First of all, thank you for an excellent platform — the tools, flexibility, and research capabilities in Portfolio123 are truly outstanding.
I wanted to ask a more detailed question regarding data expansion.
Is there any plan to introduce alternative data sets into the platform?
If anyone has experience integrating external alternative data into Portfolio123, I would be very interested in hearing how you approached it.
In addition, I’m curious whether there is a roadmap for adding more fundamental variables, such as backlog, industry-specific KPIs, or other operational metrics that appear in filings but are not currently included in the standard database.
Overall, I am trying to understand whether there are upcoming plans for expanding the data universe, or whether users have found effective ways to combine external data sources with P123’s Platform.
Thank you in advance to anyone who can share insight or experience.
One avenue to explore is importing stock factors. That allows you to integrate alternative data into your Portfolio123 systems. Another Portfolio123 user and I are beginning to import options data; apparently there are some very good signals in that data.
You mentioned "other operational metrics that appear in filings." Because I can get access to some fundamentals that aren't on Portfolio123's list of line items, I'd be curious as to what you'd recommend importing. I haven't done anything with these yet; it would take a bit of work to access them first, so I have been putting it off.
Very interesting. I look at volatility skew to get an idea of participant expectations sometimes. Would be interesting to be able to use that data more systematically
One issue in this regard is that there is a hard limit of 40M rows for importing stock factors.
40 million rows sounds like a lot, but if you import weekly data from 1999 - 2025 for the Russell 3000 tickers, you get about 52 * 26 = 1456 weeks x 3000 stocks = 4.3M rows.
So that means you can upload less than 10 factors, even if you limit yourself to the Russell 3000.
Yes because of a terribly inefficient implementation which we plan to fix, together with other refinements to make imported factors more usable (automatic expiration and series operations for example). As usual, it's a matter of prioritization.
Yes, we are looking to add more:
Alphavantage offers some interesting factors and they are reasonably priced.
SEC factors derived from the Business Overview, Risk Factors, Management's Discussion & Analysis (MD&A). In other words using AI/ML to extract scalar values from text. We're working on this with someone
Financial ratios we already get, but have not exposed (oversight most likely)
Our own factors derived from what we already get. For example the insider transaction ratings
Depends on the cost. Currently we're focused on adding Asian data which has many backers.
Thank you for the response — much appreciated.
I completely agree that options (and corporate bonds) can be a rich source of additional signals.
I’d be very interested to learn more about your experience.
If you don’t mind sharing, where are you sourcing your options data from, and what does your dataset include?
Regarding the fundamental variables, I have a few ideas that I believe could be useful — and, in principle, not too difficult to implement:
Separating Sales & Marketing expenses from the broader SG&A item
Separating IFRS 16 lease assets and liabilities
Backlog and Bookings (in the Israeli market, revisions in these items are very strong signals — it would be interesting to test this across other markets as well)
Patents and trademarks data
Sector-specific fundamentals, especially for banks and insurance companies
I think these would be valuable and low-hanging-fruit additions to P123.
Regarding the SEC-derived factors from the Business Overview — that would be huge. It would be very interesting to hear more about this initiative and what the plans are. If users will eventually be able to derive or create their own factors from these disclosures, that would be a significant development.
As for the financial ratios — if you can share which ratios are planned to be exposed, that would be great to know.
Yuval — if the options data works out, I’m not sure whether it’s something you’d want to explore, but I’ve become fairly convinced that the most efficient way to hedge is at the portfolio level using deep out-of-the-money, very short-dated index puts.
The basic idea is that by accepting small, consistent losses on the hedge, you can justify running the portfolio hotter than you otherwise would. In good environments, the additional equity exposure pays for the hedging cost and adds incremental return.
Crucially, during sharp drawdowns, the hedge doesn’t only reduce damage — it creates an influx of liquidity that can be redeployed into assets at depressed prices.
I’ve been running and refining this approach since pre-COVID (roughly 6–7 years now). While it’s anecdotal rather than rigorously backtested, it has worked well for me in practice. For the most part, the process is highly procedural: puts are bought and rolled consistently, almost always for a loss.
The only part that becomes more art than science is managing a payout. For example, I navigated this summer’s VIX spike well, and that single episode covered most of my annual hedging cost. It’s possible that even this phase could be further systematized, although the rarity of true stress events means that even a decade of data may not provide enough signal to do so reliably.
You posted something about this a couple of weeks ago and I backtested it, using QQQ puts. I found that it didn't work as well for me as puts on individual terrible stocks and shorting a large portfolio of terrible stocks. I was using 8- or 9-week options at about 75% of the current price of QQQ. But when you say very short-dated, maybe that's a lot shorter than 8 weeks? Are you using one-week or two-week options? Those do tend to be extremely cheap (low IV), so maybe I should backtest those? And when you say "deep OTM" do you mean about 75% of the stock price? But how often does an index fall 25% in two weeks?
