24

First, let's speak about perceptrons in general: their input $X_0$ is a $K$-dimensional vector. So if you want to use $(P_{bid}(t),P_{ask}(t), Q_{bid}(t),Q_{ask}(t))$, it would mean that without any effort (but later we will see that is would be better to do some efforts, as usual): $$X_0(t)=(P_{bid}(t),P_{ask}(t), Q_{bid}(t),Q_{ask}(t))'\in\mathbb{R}^4$$ ...


24

Two aspects of statistical learning are useful for trading 1. First the ones mentioned earlier: some statistical methods focused on working on live datasets. It means that you know you are observing only a sample of data and you want to extrapolate. You thus have to deal with in sample and out of sample issues, overfitting and so on... From this viewpoint, ...


8

Deutsche Bank's Quantitative Strategy (US) team put together the following piece on this topic (note: their research is available for clients, but I found that somebody uploaded the piece to a sketchy web site). In case the link dies, some of the academic papers they site are: Akbras, F., E. Kocatulum, and S. Sorescu, 2008, “Mispricing following public ...


8

The bias comes from the paper Stambaugh (1999) and has nothing to do with small sample bias. It has to do with point (1) below. The argument goes as follows: Typical lagged explanatory variables for stock-return regressions are correlated with contemporaneous stock returns This contemporaneous correlation biases forecasting regressions First review OLS ...


7

A cautionary tale on all these approaches it told by Tim Loughran and Bill MacDonald in the Journal of Finance, 2011 (When Is a Liability Not a Liability? Textual Analysis, Dictionaries, and 10-Ks, here). In their analysis they show that the commonly used Harvard Psychosociological Dictionary is inadequate for sentiment classification in a financial ...


7

Without seeing your trading desk's P&L it's impossible to say whether it is predictable or not. But here are a few thoughts - There's no reason to think that it isn't predictable. In general, financial time series are hardest to predict when the represent the return stream of an investible asset. A trading desk's P&L isn't really investible, so ...


7

Let me start with a simple example. Suppose you have a dividend strip that pays an unknown dividend $D_T$. The gross return (something like 1.05 and NOT 5%!) on this security is, by definition, $$R_{t\to T} = \frac{D_T}{P_t}$$ where $P_t$ is the current price of this security. If we use lowercase letters to denote logs (i.e., $\log D_T = d_T$ etc..) we can ...


6

People seem to think that using ML is going to circumvent the process of actually learning to trade, it doesn't. ML can be used to refine trading ideas, but it doesn't generate them, you need to use your brain for that.


6

Have you considered fitting ARIMA with exogenous regressors model? Linear regression with autocorrelated errors might be appropriate. R can do this with the arima() function via specifying the xreg argument.


6

I'm currently working on this task, to apply machine learning to stock trading. However, the concerns raised in other answers are major obstacles. So, I'm taking a different tact. My strategy is more akin to teaching a car to drive - the machine learning is not based on the underlying data, but rather on the driver's reaction to the data. So based on what ...


6

No I believe there is no directional predictive value derived from looking at divergences between futures and their underlying price value. The reason for divergences are of the no-arbitrage argument type. Futures could be arbitraged (and are immediately if such arbitrage opportunities surface, even those opportunities may only fill the stomach of a single ...


6

My favorite tool is Sornette's own Finanical Crisis Observatory: http://tasmania.ethz.ch/pubfco/fco.html If you are interested, I have developed my own tool in Java and JavaCL which can be found here: https://thebubbleindex.codeplex.com/ Update: Code moved to github: https://github.com/thebubbleindex/thebubbleindex


5

The mean could be the long run variance which is sig2 = fit.Constant/(1-fit.GARCH{1}-fit.ARCH{1}); I hope this explains. If not, note I ran this model through Matlab, I get different values. you can paste your m1 and m2 values and some other intermediate results so I can see why Matlab differs. EDIT: The question refers to forecasting the returns. ...


5

Sorry, but despite being used as a popular example in machine learning, no one has ever achieved a stock market prediction. It does not work for several reasons (check random walk by Fama and quite a bit of others, rational decision making fallacy, wrong assumptions ...), but the most compelling one is that if it would work, someone would be able to become ...


5

One possibility worth exploring is to use the support vector machine learning tool on the Metatrader 5 platform. Firstly, if you're not familiar with it, Metatrader 5 is a platform developed for users to implement algorithmic trading in forex and CFD markets (I'm not sure if the platform can be extended to stocks and other markets). It is typically used for ...


