August 15, 2016

Systematic traders - keeping it simple and divergent

"[A] model will be a simplification and an idealization, and consequently a falsification. It is hoped that the features retained for discussion are those of greatest importance in the present state of knowledge." 
- A.M. Turing.

The real trouble with this world of ours is .. it looks just a little more mathematical and regular than it is; its exactitude is obvious, but its inexactitude is hidden; its wildness lies in wait.

- GK Chesterton

Models break. Because the world changes, the factors that drove performance yesterday may not be the same as tomorrow. This is why simple models are best in a more complex uncertain world. Complexity means there are multiple factors necessary to explain change. Uncertainty is not volatility but our inability to measure some factor or to even know it exists. A world of multiple factors which change in importance is the hallmark of financial markets

So what can be done about this problem with models and complexity? 

Keep it simple. This is not the same as simplistic. You just have to realize that for each extra factor there is greater potential for something to break. Data could be wrong. The relationships can change. On the other hand, if there are not enough factors, the model will to capture a key relationship. In keeping it simple, there is the realization that a model is only an approximation and that failure is real.

We also like the construct of convergence and divergence effects to at least think about how the world operates. Markets are dynamic and move between converging to some equilibrium price or moving away from some equilibrium in response to some shock.

There are market dislocations or divergences. These "shocks" are unanticipated and will have the greatest impact if they have not even been modeled or considered. An awareness of divergences is critical to understand models as falsifications or reality. Similarly, convergence suggests that dislocations in price will move back to fair value or to what may be considered equilibrium as structured or explained in a model.The world is still driven by some fundamentals relationships.

An ebb and flow of divergence and convergence can frame why models in the short-run may not work but in the long-run may be effective representations. As Herb Stein said, "If something cannot go on forever, it will stop." The divergence of today may be the convergence of tomorrow. The model often cannot explain all divergences but may still be effective as stating that prices may move back to some norm.

November 26, 2015


October 14, 2015

PTJ on Investment Philosophy

"I think I am the single most conservative investor on the earth in the sense that I absolutely hate losing money. My grandfather told me at a very early age that you are only worth what you can write a check for tomorrow, so the concept of having my net worth tied up in a stock a la Bill Gates, though God almighty it would be a great problem to have, it would be something that’s just anathema to me and that’s one reason that I’ve always liked the futures market so much, because you can generally get liquid and be in cash in literally the space of a few minutes. So that always appealed to me because I could always be liquid very quickly if I wanted to. I’d say that my investment philosophy is that I don’t take a lot of risk, I look for opportunities with tremendously skewed reward-risk opportunities. Don’t ever let them get into your pocket – that means there’s no reason to leverage substantially. There’s no reason to take substantial amounts of financial risk ever, because you should always be able to find something where you can skew the reward risk relationship so greatly in your favor that you can take a variety of small investments with great reward risk opportunities that should give you minimum draw down pain and maximum upside opportunities." -- Paul Tudor Jones

August 13, 2015

Are you a trend Follower ?

From Jerry's Parker Twitter:

Non trend follower:
what's going to happen with this trade?
Trend follower:
What's going to happen with all of my life's trades?

June 14, 2015


Modernity has been obsessed with comfort and cosmetic stability, but by making ourselves too comfortable and eliminating all volatility from our lives, we do to our bodies and souls what Mr. Greenspan did to the U.S. economy: We make them fragile. We must instead learn to gain from disorder.” – Nassin Taleb

May 31, 2015

Smooth Risk ?

"A imperfeição da estratégia trend following é a principal razão da sua robustez, reseliência e estabilidade." --Jerry Parker

Blogpost 3 - smooth-sea

By Moritz Seibert, edited by Niels Kaastrup-Larsen
The CTA industry has changed a lot during the past ten years. New regulations and increased demand from institutional investors, especially pension funds, have caused many CTAs to change (i) their trading programs and (ii) their business operations.
Instead of representing the core investment philosophy of the manager, it seems that many of today’s trading strategies are designed to suit the investment preferences of large institutional clients: low volatility, small drawdowns of – ideally – short duration, and a consistent return stream uncorrelated to other investments. Obviously, such a “feel-good” program can also make for a great business (steady fee revenue).
However, investors should be careful what they wish for.
“Return-smoothness tends to come at a cost and the risks are often hidden.”
The rough side of a smooth return stream
Traditional trend following is the opposite of smooth. Maybe counter intuitive, the roughness of trend following is the primary reason for its robustness, stability and resilience over time. But the lumpy returns make it a trading style that’s very hard to like for most investors (relies on price only, shows more losing than winning trades, almost always in drawdown, high return variability, and so on). Smoothness, in contrast, is much easier to like.
To smooth the performance of a trend following trading system, managers may
  • increase portfolio diversification by adding more (different) markets to the portfolio, spreading trading signals across multiple time frames, and expanding the range of entry and exit signals (e.g., by combining breakout and moving average-based trading styles)
  • alter the workings of a “rough” trend following model, or
  • add non-trend following models to the portfolio.
Out of the above, (a) is a great choice because it adds to the stability of a trend following portfolio,but (b) and (c) tend to come with considerable side effects.
With respect to (b), when altering the design of a trend following trading model with the aim of making its returns more likeable, the risk is that the additional complexity that’s required to develop a system that prevents large losses, reduces drawdowns and takes a higher percentage of winning trades, actually weakens the overall portfolio. That’s because additional trading parameters tend to lessen the sample size and make the system increasingly dependent on past market environments which are unlikely to recur in the future.
“Simplicity on the other hand avoids over-optimization. It increases a system’s robustness at the “cost” of roughness”.
Regarding (c), the introduction of non-trend following models to the portfolio (e.g., mean-reversion) can indeed work wonders. However, many of these models don’t represent skill (alpha) and come with hidden risks – especially tail risks. Most mean-reversion models work most of the time and hence a backtest will suggest that they are a nice counterbalance to trend following. What investor’s should understand though is that most of these models include “time bombs”. The question is not if these bombs will explode, but when. And, when they do, the damage will make it clear that these models have been traded at the investor’s expense: the investor has all the downside, but the manager just goes out of business. Long Term Capital Management traded in the markets on somebody else’s dime in the same way.
The above risk does not only apply to mean-reversion but to any system that’s focused on a part of the return distribution where theoretical probabilities are tiny but real-word probabilities are not. These models make for great back-testing, but they tend to be very fragile. Personally, I think that trend following – despite its roughness – is the more honest trade. At least it gives the investor a return ticket from a drawdown and, as a result, it’s fair to say that pure trend following is actually more protective of investor capital than some of the cleanest looking trading systems.
“To be fair, not all non-trend following models are to be dismissed”.
There are certainly some good and sustainable ones out there and Kudos to those managers who are trading them for their clients. But when returns look too smooth and Sharpe ratios start to exceed 2 or 3, investors should ask themselves and their “Smooth Operator” (Sade, 1984) why they are invited to participate in the trading program in the first place.
Summary: investors do not necessarily lower their risk by lowering the volatility of their returns. Volatility and risk are two very separate matters. If investors are too risk averse and too keen on obtaining a smooth “all weather” return profile, odds are that at some point they will end up with the opposite.