What kind of moving average of the last x days is the best predictor of current and future happiness, and how does this relate to markets?

Anatoly Veltman writes: 

The widespread misuse of MAs concept is what gives it bad name. 90% of testers and users look at crossovers, and the remaining 10% look at break of MA from above or below. All wrong

The only proven way to apply MAs from trend-follower stand point is to look at nothing else but SLOPE. (Trading Days) Is 14-day MA sloping upward? If so, then is 30-day sloping upward? If so, then is 50-day sloping upward? If so: then Shorting is forbidden! Mirror test may save you from disastrous bottom-picking.

Bill Rafter writes: 

I beg to differ. There is no way the "average of the last x days is best predictor…" It by definition is at least a coincident indicator and more likely a lagging indicator. BTW the same can be said of the SLOPE of the last x days.

However, you can construct a leading indicator by comparison (difference or ratio) of the coincident to lagging indicators. For this newly created leading indicator, there tends to be a lot of false signals, due to random market action. To guard against that you need to have very smooth coincident and lagging inputs. Making them smooth also makes them more lagged, but that will not hurt you as you are not going to look at them outside of a difference or ratio, which will be quite forward-looking.

The real problem is that investors want to identify a static x. In doing so they are insisting that the market be modeled by x periods. Well, the market doesn't always feel like cooperating. At times the market may be properly modeled by x periods, and at other times by x+N, in which N can assume a wide range of positive and negative values. The solution is to first identify the exact period over which the market should be modeled for the coincident valuation. And then go on from there. Rinse, repeat.

Russ Sears writes: 

This would be a good question to ask the trading expert psychologist Dr. Brett.

It seems that with the same brain imagery he uses is being used in the study of the science of happiness.

While I am no expert I have read Rick Hanson, PhD book "Hardwiring Happiness"/ It has been awhile since I enjoyed this book, my summary of it is "focus on the life/good in the present. Placing things in context to how it has brought you to this moment, then enjoy the moment is enjoying life."

Presence seems to be the buzz-word in studies of contentment and psychology of success. Being aware of all your inputs, your feelings and recognizing them as part of life, then celebrate living. Presence gives you the fulfillment in your life needed to be loyal and disciplined enough for what is working well in your life. Thanksgiving is a day built on this idea, But presence also gives you the courage to turn things around, admit things are not as you want, and gives you Hope for the future. Happiness is more about living your life, being in control, then it is circumstances. Some of my happiest times have been after running hard for over 2 hours exhausted after 26.2 miles, cold and totally and dangerously spent but knowing I gave it my all.

So I would suggest that MA, trend following, momentum, acceleration, nor death spirals nor reversion to the mean, value investing should not ever be the "key to Rebecca", rather judge them in the context of everything else. Some days "the trend your friend" other days "the sun will come out tomorrow". 

Brett Steenbarger writes: 

It's a really interesting area of recent research. It turns out that happiness is only one component of overall well-being. What brings us positive feelings is not necessarily what leads to the greatest life satisfaction, fulfillment, and meaning. I suspect the market strategies that maximize short-term positive emotion have negative expected return, as in the case of those who jump aboard trends to reduce the fear of missing a market move.

Ralph Vince writes: 

Too many times in life I've found myself in darkened parking lots with a small gang of characters who intend me harm, and saw how the pieces would play out enough in advance enough to get out of it, or at least to realize there was only one, very unpalatable way out of it.

Shields up.

Too many times in life, I've had an angel whisper in my ear with only a few hours or seconds to spare to keep from being robbed blind by people I made the mistake of trusting.

Too many times in life I've paced in some anonymous hotel room, wondering "How the hell am I going to do this once the day comes?"

Too many margin calls have had to be met.

Far more times than I would care to, I've found myself confronted with the proposition of having to throw boxcars to survive, and I find myself, yet again, with that very proposition in a life and death context.

Only someone who really loves the rush of the markets, could enjoy wanting a given market to move in a specific direction. I've come to the conclusion it's far better for me to set up to profit from whatever direction things move in on a given day. Those that dont move in a manner so as to profit from this day, will tomorrow, or the next day, or the day after that… I need to just show up on time with my shoes on, collect on that which comes in today, sow the seeds today for taking profits on something at some future date. It's not difficult, and a lot more satisfying.

There's enough episodes in life we need boxcars to show up, and yeah, "Baby needs a new pair o'shoes."

Victor Niederhoffer writes: 

I like all these untested ideas about moving averages but my query was of a more general nature. What kind of moving average, perhaps its top onion skin an exponential average, is the best predictor of human happiness. I.e. if you are happy yesterday and unhappy the day before, are you happier or sadder. I mean vis a vis the pursuit of happiness, not markets, although the two are related I think.

Alexander Good writes: 

My answer would be a medium term moving average works best - about 6 months. We're naturally geared to notice acceleration not speed. After accelerating happiness, it's virtually certain to decelerate which we would have a heightened awareness of. Thus a 5 day moving average would have too much embedded acceleration and deceleration to yield a good outcome.

I would also say 6 months is a good number because there's a fear of 'topping out'. I.e. if you're at the peak happiness of the past 5 years you might get afraid of a larger mean reverting move. 6 months is short term enough not to be victim to noticeable accel/decel, but not too long to be subject to such existential thoughts that lead to unhappiness. 2 quarters is also a good timeframe for evaluation of back to back 3 month periods which seems like a relevant timeframe to most people professionally.

My meta question would be: does measuring one's happiness with a moving average make one more or less happy? 

