In short: a company can raise funds from non-accredited investors up to $2,000/investor. So 1,000 investors/P123users = $2M. Throw in a few accredited investors and we could easily raise $3M-$5M. We can do everything with that. Its new, different, anti silicon valley, robinhood-esque, and it all fits in with our philosophy of cutting out the middle-men (management fees and now VC fees). This summer I invited 4 MBA students to brainstorm a business plan. Who is in for a September pitch ?
Hi Marco,
that sounds great to me. I like the “Robin Hood” part
What worries me a little bit is the “quality control” part of this whole endeavour.
Growing fast and going into new directions is all fine. P123 success will rest on quality on all issues. Make sure, “quality control” is strictly enforced on all fronts.
I just downloaded and tried EquitiesLab. This is in no way comparable to P123, but rather seems in a very early beta stage. Not something that could be reasonably titled “competitor”.
However, I’d like to mention another more mature site, https://screener.co/
which covers only SCREENING, i.e. it caters only for an extremely narrow part of the broad spectrum P123 offers.
A big PLUS there is the large number of stocks covered (30k+ companies), including EUROPEAN and other markets.
@Marco: looking forward to hear from you about this. If the 4 MBA guys will not focus on this topic in summer already, I’d like to offer my support.
Please take another look at TradeStation. They have a much more flexible and powerful user interface via Easy Language. For technical systems they are superior to P123. Their fundamental side is weak. You could develop P123 into a truly awesome machine by including an interface that's as flexible and powerful.
Count me in for the Sept Pitch. My (strong) wish is to see further development of the research platform with (for instance) an iPython Notebook IDE and international data. The current focus on ease of use is wonderful to start, but entirely hamstrings proficient users later. A few useful functions such as FRank() in the Screener are great - and can be armtwisted into accomplishing quite a lot - but they are not sufficient. There should be one interface that is useable across the entire platform that doesn’t constrain users except to ensure that data use limitations are observed.
What about creating a type of “superuser group” that can more fully detail some of these requirements and motivate the rationale for investing in them? There are tensions between R2G developers / users and research-driven users that should be addressed. The focus could be working with your MBA students to move through strategic / growth objectives in a systematic manner for R2G user needs, research user needs, etc…
I am also very interested in European data,
I had a subscription to valuestockscreener
they offer high quality world wide data including Europe with screening but without backtesting.
.
As far as ideas from the competition. Bloodhound system handles range of returns in this way: Bad Screen Shot Here. Click on return probability.
I’m not sure they did it right (e.g., Monte Carlo?). Just consistent with the original post in the thread.
For now, the most accurate method might be to randomly select a year from The “Annualized Performance By Calendar Year” in the performance statistics, plug several (n = target years) of these randomly selected returns into a spreadsheet, average (consider using the geometric mean) and do it a few times. Use you best and worse value as your range of annual returns. Realize there will be a Black Swan (something way outside of your sample) at some unknown date in the future. If I really wanted to know this, I think that is what I would do.
Not a priority: in fact I’m happy to do it on my own rather than use P123 computer resources. I get that some of this may be for the SEC. If so, a modified (improved) version of the short algorhythm above could give a wide range that may protect you well–there will be a wide range over the shorter time periods and this may be realistic.
What I found interesting about Quantopian was the possiblity to import your own data.
The only problem is that you have to upload it again for every backtest.
Probably they do not want to store your data permanently on their servers.
I would like to see a Quantopian type of fundamental portfolio creation/management platform.
I’d love to see P123 get there first but also want to see P123 deliver on daily data, auto trading and Canadian TRADE.
The “industrial strength” ClariFI Modelstation combines PIT data sources, time-aligned database, ability to create robust factors and flexibility in creating ‘scripts’ of how the portfolios are created (screens and ranks), rebalanced and risk-managed. PIT data is external to the platform. Time alignment is now solved in many languages, in Python everyone uses the Pandas package. Creating factors robustly is not very hard, for example if a data point is NA, you can look back and carry a number forward, or forecast it, or insert an industry mean or median.
The portfolio management programming flexibility is the key to the value add. For example, every time new data is updated for a company in the portfolio, rerank and if out, you have the choice to continue to hold till next rebalance, replace with new company, re-weight existing holdings, hold cash, buy sector ETF or buy index. This is done daily. “Rebalance dates” can be selection and reweighting dates or just reweighting dates with the portfolio selection done daily or based on tracking error drift or on returns of individual positions. “Reweighting” can be done for targeting minimum variance or tracking error or volatility targets.
