P123 Competitive Analysis

Dear P123’ers

We always try to keep an eye on our ‘neighbors’, but it’s easy to loose track of ball. We either dismiss an idea too quickly or are too busy with our own details to see the bigger picture. Maybe you can help out.

Who is P123? We’ve added many tools recently to allow you do do many things: research, trade, point-in-time charts, but our core offering is ranking and mechanical strategies. This could be one of our mottos: “P123 is the shortest path for an investor to directly own stocks by following a hedge-fund quality strategy”. Sure, we need to polish many things, but that is in essence what we do, and we want to make it super easy and affordable.

What are our neighbors doing?

Within the “toolset” space:

Quantopian - geared for intra-day, super hard to use. Dabbling in fundamentals but what’s the point? You are either intra-day or you use fundamentals. We’ll keep an eye on them.

Bloodhoundsystem - it’s quiet over there… couldn’t do much when i tried.

Zack research wizard - 1990’s windows front end, “lagged data”. Could be a serious competitor if they improve it. Be ready for lots of emails when you signup.

Tradestation - I need ot open an account to see it in action. One that I keep meaning to research

Vectorvest - Not very flexible. Pretty much insistent on selling the founder’s key indicators (which does make it easier to use). Also, multi continent data. They have a following and must be watched.

MarketSmith - mostly a screener, also founder centric: CANSLIM. Lot of money spent on user interface.

Wealth-lab - Fidelity bought them, very hard to use, mostly technical.

AAII - not much R&D. Reuters data

Stockrover, gurufocus, investopedia - mainly for reseach. Lots of data… Everywhere!

Stockcharts, Barchart, MetaStock, etc: technical analysis

Within the “models” space:

Foliofn - great broker, offering basic models (very basic). Most of their time is spent growing the brokerage side. Potential ally.

Collective2 - I don’t know… I always get lost there

MotifInvesting - raised a ton of money. Thousands of user submitted motifs. So many that I get lost there too. But ones to watch.

Covestor - Looked at them a while ago. I didn’t like the way you search for a model, which lead to the R2G design of “ranking” a system based on data-points you choose.

Validea - Prebuilt strategies. Has not changed in a long time.

Within the advisor space:

Then there is the whole new crop of robo-advisors like wealthfront and betterment. They are basically charging you fee to get you into funds and etfs which also have a fee. They are innovative , but not revolutionary.

Thanks for your feedback. We will spend our summer “strategizing”, looking around, and planning.

Marco - one company not mentioned is Stockopedia. I’m not sure how successful they are but there is no reason P123 couldn’t get into their market ($60 / month for US stocks). You’ve already got stock report generation capability, it just needs to be given higher level of prominence and requires strong marketing.

Then as a follow up you could provide a simplified P123 for non-geeks. Provide only the factors from the attached screenshot (down the right side) for backtest.

Steve


Hi Marco,
This link has a list of ‘backtesters’. I have not looked into all of them. Some are institutional level, but there may be a few in the list that are direct competitors.
http://quantpedia.com/Links/Backtesters

Expand internationally.

I’m back from China. Things are insane over there. Everybody is talking about the stock market. And I mean everybody. The old, the middle-aged, the young, the guy who sells you noodle on the street etc etc. Never seen anything like it.
For every US citizen who thinks maybe to invest cautiously, there are 4 Chinese citizen thinking they can make their fortune in the stock market.

So this is what you do:
1- let me buy a stake in your company
2- expand your business to China before someone else beats you to it.
3- let’s all be billionaires by 2030

On that note, there are lower cost data vendors. I don’t know how good the data quality is though.

https://www.quandl.com/vendors/dy

Steve

To add a few comments to what Marco presented

Quantopian - geared for intra-day, super hard to use. Dabbling in fundamentals but what’s the point? You are either intra-day or you use fundamentals. We’ll keep an eye on them.

Marc: Speaking of fundamentals, somebody in the Quantopian community is complaining about inability to get ROE further back than the current year. A response pointed out that he could do it on p123.

Zack research wizard - 1990’s windows front end, “lagged data”. Could be a serious competitor if they improve it. Be ready for lots of emails when you signup.

Marc: No stock ranking/quick-rank capability, other than the black-box single-style Zacks ranks

Tradestation - I need ot open an account to see it in action. One that I keep meaning to research

Marc: I have a friend who uses them. He also subscribes to p123. He does the serious work to get ideas on p123, and then uses TradeStation technical capabilities (that’s their specialty, they have fundamentals, but more as a box they can check on a list of features) to help him time his trades.

