1. “The less you know about how sausages and models are made, the better you sleep at night.”
    Sometimes these blind spots don’t matter. When we ask Google Maps for directions, it modeled the world as a series of roads, tunnels and bridges. It ignores the buildings, because they aren’t relevant to the tasks. When you operate an airplane, it models the wind, the speed of the plane and the landing strip below, but not the streets, tunnels, bridges and people, because they aren’t relevant to the tasks.
  2. All models have flaws, and no amount of revision removes all uncertainties. There is always need for further research to provide deeper perspectives.
    The differences between a minor glitch and a full-blown catastrophe often hinges on how much goes by before somebody notices it. The best control flags potential trouble before serious damage is done.
  3. I brush my teeth every day, twice a day. That doesn’t mean I’m ready to perform a root canal on somebody.
    As far as Arps being old, the wheel was invented a long time ago, but it still comes in handy.
  4. Trying to predict the weather that far out is like trying to calculate the last decimal in pi.
    Mean reversion can be delayed, but it cannot be averted.
  5. The big in big data matters, but a lot less than many people think. There’s a lot of water in the ocean, too, but you can’t drink it. The more pressing issue is being able to use and make sense of data.
    A model is just a toy, although occasionally a very good one, in which case people call it a theory. A good scientific toy can’t do everything, and shouldn’t even try to be totally realistic. It should represent as naturally as possible the most essential features of the system, and the relationships between them, and allow the investigation of cause and effect. A good toy does not reproduce every feature of the real object; instead it illustrates for its intended audience the qualities of the original object most important to them. A child’s toy train makes noises and flashes lights; an adult’s might contain a working miniature stream engine. Similarly, good models should aim to do only a few important things well.
  6. Tomorrow is like a race to be run over unknown terrain. We don’t know whether a car or a horse or a boat will have the advantage. If tomorrow’s race is over water, the kind of car you have does not matter. Good cars or bad cars both sink. However, if the race is run over a road, a car in good condition has a better chance than a badly maintained car.

    Risk management is like car maintenance. It anticipates and prevents some problems. It identifies and facilitates some opportunities. While it is true a badly maintained car can still win a road ace, that does not mean risk management and car maintenance are not valuable.
    Melting Ice Cube and Statistical Models

    If you see an ice cube, it’s easy to predict the shape of the puddle that will result when it melts. But if you see a puddle, it’s nearly impossible to determine the shape of the ice that created it. Similarly, a statistical model will give precise prediction of possible outcomes. However, observing events does not help you build a model. There are an infinite number of models that could possibly underline historical events. Unless you have a strong theoretical reasons for knowing the type of model, and there are a reasonably small number of parameters relative to the amount of data, modeling is as hopeless as trying to guess the shape of an ice cube from observing a puddle.
  7. Whenever we make a mathematical model of something involving human beings, we are forcing the ugly stepsister’s foot into Cinderella’s pretty glass slipper. It won’t fit without sawing off some essential parts. Financial models, because of their incompleteness, inevitably mask risk. You must start with models but then overlay them with common sense and experience.
    The South Sea Bubble of 1720 is one of the earliest, largest, and most studied instances of investment manias and crashes. It is frequently cited as the prototypical case of irrational exuberance. Isaac Newton’s role in it continues to fascinate the public. Tales abound of how he invested early, and cashed out with 100% profits as prices went to what seemed to him unjustified levels. But then, as prices continued to advance, he supposedly invested again at the peak and lost most of his fortune in the crash that followed. He is claimed to have said that “I can calculate the motions of the heavenly bodies, but not the madness of people,” and supposedly could not bear to hear of the South Sea affair to the end of his life
  8. Management Principle - Triangles

    Conceptual Selling, Rapid Prototyping, Discovery-driven Planning.
    With our team, we are delivering more innovative, efficient and effective solutions to our clients. We do this by following 3 simple steps which we call modeling best practices.

    First, listen to the customer to really understand what the customer needs (conceptual selling. Conceptual selling is all about putting yourself in client's shoes. Understanding the challenges the client faces, so that we are best positioned to help them). Second, rapidly develop some potential solutions and then engage the customer in a round of discussions until the customer is comfortable that the solution meets the need. (Rapid prototyping. It allows clients to tease out their own wants and be a partner in model designs). Third, track important metrics throughout the process to make sure that the solution delivers the expected financial results. (Discovery driven planning. It is about having well articulated milestones).
  9. This reveals a view commonly held (even today) that there is some single model that describes the data generating process, and that the job of a forecaster is to find it. This seems patently absurd to me - real data comes from much more complicated, nonlinear, nonstationary processes than any model we might dream up - and George Box himself famously said "All models are wrong but some are useful."
    The perfect model doesn't exist, so we have to use imperfect ones intelligently. Smart traders know what you have to combine quantitative models with heuristics.
  10. The world of markets never matches the ideal circumstances a model assumes. Whenever one uses a model, one should know exactly what has been assumed in its creation and, exactly what has been assumed in its creation and, equally important, exactly what has been swept out of view. A robust model allows a user to quantitatively adjust for those omissions.
    Mathematical modeling holds two benefits that are much more important than the direct results the model produces. First, building the model itself forces the creator to understand deeply the underlying problem. Modeling forces you to think. For me, everything that is important in finance is self-taught through the labor of model building.Second, the most important single element in all of finance is data. Good mathematical models tell you what data are important and why they are important. Many times I think of my models as merely diagnostic tools that exercise the data. When model results are questionable or inconsistent, I often find that data errors are responsible. As our team builds stronger and more complete mathematical models, we clean and improve the data.
  11. “Black Swans” is the term used to describe events with three characteristics: they are rare, they have extreme impact and they are rationalized retrospectively.
    The next question is whether the bailout was a good idea. It really comes down to Coke vs. water. If you are thirsty, you have choices. Coke tastes better and provides an immediate sugar rush and caffeinated stimulus, while quenching thirst. Water also quenches thirst, but it isn’t as stimulating. It purifies your body. It doesn’t make you fat and is much better for your long-term health.
  12. You don’t have to be able to solve new problems to understand how the old ones were solved. Explorers who chart new territory are rare, but lots of people know how to read a map.
    As the saying goes, “in God we trust, all others are subject to controls, limits, validation and backtesting.”
  13. “Creativity in business is often nothing more than making connections that everyone else has almost thought of. You don’t have to reinvent the wheel, just attach it to a new wagon.” – Mark McCormack, What They Don’t Teach You at Harvard Business School.
    All models by necessity distort reality in one way or another. A sculptor, when modeling in stone or clay, does not try to clone Nature; he highlights some things, ignores others, idealizes or abstracts some more, to achieve an effect. Different sculptors will seek different effects. Likewise, a scientist must necessarily pick and choose among various aspects of reality to incorporate into a model. An economist makes assumptions about how markets work, how these assumptions, considered alone, is absurd. There is a rich vein of jokes about economics and their assumptions. Take the old one about the engineer, the physicist and the economist. They find themselves shipwrecked on a desert island with nothing to eat but a sealed can of beans. How to get at them? The engineer proposes breaking the can open with a rock. The physicist suggests heating the can in the sun, until it bursts. The economist’s approach: “First, assume we have a can opener …”
  14. "Developing good models requires iterating many times on your initial ideas, up until the deadline; you can always improve your models further. Your final models will typically share little in common with the solutions you envisioned when first approaching the problem, because a-priori plans basically never survive confrontation with experimental reality." - Francois Chollet.
    No model, no matter how sophisticated, will be able to account for all possible environmental, circumstantial, and freak factors.
Card Set