Elements of strategic trading guideline
In this article, I woud like to share with you a few elements of some thoughts on probability and trading that built up through my observations of the markets over the years. I would not delve into complex mathematical concepts or formulae here but simply refer to intuitive statistics and probability. It is not my goal to deliver definitive trading edges as I would merely explain some ways to ultimately elaborate some through technical analyses and possible implementation into complex algorithmic trading systems.
Before going further, I would like to make a very quick and simple recap about probability space and some parallels with real life and trading examples.
In a probability space, the probability that an outcome occurs is comprised between 0 and 1
The probability of an event that will necessarily occur is 1. An event that is impossible has a probability of 0 to happen.
Example 1 (Real life)
For instance, our lifespan is positive and this is obviously true. A negative lifespan is impossible and therefore its probability is 0.
Example 1 (Trading)
The price of a stock cannot be less than 0. This event is impossible, however a trading account can be negative if leverage is used and that a gap appears in a direction adverse to that of our position.
This first instance brings up a first important notion, that is “integration“, meaning that a trader must incorporate different elements in its scope that are outside of the limits of the initial intrinsic market (in this case, integration of leverage skewed potential risks and profits and ultimately created new possibilities that were not initially possible).
Exemple 2 (Real life)
In real life, we are mostly certain to “die” with a probability of 0,999…999…; and the limit tends toward 1 given enough time.
Exemple 2 (Trading)
Depending on the markets we are dealing with, if we do nothing over a long period of time, it is more likely that we lose everything even though potential gains are theoretically infinite (ie. end of financial markets, bankruptcies of brokers and/or banks …). We would not be able to benefit from it anyway since we would not be here anymore (see previous example).
Nonetheless, it is possible to manage dynamically our trading (profits/losses, positions …) and this is where the notion of “flexibility“ kicks in.
It is preferable to speak in terms of probability space when it comes to trading, because nothing is 100% sure in the markets, if not this claim itself.
We are never free from Black Swans and one of the main objectives of the trader is to find a trading edge in any market condition that would allow him or her to “play with chaos”.
STRATEGY AND PERFORMANCE
Repeating the same action over and over can lead to very different results that can be counted and used to create statistics.
What worked in the past is not bound to work again in the future. Past performance is not necessarily representative of future performance. This is also an issue when models that are overfitted are being backtested lead to results that cannot be positively expected all time in the future.
Out of sample data and time series must be analyzed, especially when other algorithms might eventually find out what our strategies are and integrate our model into theirs.
There are many questions of interest that can be tackled: how is for instance short term “noise” impacting longer time frames and the latter impacting those same moves that appear at first to be very chaotic? There are indeed some correlations and patterns that are difficult to detect and define but that seem to be existent.
I would like to address next a particular point that can drastically increase profits if used cautiously and correctly.
What is the probability that we perform a winning operation. Likewise, what are the odds that we recoup our losses or even that we generate some profit from this reverse position? Various vital questions must also be considered: should we be correct to change our mind, when should we secure our gains from that reverse position? Can we only partially recover from our losses or is it possible to breakeven? Different psychological impacts have to taken into consideration by artificial intelligence in scalping/intraday trading, including whipsaws and bull/bear traps.
We should admit and accept to be wrong and be flexible. We need to become systematically flexible because the market is always “right”. Hence the notion of “flexibility” and “systematicity“.
Example of adaptation to western technical analysis (use of trendlines, mathematical indicators …)
Supports and resistances: what is the probability that they are hit? How many “hits” have to appear so that the chance of a breakout increases?
We need to pay attention to the fact that the price doesn’t always stop at the same value. We should setup arbitrary zones of resistance and support (ex: percentage …).
In terms of probability, we can ask ourselves what are the odds that the price remains inside a range after 4 contacts with extreme band levels? 5 contact points? 6 contact points? On the contrary, how should we interpret that if we witness 10 hits at those key levels? Are we simply observing a singular event or should we completely questioning ourselves (we are maybe seeing things that “do not exist”)? One simple example would be to consider a security which is being sold at 101 then bought at 100, to be later sold again at 101.
The process is repeated 4 times. Someone who would be aware of the concept might believe there is a resistance and support at those specific levels or even better, a range. This could well be the case for a scalper, but this price action could be insignificant to a day trader who would not even consider that a pattern has been drawn. Depending on the situations (here, the time frame among other things), things can be perceived differently.
Therefore, it is crucial to focus on the notion of interpretation and to dynamically adjust our methods in order to increase our probability to succeed and/or to avoid to being stopped. In fine, even some tiny tweaks could greatly impart our P&L.
What is the probability that a moving average (ex: EMA20, 100 etc.) is touched over a defined period? Broken through for more than two days? A month? A year? Ten years? Many false penetrations and signals could indeed happen, and only the combination of different confirmations could correctly strengthen our opinion. A trader should also learn to recognize and use arbitrary yet popular elements of technical analysis (sets of patterns, indicators …). Would they increase the odds of “self-fulfilling prophecy”?
What is the probability that a fractal pattern appears over a given period?
Example of probability independence and conditioning
In order for a pullback or a throwback to materialize, is it a prerequisite that price pierces through resistance or support lines (as long as a resistance or support area has not been gone through because of a price increase/decrease or a gap, this probability is 0 because this could not possibly happen if the price below/above). This depicts that many events are conditioned by other ones (cf. Bayes’ Theorem).
What is also the probability that a technical pattern transforms into another one? For instance, that a triangle becomes a rectangle? Needless to say, for this situation to appear, the triangle must already exist in first hand.
Dynamism, agility and mastery of the basics and statistics are fundamental.