Vectorvest - Not very flexible. Pretty much insistent on selling the founder’s key indicators (which does make it easier to use). Also, multi continent data. They have a following and must be watched.

Marc: I saw them all the time at the Money Shows, where they pay big $$$ for a monster-sized booth/showing area. The founder gives shaded pitches talking about his black-box systems. That’s what they are really selling.

MarketSmith - mostly a screener, also founder centric: CANSLIM. Lot of money spent on user interface.

Marc: I’ve been there. No ranking capability except for IBD proprietary black box systems (they look like they disclose a lot, but if you try to replicate the details, you notice that they omit just enough ti preserve their IP). And, of course, no backtesting/sim

AAII - not much R&D. Reuters data

Marc: know them very well,not just the tool but the people. No backtesting at all. No rankling capability at all. Non technical-analysis factors at all.

Foliofn - great broker, offering basic models (very basic). Most of their time is spent growing the brokerage side. Potential ally.

Marc: Not a competitor at all. In fact, my continuing dream is to work with them as a partner. Their pre-set folios are not strategically designed the way ours are and they know it.

Collective2 - I don’t know… I always get lost there
Covestor - Looked at them a while ago. I didn’t like the way you search for a model, which lead to the R2G design of “ranking” a system based on data-points you choose.

Marc: Besides the issues Marco points out which are quite real, their models seem pretty much free-for-all. Some may actually use a disciplined system, but often it seems like they are seat-of-the-pants.

MotifInvesting - raised a ton of money. Thousands of user submitted motifs. So many that I get lost there too. But ones to watch.

Marc: A very interesting company. I love the way their strategies are presented. They tell exactly the kind of stories investors want and need and I think R2G designers can learn m,such from examining their presentations. Their weakness, in my opinion, besides inadequate search capabilities, is in the strategies themselves. Their themes (motif stories) make for magnificent marketing, but performance-wise, I’m not so sure clients will be as happy down the road as they think they’ll be on day one. Interestingly, they do present backtest info (notch, but something) and oddly, lousy backtest performance may be a plus for them. They are not implying in any way shape or form that past performance of even sets of probabilities, determines future outcomes. It seems almost as if they relish in the proposition that past performance sucked but the future will be different because this motif captures a trend that was meaningless in the past but is now starting to catch on. (Again, it’s magnificent on paper, but when you look closely at how solid the “plays” are, I’m not confident in the delivery.) And, by the way, they seem to understand that the motif stories are 1005 marketing. Notice how, with a free account, you can see and keep updating all the socks i a motif. They make their money on trade execution, and I assume, float, and they’re hoping the motif stories will lure you to give them that business.

Validea - Prebuilt strategies. Has not changed in a long time.

Marc: All guru screens all the time. They are all prebuilt and the only way you can screen is to say Lynch or Buffet, or Lynch and Buffett and Zweig, that sort of thing. Very minimal backtest and I have no idea how its donee (black box) or even if they use PIT data. The business is really asset management. The guru stuff is more a subtle sales pitch; they want you to open an advisory account.

Marc additions:

Morningstar (they are the ones who supply data for Quantopian): Premium Screening that looks very pretty but is not not necessarily easy to use, and not too flexible. No ratings except for Morningstar ratings. You can’t download results and what you see in results reports is horribly limited; in fact, they don;'t even give you tickers. But they do let you click once to create an on-line portfolio on Morningstar.com. You can also click on a company name to go to their on-line data presentations; they have analyst reports, but very little coverage outside the large-cap group. And I’m not sure what’s up with their data. At one time, I know they collected their own but they hated doing it and were constantly coming around to Market Guide/Multex/Reuters to talk about licensing. Not sure if they switched yet, but if they haven’t, it wouldn’t surprise me if they once do it.

Microsoft MSN Money: Backin the day, the early day of on line, all the reviewers gushed over their screener and named it number one by a landslide. But they never had rankings other than proprietary ones they bought from an outside quant shop (I forgot the name) ands no testing, and no technical analysis]. And the factors that could be created were very rigid. Ultimately, though, it’s irrelevant. They terminated the offering a few years ago. They found it too burdensome to manage the data (and that was without PIT backtesting, or even any backtesting).

