am hitting a wall as a consultant that people want certain features built in that I don’t have access to. I use a ton of analytics at WorldQuant - lots of them aren’t very good - but still very important when making models.
I get the impression that for a while this will be mostly custom series or time series data. All good. And probably I am showing my lack of programming skills by even posting this.
Again, this is all good and I hope to improve my time series skills and use this at some point. When I do I would be interested in trying deep learning (or ARIMA)—not just be limited to the present tools at P123. In the near future the stuff I would want to use (and manipulate) would be MORE like a cross section. An example, would be the Zacks data.
I remain very much the student and I am actively working on learning more about the time series aspect of what we are doing here.
Anyway, those more familiar with programming have already probably thought of all of this and I appreciate the continued thoughtful improvements by P123.
How about working out a deal with Intrinio, Quandl and others, were you would offer their DBs for a percentage of the subscription price. The cut would pay for the integration/support of the new data into P123.
The initial post was about importing your own stock data for ranking , screening , etc. We have already implemented this, it’s just lacking a front end. However, importing factors for each stock might be a bit much for most people, and better suited for researchers. You’d have to generate the values for every stock in your universe for every period. Nonetheless we’re getting the front end ready and should launch soon.
What most of you seem to want is simply to import your own time series, like MACRO data, market timing signal, etc. We’ll leverage the current “Custom Series” to allow for also adding “Data Series” ( date,value pairs ) that can be used using the GetSeries() function. This will launch soon as well.
As far offering your datasets to others that would be on a subsequent release.
To import directly from quandl and Intrinio we probably just need to add a way to use your credentials to automatically import data into your P123 account. Naturally this data would be for you only, not shareable.
Exporting data to R , RStudio, etc, gets tricky. You’d need an S&P license.
As far as what our APIs will be able to do, that is to be determined. Naturally they will allow you to import data. They will also allow you to download holdings, run screens, etc. For users with S&P license they will allow you to download factors. We have not started the API project but it’s on the agenda for this year.
Thanks for your feedback.
PS. We added the import feature to test the ranks generated by an AI system that learns how to weight factors . It’s looking promising and has a good shot of becoming another tool in the arsenal.
It would be easy for P123 to develop an API to fetch Quandl data without having agreements that are unlikely ever to happen.
Quandl users have a unique token for accessing the Quandl data so that Quandl can identify the user and verify access privileges. From there the act of downloading data is pretty straightforward.
The concept is … P123 downloads Quandl data using the user’s token at the time the data is required by the user. i.e. when a sim is run or port is updated. The user pays Quandl for the data subscription. The user pays P123 for the ability to use third-party data. Everyone is happy. P123 doesn’t pay Quandl for a lot of data that isn’t used. Sims would take a lot longer to run but worth it for the extra functionality.
The only thing that I would want to see out of this is that the external data series be capable of being designated as the trading instrument. In other words, I would want to be able to test my sector series as if they were ETFs or stocks.
Marco, this is all very nice.
But I am still waiting to have a simple hedge module added to ETF sims, which you said many years ago was no big deal.
So why don’t simple things get done first, like incorporating FRED into the live data base for example…
Marco - I just saw your last post. It sounds awesome. The only thing is I would like to be able to use custom data as if it were a stock or ETF to be backtested and made into a port. This is probably beyond the scope of what you are thinking. But you have one heck of a platform and I don’t see why you should limit it’s capabilities to stock / ETF trading systems. It would be nice for users to be able to predict economic indicators using the same platform or test the proxies for sector ETFs… User models for economic indicator prediction could be fed back as inputs to stock/ETF simulations/ports, adding a new dimension to the tools we already have. Imagine, for example, what you could do if you had a model that predicts the unemployment rate three months out? And then use that model as an input for market timing.
George, Yuval will be handling the development of the long list of features requests. They are not going to be forgotten. He’s got a better feel for the community gripes and priorities since he was one of you. He still is but now he can walk over to the dev team and get something started
This new import functionality is something I needed for our current R&D in the Ai space . Since the import can be used by others we decided to release it.