Theo Brossard writes: 

I would pose that happiness would exhibit similar behavior with market volatility. Short-term clustering (which makes exponential average a good choice, if you are happy today chances are you will be happy tomorrow) and longer-term mean reversion (there must be some thresholds defined by values and time–you can't be very happy or unhappy for prolonged periods of time).

Jim Sogi writes: 

A good way to study this is to rate and record your happiness each day. Also record your acts: exercise, diet, work, family, vacation, tv, meditation, etc. Over time you can correlate the things you do that make you happy. You could correlate day to day swings as Chair queries in a univariate time series.





Speak your mind

8 Comments so far

  1. Ed on November 25, 2015 10:31 pm

    IF it is slope then over what period. In my view u don’t need an average. Magnitude and + or 1 for one period is enough with careful attention paid to survivorship bias on reversals from low periods.

  2. S K Rahman on November 26, 2015 2:15 am

    The Dominating influence in MA trading idea is underlying mean of a particular instrument. The returns will hover around what ever mean that instrument has exhibited in historical data. Assuming the large amount of data is selected (5 years?) for relatively stable mean, the interval selection should have trade off between number of opportunities vs whips (and pray that law of large number will hep). If instrument stays near MA dividing line, too many whips (Shorter Duration) and if it stays far away too often (200 day MA) less entry point.

    Alternatively a “correcter” % can be applied that can possibly minimize false positives (for any preferred MA interval) however this will change the format of trading to Long-Stop and Short-Stop mode

  3. drdimick on November 29, 2015 4:56 am

    To Average or Not To Average… That is the question

    A selection from wiki…

    “In colloquial language, an average is the sum of a list of numbers divided by the number of numbers in the list. In mathematics and statistics, this would be called the arithmetic mean. However, the word “average” may also refer to the median, mode, or other central or typical value. In statistics, these are all known as measures of central tendency.

    Signal averaging is a signal processing technique applied in the time domain, intended to increase the strength of a signal relative to noise that is obscuring it. By averaging a set of replicate measurements, the signal-to-noise ratio, S/N, will be increased, ideally in proportion to the square root of the number of measurements.

    In telecommunication, the term frequency averaging has the following meanings:

    The process by which the relative phases of precision clocks are compared for the purpose of defining a single time standard.
    A process in which network synchronization is achieved by use, at all nodes, of oscillators that adjust their frequencies to the average frequency of the digital bit streams received from connected nodes.
    In frequency averaging, all oscillators are assigned equal weight in determining the ultimate network frequency.

    In machine learning, particularly in the creation of artificial neural networks, ensemble averaging is the process of creating multiple models and combining them to produce a desired output, as opposed to creating just one model. Frequently an ensemble of models performs better than any individual model, because the various errors of the models “average out.”

    My research in the study of quantitative relativity indicates that averaging as a stochastic indicator is not relevant (to the physics of electronic exchange) because of predictive outcomes in (non)correlation to random variable(s). Granted, many if not most mathematical operatives within a given program trading system include some form of averaging… Hence Victor’s query.

    Nevertheless, an electronic exchange market so operates (and is subject to manipulation) by being deterministic. That fact appears to allow technical impressionists and algo-quant theorists to claim that central tendencies found during a given exchange may constitute predictive trends. Either possible or probably, no doubt about it… they do.

    A moving average does just so… quantify central tendencies displayed during an exchange of a given time period. Thus, to ask — what kind of moving average — is less so determined by variable denomination (x, for instance) than the underlying assumption (or rule) governing the average output. As a result, the exercise itself (ie., averaging) is nondeterministic… more so the random walk.

    Only when such a form of quantification determines the convergence/divergence of an outcome (or price) for entry/exit, does one say, ah-ha, herewith the output is a state relative to a system of coordinates; therefore, buy/sell.

    The rub? That said determination of price is or was relative to a period of an exchange process, but it DOES NOT constitute a trend, average or otherwise. It merely represents sequencing that is or is not directional indicative.

    What factor decides?

    That is the point here… The process of averaging is determinant not the input per se.

    To that end a la V’s question, my suggestion would be signal averaging… “By averaging a set of replicate measurements, the signal-to-noise ratio, S/N, will be increased, ideally in proportion to the square root of the number of measurements.”

    c^2, for example.


  4. Steve on November 30, 2015 6:42 pm

    What kind of moving average of the last x days is the best predictor of current and future happiness?

    There cannot be a quantitative or objective measure for something that is qualitative and subjective.

    The literal answer to your query would be, “it all depends.”

  5. Efthymios Karatzas on December 1, 2015 4:15 pm

    What kind of moving average could be able to predict our chaotic behavior?

  6. ED on December 2, 2015 5:16 pm

    One very sad day changes a person. You can have a happy day again but it is never the same - at times better it is better, many times it is worse.

  7. Patrick Raab on December 9, 2015 2:31 am


    The best predictor of my happiness is the moving average of the last x days of your happiness.

    Unfortunately, the majority just talk and talk of their own exponential average pursuits.

    And maybe this why I feel so up and down as I hear the limiting notes echo from their jazz flutes.

    I overcome this with my pursuit of happiness to learn about each and everything, everyone, someone and something.

    This gives me greater reference and clarity. The perception of my own change. Not just another flutist reacting to some odd or boring thing.

  8. Patrick Raab on December 9, 2015 8:29 am

    **Please post this version for my response**


    The best predictor of my happiness is the moving average of the last x days of your happiness.

    Unfortunately, the majority just talk and talk of their own exponential average pursuits.

    And maybe this why I feel so up and down as I hear the limiting notes echo from their jazz flutes.

    I overcome this with my pursuit of happiness to keep learning about each and everything, everyone, someone and something.

    This gives me greater reference and clarity. It is my moving average of x days that provides the perception of my own change.


Resources & Links