FactSet: The big gorilla in our space. They charge $60,000/yr. for platform access, but then, you also have to negotiate and pay separately for your own data license. And you can’t build serious ranking systems there (a limited number of factors) and my experience was that they are not P.I.T. (something confirmed recently by someone else in a forum post).

yes there are, and I know some of them. Crowd-sourced collection is something some rely on, which is cheap, but not well controlled. With XBRL. they may be able to do enough in the way of edit to get the field numbers picked up correctly. But that’s actually the low-value aspect of the enterprise. Effective standardization (we can’t live without it; trying to model with basic collected simple data is like trying to put crude oil in the gas tank of a car and expecting to drive; both are impossible) and P.I.T. friendly database design and delivery protocols are where the money is spent.

This stuff looks easy but it takes a lot of man hours and expertise to make it look that way. Providing a great product requires a lot more than reading an article on “fitn-tech” and saying: “Wow, that’s the next big thing. I’m going to grab a piece of that action.”

So my quick take on 3 that I know

Zacks:

Pros: offers earnings estimates revisions. PIT? Not sure. Zacks rank is probably a PIT-like ranking as it is generated in house. I used the premium screen at Zacks to test if the earnings estimates are better. I used Earnings Estimate Revision > xx% and created a universe for P123 using In-List . I created a similar universe for P123’s earnings estimates revisions. I then used the same ranking system to rank each universe for 2 separate ports. I ran each port with real money. My very limited experience: no difference in results. There were some differences in the universes as was expected.

Cons: No ranks other than Zacks rank.

Bloodhound System:

Pro: According to the web page: 27 years of data. Daily ranking for backtests. Good PIT data. Can write formulas (as opposed to just factors). Same cost as P123

Con: No earnings estimates except for what is in Value Line

Note: If P123 did not exist this would probably be one of the services I would use

P123:

Pros (the things I use): Wrote the book on Sims or at least it is where I read the book. Also very good data. Ports and port ranks can be done daily. Marco is very responsive to questions and concerns: good support generally (thanks Marco). My out-of-sample experience is that you can make money here.

Cons: P123 is planning to market to a different type of customer and make some changes. Most (probably all) of the changes will be good.

Quantopian is extremely early in their development and not hard to use if you know Python.

Bloodhound is interesting and worth a look but like P123 its a work in progress, but we don’t know much about its capabilities yet. In a call to inquire they mentioned their target market is asset managers and “haven’t opened a retail account in over a year”. Also mentioned that they’re extremely flexible to the needs of their users and can create simple add-on features (such as what we’ve been asking repeatedly for) in a day or no longer than week.

“A rose by any other name…”

They call me “professional” level here too. I am in fact a retail investor and the price is the same.

Edit: I know ClariFi Xpress is really for pros and significantly more expensive. I had it on my list to see how much more expensive. Anyone know?

Quantopian:

It seems to me that Quantopian is weak where P123 is strong and vice-versa. While the Python language provides a lot of flexibility, one needs to code most everything. P123 provides an excellent framework - dynamic universe, ranker, hedge module, etc - that allows one to just plug in rules and go. Quantopian isn’t there and probably never will be.

Macroaxis - http://www.macroaxis.com/ - is interesting. They provide MotifInvesting like portfolios and portfolio mean-variance optimization. One can import stocks lists and run MVO, too. Someone there apparently has a lot of time, since the UI is loaded with small animations. The most basis account is free.

Walter

From what I can see over there, they are not a competitor to p123 but a possible add-on; to get started setting something up there, you need to have a list of stocks.

It looks from what i can see that they relay on MPT/mean-variance optimization. If that’s so, I would not look for them to have a good future. That’s an ac anemic cornerstone (supported by a Nobel Prize) bit in practice has by now been pretty largely discredited due to the impassibility of coming up with reasonable estimates for the inputs and the sensitivity of the results to imperfect inputs; academicians today describe it as an “error maximization” model. If you run into a site that offers portfolio optimization, they should at the barest minimum be using the Black Litterman MPT adaptation. Failure to do that is an easy way to distinguish between one of the many who are simply looking to ride the fin-tech craze and somebody who is for real.

Hi Marc,

You’re right about the macroaxis MVO aspects.

I interpreted - http://www.macroaxis.com/invest/investingOpportunities - as MotifInvesting like. But since I haven’t really tried either, I may be wrong.