I keep bringing it up in hopes someone with better technical chops and data can help me see what I may be missing : )
I consistently use SPX puts, typically 2–3 months out. SPY puts should behave similarly, though index options have some advantages around taxation (not advice). I size them off delta, but they’re usually around 35–40% out of the money. So yes — it’s a persistent premium drag.
Almost never — but isn’t that the point? Isn’t crisis risk the primary thing we actually need to hedge?
For slower, grinding bear markets, I rely on my tactical and timing models to adapt.
My view is roughly this:
a) Hedging should primarily target extreme, discontinuous events, not routine volatility
b) The hedge needs to be cheap, convex, and explosive when it hits
c) To afford that insurance, you need to lean on it to justify higher baseline risk exposure, allowing higher returns during normal regimes to fund the hedge
d) Even rare, infrequent payoffs can materially improve geometric return by injecting capital during drawdowns — capital that can also be redeployed into risk assets at distressed prices
e) Because long-term wealth is path-dependent, a single large convex payoff — even if extremely rare — can materially improve outcomes by truncating left-tail losses
When I traded discretionarily, it was relatively straightforward to pair position-level stop losses with a portfolio-level hedge to manage gap risk.
Moving to tactical ETF and stock models has forced me to think more carefully about the tradeoff between building a more structurally durable portfolio versus running higher-risk models and relying on this hedge layer for tail protection.
I think this approach has quite a few virtues. Besides those which you mention, these puts are relatively cheap since they're so far OTM. Not only that, they're very liquid.
Personally, I prefer ATM medium-range puts on low-ranked stocks, because they'll definitely pay off during crashes but will also occasionally pay off during normal markets too. In my own backtesting, it seems optional to use a mix of modest leverage, a small put-based hedge, and a small shorts-based hedge. But your approach sounds good too.
I appreciate your thoughts, Yuval — thank you for sharing. I don’t have the experience to design my own position-based short or hedge, but it’s something I’d be very interested in exploring, as I prefer using a diverse mix of tactics rather than relying on any single approach.
The TZA-pairing discussion was particularly insightful because it allows me to directly hedge a portion of my portfolio. That, in turn, lets part of the portfolio run without the SPX hedge, while also helping smooth the ride in terms of day-to-day volatility. I know I shouldn’t care too much about short-term fluctuations — but in practice, it’s hard not to.
My suspicion is that running more straight long exposure (i.e., not TZA-paired) and relying primarily on tail-risk hedging would likely perform better over the long run. That said, at the end of the day I still need a structure I can live with through the inevitable drawdowns and volatility.
Your model will capture the 80 vol moments but hese are very rare. The last 40 years we had two: a financial meltdown and a global pandemic. We may see them more regularly but, who knows, all those liquidity valves that governments like to use these days may make them less frequent too. Most crashes and bear markets drop not fast enough. 2000 and 2022 are two examples. If your models can capture the slow grinds down, that may be not an issue but for most of us I think it is.
I buy an 18 month out SPX bear spread (typically a 15% OTM long put and a 40% OTM short put). I sell it after 6 months (or when the long is singificantly ITM). You don’t have the vega explosion but the impact of theta, premium costs and transaction costs are more manageable. I guess it is a different strategy. Mine is a portfolio hedge, yours is a tactical tail risk hedge.
I’ve merged with the CBOE options database, its cheap data and has lots of signal. I would recommend that P123 add it, or at least summary statistics for the chain for each stock on each day. For example, when there is above average put volume for a stock, thats usually a bad sign.
The only drawback with the CBOE data is that it starts in 2012 and is quite sparse back then. Although that might just reflect reality, that options just weren’t as popular back then as they are now.
I would also like to add that the most significant alpha I’ve ever found has been in borrow rates. Its expensive data from S3 but highly profitable. I think there are some other providers but its all very expensive.
I try to layer defenses within the portfolio and, importantly, rely on the models themselves to adapt to more prolonged, grinding bear markets.
Using Yuval’s Microcap Model as one example, the 1999–2003 period produced returns of 78%, 65%, 133%, 51%, and 260%, while 2022 came in at roughly –10%. In most environments, then, the hedge primarily serves to justify staying more risk-on. The excess return generated during normal periods helps “pay for” the protection needed in true tail events.
Where I still wrestle is whether my current mix of models mutes returns more than it should. If mitigating tail risk is the most critical objective, as I believe it is—and if the puts are already handling that—then maximizing participation during the good times becomes just as important.