Some Basic Observations when Trading

When you get a new toy, the temptation is always to "turn all the knobs up to 10" or even a la Spinal Tap to 11 (- "it's one louder ..."); and StockWave seems like a nice toy, a real change from the monotony of the usual charting packages and technical analysis programs, and we all like money, are eager to earn some profits, to make it pay, so why not just algorithmically-throw-the-kitchen-sink at some stock and pile in ... ?

Well, for one good reason - just because StockWave has "A B C and X Y Z", doesn't mean it absolves you from the responsibility of thinking about what you are doing, far from it - one thing about computers and this is the real problem about creating a general purpose AI system - is that they don't get "context" - it takes a human being to step back and ask "does this make sense?" or "am I happy about this?" or to say "something doesn't smell right here!"; StockWave serves up a set of analyses to help the human trader make a decision about whether to trade or not - it does not take the decision for you, and it would be dangerous if it did - that would be asking for trouble (- Knight Capital, anyone?)

There is a saying in the martial arts, something along the lines of "one technique mastered is worth a thousand sampled" or more bluntly "do the simplest thing that works!" - trading is just a game of "find the lady" of guessing "up, down or sideways" a wee bit better than random chance, ... then you can make money, real money.

The simplest technique in StockWave is our basic random walk or "stochastic analyzer" - it makes the fewest assumptions possible - we don't, or even can't know anything about this stock apart from the share price, which is a "stochastic process", it has no memory and is not affected by external events; all we can know is the distribution of movements, which we measure. It is a very conservative, very safe set of initial assumptions, i.e. "we know nothing"!

You create one of these and the result gets plotted on the chart as a heatmap, with contours of probability overlaid; it looks nice and tells you something, but at first look one might say "it doesn't tell me where the stock is going to go - so how do I trade this?" or "even if I work out a way of trading this information, aren't the market makers and all those other people using models similar to this anyway, so what trading edge do I have ...?" - these are good questions, and you might not expect it to be useful, just another piece of pointless eye-candy, but remember, this is your model - you chose the input data and the model parameters, you can see it on the chart, you can vary the parameters to check the model is robust - it did not come from some "black box" pricing calculator, or a "fair value" price from some website - you should be able to trust it; in trading, this is not a trivial thing.

Some anecdotal observations which may help you (and just using the basic random walk) -

  • "The trend is your friend" is a ropey old trading wisdom "truism"; trend following works so we are told, er, ... except when it doesn't! - when it all goes into reverse and you start getting killed! So when do you decide the trend is over and a new trend has begun? Yeah, there's a lot of "wisdom" that doesn't bear examining too closely, but ... there is something to this, but not a lot. To clarify, if you are looking one month ahead for say an options trade and you use the last 4 months price data as an input, then it is more likely than not that the trend over the input period will persist ... thus, if you are betting against the trend, you really need to be confident, and have good reasons for it.
  • But you aren't trading the trend anyway, you are trading the price! "So what"? if the "trend" persisted? - the actual price bounced around like a rubber ball in a concrete car park! But ... here is something for the swing trader - the 50 and 66% isocontours often act like "barriers" to the bouncing share price - as they approach the contours, they will soon reverse; this is like a more sophisticated version of the old ideas of "support" and "resistance".
  • With a probability model you are more likely to be trading options and not stocks or CFDs or spreadbets, which are simply directional bets - options trades, especially combinations can allow you to make money from any type of information, they are thus most flexible. With a probability, start thinking options.
  • However, crafting options trade combinations by hand is a pointless, needle-in-a-haystack style exercise.
  • Your heatmap, your probability model will not be quite the same as those of the market makers and other players, but it won't be massively different either, so your chances of finding profitable trade combinations, which are the result of "mispricing" are likely to be small; remember options trading is a zero-sum game where you have to beat the other guy - he has to lose, for you to win, i.e. your knowledge, your prediction, your model must be better than his.
  • Having said all that, automatically and systematically searching for trading combinations will allow you to find viable trades; the computational resources to do this kind of search are large, and so not that many people are doing it; thus you may find yourself "panning for gold" in a relatively quiet environment.
  • Trading costs, spreads and fees, are really important - they constitute a drag on your trading, holding you back; costs also eliminate a great many high probability, small profit trades that the small investor cannot access. Especially make sure that multiple leg combinations trades are charged efficiently - i.e. as one single atomic trade, not for each separate leg - that will kill you!
  • Starting out with the random walk analyzer allows you to become comfortable with its strengths and weaknesses, in particular, when you need something better (- and what you want out of the "something better"), and when-and-why the folk-wisdom above stops working, e.g. the "bouncing at the 50 contour" fails dramatically from time to time, the price crashing straight through to a new level - this is almost always down to an external news event ... so now you know why StockWave has a huge and unique machinery for doing news analysis!
  • You soon realise and understand that you could trade a lot better, and lot heavier, with higher leverage and less risk, if your probability contours were tighter, or you could see sub-structure in the heatmap ... and this leads us to the advanced analyzers, those "random walks with machine learning". Note that it is not good enough to know some technique is available, you need to know why you need it and when to use it! Furthermore when you start using the advanced analyzers do not resort to the old "turn all the knobs up to 10" mentality - explore their parameter spaces, find stable, corroborated results.
  • Get to know the options trade combinations like they were your own children - know them by sight and get an instinctive understanding of their character; when you see the payoff profiles, remember green is good and red is bad; a horizontal line means a capped loss(red) or a safe, stable profit(green) while diagonals on the sides indicate unlimited profit or loss. Red triangles thus should signal danger to any trader. A capped loss should be something bearable to the trader and likewise the capped profit should be worth the effort; be wary of sharp transitional regions where you can slip quickly from profit into loss - these need a lot of watching.
  • Profitability is really about trade management - obviously timing when to get in and get out at the right time makes all the difference in the universe! - but what I really mean is when you are watching an open trade to take your profit or loss in a sensible fashion; for most people this will mean resorting to mechanical rules in order to force themselves to be decisive, to free themselves from the emotional anxiety of excessive greed or pointless hope.
  • Despite the previous item, be cautious about using brokerages own "stop loss" or "exit plan" features - they can be "twitchy"; some brokerages also have a tendency to intervene in open trades, closing positions for arbitrary reason.

Advanced observations to follow, experimentally yours.

A scary story, to frighten the children.