Walter

EDIT: Just wanted to add that I agree that they are not a direct competitor to P123. However, they have features I would love to see in the Book module - like a stock level correlation matrix.

EDIT: On 2nd thought, maybe a 50+ x 50+ matrix would be too much.


My 2 cents on Tradestation - it’s absolutely fantastic if you want to trade technical systems, great for developing simple to extremely complex systems. Their reports and stats are extremely useful. Their programming language is easy and flexible. I used it all the time when I traded futures. In my opinion far better than Metastock or other technical-based programs.

Marc - Quandl started out as crowd-sourcing but now they are pulling in a lot of vendors such as Zacks (fundamentals/estimates/EOD pricing+graveyard stocks) for what seems like a reasonable price.

Steve

I’ll leave to Marco the task of discussion what it really takes for a database to work with p123; if I try to answer in detail, I’m likely to screw up the explanation. Suffice it to say the sort of P.I.T. quality that folks here expect (just keep your eyes open as people raise data issues) was incredibly challenging for Reuters and are so for Compustast too. It’s easy to just present data; but the issues we have to deal with are quite complex, which is why when Marco calls me to talk about the issues you guys find and raise in the forum, my first response is usually something along the lines of “Oh sh**, I don’t know” followed by a lengthy discussion with Marco and discussions between Marco and Paul and often sometimes lengthy debates between me and Paul as to how to handle things. I won’t say we’d never use Quandl; I don’t know enough to have a specific opinion about them. But it takes a lot more than a cheap-and-easy story to satisfy our incredibly demanding needs. We have encountered other low-cost data vendors, but they tend to fall away after failing be able to give satisfactory answers to Marco’s preliminary questions.

Marco,

From a user (not a builder) perspective, I think you guys should maybe look at things like “Sector Surfer” and “ETF Replay” as the interest in the type of strategies has become huge.

And if your best strategy builders started to imitate the more complex models we read about on web forums (see Alpha Architect’s blog where they review all the latest research in finance theory…or Research Associates, CXO Advisory…etc) then you guys would get a nice piece of the rule based active management / smart beta market market that’s now heading toward ETFs, mutuals, and hedgies.

Fwiw, if a well built model was open and not a black box I’d gladly pay $50/mo since when there’s larger sums involved the 1%-3% fund fees are much more than $600 a year model rent. Also I think you guys should vet some of the crazy models that are curve fitted with back-tests that have no chance of represent any future reality.

Anyhow I’d love to see something like this implemented as a P123 model: http://blog.alphaarchitect.com/2015/05/11/momentum-investing-skewness-enhanced-momentum-yields-double-alpha/

PS…I love your site and you have done a great job bringing quant tools to the little guy and in crowd sourcing a bright quant builders pro bono workforce =)

Marco,

Thank you for sharing your thoughts on this topic.

I may be wrong, but would not P123 with international data and the Quantopian iPython Notebook IDE be a viable competitor to ClariFI? If yes, wouldn’t that draw a highly attractive and sticky customer demographic to P123?

You do have an affordable platform that is generally super easy to use. But, why not extend the offering in the direction of professional or academically guided, fundamentally-oriented quantitative investors? Perhaps one should consider where the population of potential users in this demographic will be in 3 years. I would speculate in the direction of rapid growth.

Python would offer so much more flexibility and the opportunity to conduct more fine-grained research on existing strategies (as well as new ones) and efficiently improve them (through deeper understanding, not over-optimisation) over time. And, the iPython Notebook is such a great tool for documenting work and presenting results. It’s possible to replicate some of this in a limited fashion with P123 today, but only with great creativity, work-around skills and time.

All this being said, none of your direct competitors comes close to facilitating the development and implementation of intelligent, fundamental quantitative strategies like P123 does.

D

“Also I think you guys should vet some of the crazy models that are curve fitted with back-tests that have no chance of represent any future reality.”

bazooooka - I agree that “crazy models” should be vetted but there some practical problems associated with that. First of all, resources would be diverted form other important tasks. Second, who is to say what sound model design is or is not? In some cases it is blatantly obvious that a designer is cheating. In most cases it is not. I posted this link in another thread but I’ll post it here as well. Pay particular attention to 1 (a).

http://www.gipsstandards.org/about/governance/documents/meetings/2009/tab_17_usipc_suggested_practices_theoretical.pdf

Steve

Quantpedia has a nice summary of backtest software/websites with company, products and pricing here . There are 28 listings, with P123 being one of them.