Quant Blogosphere Herding
It’s interesting to watch the herding in the quantitative blogosphere over time (and I’m just as guilty as most).
In 2008, when I first started blogging, it was hard to find anyone not talking about RSI(2), DV(2), or some other whiz bang indicator that measured short-term mean-reversion. That inefficiency dried up and so did the subject.
Today, the topic du jour is strategies of the longer-term variety: trend-following and momentum, tactical asset allocation, risk parity, volatility premiums, permanent portfolios, market valuation models, etc.
I assume that’s partially a result of the market closing a lot of the short-term plays we quantitative types were profiting from, and partially a result of these longer-term approaches doing such a good job of timing the two big bear markets of the new millennia.
Of course, there are deviants from the herd.
I’ve focused a lot of energy on volatility trading strategies (along with VIX & More, Six Figure, Only VIX, etc.), seasonality plays always get some ink (although I’m less and less inclined to pay them much heed), and Rob Hanna has been doing good things analyzing the overnight market, just to name a few.
But generally speaking, we quantitative-types herd.
I just can’t help wondering, where is the herd going next?
Those longer-term strategies are great for what they’re intended for: long-term investing, but they fail to produce the oversized return stream today that most of us are chasing.
And we’re all effectively cut out of the HFT game, so despite that being Wall Street’s play du jour, we have to take that off the table.
For a number of reasons (but mostly because of the asymmetric risk/return tradeoff resulting from the VIX futures term structure) I’ve planted my flag on trading volatility, but that’s by no means the only answer.
To the other quantitative minds out there: where are you planting your flag? What’s your next big quantitative play? And how are you breaking from the herd?
Happy Trading,
ms
. . . . .
To stay up to date with what’s happening at the MarketSci Blog, we recommend subscribing to our RSS Feed or Email Feed.
Filed under: Random Stuff | 9 Comments
As a researcher switching from industry to academia, I am heading for summarizing interesting and practical academic paper and publishing on my quant blog.
Hello abiao – I just recently got turned on to your blog. I like that approach (along with Turnkey Analyst as well) – very useful for someone like me who doesn’t have the time or disposition to keep up on stacks of acad papers. Good to have you around! michael
an excellent point and as I’ve mentioned here before, it’s very likely that given that hedge funds hire the VERY best talent (I’m talking PhD stat people) that any edges are by and large gone. as all market alpha is reduced to zero or close to it the “quant” people will become increasingly discouraged and will fade away, an inexorable trend.
Hello Steve – good comment – that may be over the horizon, but I think we’re a long way from that point. One only need look at how very recent, very simple, and very effective inefficiencies like daily follow-through persisted for a very long time.
I think the folks that you term the “very best talent” (and I term something much less impressive) will close non-beta-driven plays that require economies of scale, technology, etc. to capture (like HFT) much faster than the beta-driven observations that most individual traders are chasing. Beta-driven observations inherently carry risks that are more easily navigated by the smaller more nimble investor.
Just my $0.02…michael
michael, so if you’re right where are the outsized “beta-driven” PROFITS. I don’t see them. what I see are backtests and as you have previously written, they are virtually useless. in fact I would argue worse than useless because they delude the observer into thinking there’s something there when there isn’t.
show me the money!
I hope that indignation isn’t directed at me. MarketSci (and for the most part, only MarketSci) has always leaned heavily on verifiable returns and always refused to release backtests. Our strategies crushed the market in absolute and risk-adjusted terms from 2006 through mid-2009. We sucked for two years after that, but there were plenty of folks who didn’t. And now the strategy that I’ve planted my flag on is running at more than 60% annualized for the last 10+ months. All real-time. All verifiable. All accessible by small investors. There’s the money.
Interesting article, Michael. It’s great to see you posting your thoughts/findings regularly again! David Varadi recently posted a similar blog entry regarding the markets and future of quant methods. My response to his article was similar to what immediately comes to mind after reading through the above post.
———–
Regarding the comments made by “Steve”:
For the time being, the easy money has been removed from the market (my $ .02). Just like any other business, technological advancement over the past 10-15 years has had tremendous impact. As a whole, we, active traders, have both benefit and been hurt through advancement. I stumbled upon CSS Analytics (and subsequently, MarketSci) immediately after graduating college, and saw the death of simple mean reversion indicators. As a result, I’ve found it quasi-comical to see blog-goers comment about the market being skewed in favor of the institutional players; as though they even had the ability move their capital in and out of the markets like the average retail player does. Though it’s never mentioned, the retail players do have advantages that institutions do not… we can move in and out of positions without worrying about shifting prices to any significant extent, take as little or as much risk as we please, changing trading styles at our own discretion, and use any product that we please when building portfolios. There’s something to be said about this type of flexibility.
———–
Back to the topic at hand, I think that the herd effect was/is responsible for the death of MR indicators. Moving forward, I’ve planned to allocate capital to strategies that are a bit more complex than what we’ve used in the past. I’ve started to do things like incorporate volatility risk premiums into weighting schemes within my trading models; namely, look-back periods, and leverage signals. Also, I don’t see how, at this point, it isn’t painfully obvious to all traders that overseas markets can have tremendous impact on our daily P/L. I’ve had some success trading based off of inter-market volatility adjusted divergences. As always, using/developing the latest KPI’s, I feel, will be essential moving forward (of primary interest, the Omega ratio, and variants of). Tony Coopers paper on volatility of volatility has led to me doing a bit of R&D in building an additional filter into my pairs trade models, that helps predict when correlation of the pair may soon break. However, keeping in mind that the herd will crush a real edge, I can’t help but wonder how long volatility of volatility will remain as predictable as it has to this point. I think that developing aggregate indicators that sum data into a simple signal using data from stocks/etf’s AND their derivatives will also provide an edge. Neglect of derivative markets is something that has always irked me within the blogosphere. Lastly, driven as a direct result of your most recent article(s)… I feel that the overnight session is likely worthy of developing sets of indicators/tools for. I’m quite interested in investigating how to hedge against overnight risk, efficiently (read: on the cheap!), with the intent of maximizing equity compounding.
That’s about all that immediately comes to mind, and as long-winded of a response as I think I can make before deeming my own message as absurd.
Best,
Thanks, Michael. For people doing (bottom up) quant stock selection, figuring how to time that using more top-down approaches is a big topic these days, perhaps enough to call it enough form of herding in itself.
As for edges getting competed away, personally I think that varies a lot with the type of the edge and type of investor. Shorter-term ideas with higher Sharpe Ratios both attract more competition and have less capacity (because of the greater trading rate required), so would seem the most subject to competition and so the most at risk. Longer-term ideas (say value investing) have both more capacity and more career risk for professionals (you can get fired before your investment pays off) so are more able to survive (though are subject to crowding too). Others have written about a related distinction between informational, behavioral, and risk premia strategies. And the influence of the macro environment (and plain statistical noise) muddies our ability to separate what’s crowded from what’s just out of favor (or untenable in the current environment). What to do with assessments of crowdedness should also be influenced by the kind of investor you are – small investors (outside of high frequency space) have some advantages being small (lower costs, possibly different career risk) that let them do some things well that aren’t scaleable for big institutions.