5

There is large literature on MIDAS (mixed-frequency data sampling) models, the leading scholars being Eric Ghysels and Rossen Valkanov — google their research for references. However, the motivation for these models has mostly been to forecast low-frequency stuff with high-frequency variables, updating, say, quarterly GDP predictions as weekly ...


4

Although not directly related to financial modeling, I've found the following quotation to be very instructive: "I remember my friend Johnny von Neumann used to say, 'with four parameters I can fit an elephant and with five I can make him wiggle his trunk.'" -- E. Fermi You may also read this: http://mahalanobis.twoday.net/stories/264091/


4

“Make things as simple as possible, but not simpler.” The problem you want to avoid is (near) multicollinearity. The tip-off will be that adding/removing a regressor will significantly change the coefficients on the other regressors. In practice (well, in the research that I read) I rarely see this explicitly tested. If you think that you have ...


4

Hopefully these ideas open up some solution strategies. A. Calibration approach: In the case of a volatility model such as Axioma's above, you could perform an instantaneous volatility adjustment. Procedure: You build your usual T+H volatility model. You measure the realized volatility and implied volatility of the training set. You measure the out-of-...


4

I can offer three suggestions: (a) Since any model, however sophisticated, will miss tail cases (such as Oct 2008) I would increase the number of high-frequency factors (eg weekly jobless claims - I don't know if that is a relevant example in your case - but just to give you an idea) in the model. Not only does that make the model more responsive to current ...


4

There are two excellent choices for implementing prediction markets: (1) Use book orders that stand until filled, just as intrade.com does. (2) Use an automated market maker (like Robin Hanson's) that stands ready to make trades. The book orders model is very simple to implement, but can suffer from very wide Bid/Ask spreads. And, it can be tough to bet ...


4

Van Belle describes a basic correction for autocorrelation in a t-test, although it may be hard to wedge it into the regression t-test. For the 1-sample t-test of the mean, the correction is to multiply the t-statistic by $\sqrt{\frac{1 - \rho}{1 + \rho}}$, where $\rho$ is the 1-period autocorrelation (or estimate thereof).


4

I think this one has a clear answer (I am solely talking about equities here): The change magnitude is much more predictable than the direction. The reason being that equity volatility is much more predictable than equity risk premiums. Volatility is nothing else but change magnitude and due to the stylized facts of volatility clustering together with mean ...


4

The graph you attached suggests that you were trying to find swings between major highs and lows. This can be done by simply finding local extrema in the price series. The concept is: find local extrema: minima in Low prices, maxima in High prices; find local extrema in the results, if swings are too short; repeat #2 until satisfied with the results. This ...


4

I have written an entire paper on this approach at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2828744 As to your specifics 1) "Volatility" as defined by variance does not exist, which is why it is changing. The first moment is undefined so the second cannot exist. See the paper as to why. Your fitted pdf will treat the outcomes as having a ...


4

Another way of staying "time-varying risk-premium", is saying that the risk-premium is predictable. However, that the fact that the risk-premium is predictable does not means that you can make money out of this. The best two references to understand this are: Cochrane (2008) - The dog that did not bark Goyal and Welch (2007) The first tells you what ...


4

The point of confusion may be in thinking that a predictable price process is synonymous with a mean-reverting process while using the definitions in these papers, it's actually the opposite! In the context of these papers, a random walk would be 100% predictable: the unpredictable component of a random walk (i.e. the period specific shock which has finite ...


4

There are a few exclusions that I have commonly seen: Excluding thinly traded stocks. The price that shows up in your data feed may not relate to actual tradable prices. Filtering for ADR/Pink locals. You can find stocks listed in multiple places in ways that would lead you to think that they are great for pairs trades when actually they are the same ...


3

The two components you refer to in your questions are: Market direction (the sign of the return) Change magnitude (the absolute value of the return) First, I'm sure you realize that neither of these are predictable at a 100%, otherwise there would be no way to make profit (you make profit by seeing things other didn't). To answer the question, I would say ...


3

There's no rule to answer this question for you. You need some combination of: Judgment: Are the parameters you're including reasonable? Sniff test: Is there theory to justify your parameter choices, or are you just hunting for chance associations? Hold-outs: You correctly mention that the problem is "in sample performance." The solution is therefore to ...


Only top voted, non community-wiki answers of a minimum length are eligible