VALIDATION / INVALIDATION
A trader must bear in mind that invalidation is as important if not more important than validation itself in every kinds of settings (patterns, strategies…). One must think about the best options available: Do we have in this case to reverse our open positions? Should we simply close our positions and take our profits or losses? Should we wait and pray instead? Is the market clearly telling us that our position is wrong? Success rate has to be constantly assessed with flexibility and free from biases. We also face the possibility to be tricked and whipsawed so validation and invalidation depends on criteria that could be either systematic or arbitrary.
Time could also serve as a stop-loss. For instance, an intraday bullish flag that stretches too much over a long period of time could signal that the pattern and current targets are invalidated, or at least for the time being. Indeed, traders and algorithms often “keep in memory” targets that seem to be existent one of particular shorter time frame but that would only be “perfectly” reached much later (the following day, in a week, a month …). Analysis of this kind of delay is very important to remember as it could also add to trading edges.
A second target becomes valid id if and only if the first one has been hit. That being said, this is also dependent upon the strength of the moves: Was the objective attained fortuitously? Were the momentum and/or market conviction strong? Could be it be due to some kind of overreaction or simply to the triggering of stop-loss orders? Are the moves resembling to a bull or a bear trap?
Many breakouts could be false breakouts (cf. bull traps and bear traps).
Special case: volatility
When a key level (resistance, support line …) is being pierced downright with low volatility in certain markets, the probability that we head further in the direction of the “breakout” gets larger in the given time frame and period of time. On the other hand, some analysts advocate that large volatility is preferable to clear a particular level.
What is the probability that our trading account could survive over a certain period if a given amount of leverage is used (ex: full leverage, under leverage)? What are our odds to become a millionaire if we start with a account with 30 bucks? Billionaire?
How fluctuating is our success rate and how can we improve it depending on our trading style (very short term, short term, midterm, long term). We need to involve a set of different factors such as psychological effects, position size, leverage …).
It is vital to identify our profile across different market conditions and time horizons. We can also get sort the data and get the skewness (measure of symmetry or rather the asymmetry) and kurtosis (heavy-tailed or light-tailed in comparison with a normal distribution).
We need to understand whether we are performing in a consistent manner or in an erratic and instable way (small gains, small losses? Small gains, large losses? Large gains, small losses? Large gains, small losses?) in order to determine our odds to reach our goals over the long haul.
We could think of “random walks” that wouldn’t in fact be that random on the grounds of self-fulfilling events that would be carried out once a certain set of conditions are met.
Mean reversion: Many statisticians agree that the best predictor of trading price is holds to price average. However, several markets do not reach their previous price level from where they are deemed to be overvalued before a very long period of time.
Probability and market biases: Declines in price are in general faster and more violent compared to price increases. Bubbles also show several statistics such as increased steepening. Yet, this is sometimes what others try to make us believe like in a game of poker, prompting many to open an inverse position (cf. contrarian strategies). Some arbitragists are knowingly surfing this wave on purpose and inflating further bubbles. It is also a strategy used by some professional traders who are attempting to trade what they believe to be excess momentum and buy “overbought” products instead of short-selling them.
What we know for sure is that trees do not grow to the sky. It is also possible that we have to deal with all kinds of cycles even though we cannot not how long they can last. How many consecutive ups and downs can we have? This would certainly depend on different elements, such as the product itself, its history and different cycles. What is the probability that a new game of poker is started and that new “normals” are emerging?
What is the probability that a stop-loss is hit depending on our position or even that a CFD broker traps a CFD trader by playing against him or her (ex: B route)?
What are the different factors that impact our probability to become euphoric, petrified, optimist, pessimist or angry?
Long? Short? Flat? We all possess this bias and it is quite interesting to know the probability that it would change. Equity markets are structurally bullish and should “normally” keep growing for a long time. In times of panic, momentum is naturally much higher. If we do nothing, we only waste time and maybe some opportunities, but the best trade could well be to not pull the trigger at all and wait.
Under which market conditions should a scalper become a day trader or hold his or her positions in swing trading or investing mode (see volatility)? Some funds specialized in algorithmic scalping might not also have the leisure to be that flexible owing to their systematic strategies unless they integrate more dynamism and adaptability in their algorithms (ie: Artificial Intelligence).
Depending on personal objectives and financial capacities, it could be preferable to make long-term investments or speculations with some or all of the available capital when some particular market conditions are being met.
What is the likelihood that a certain price, if reached, would attract investors or speculators? We need to take into account the speed and momentum with which we reached those targets, along with indicators and other elements in which a significant number of actors believe. We can also rely on the alignment of planets (we shouldn’t exclude spurious factors that could be randomly correlated to the result we desire to achieve).
RISK AND UNCERTAINTY
A trade should generally be jumped in when there is a high probability trade setup. We cannot be sure to always win over a certain period of time and a certain number of trades.
We might as well possess a solid strategy but not be able to be filled at the price we want due to lag time and our position in the queue of an order book. Slippage can also frequently occur in particular when market orders are thrown in illiquid markets or during periods of price shocks (ex: well defined stop-loss orders are set but no buyer is willing to take our position in case of a sharp decline when we are long the market). We also need to pay attention to busted trades and non-reviewable trading ranges.
Obviously, there are many other exogenous and endogenous issues that have to be dealt with. “Fat fingers” could appear, Iceberg orders must be accounted for (implementation of strategies using VWAP – Volume Weighted Average Price) and so on.
What is the probabiliy that a trader in possession of a confidential information performs insider trading? What are also the odds that a broker would do front-running or that a high-frequency trader use illegal spoofing techniques and that her or she gets caught by the SEC or any similar entity?
You’ll find below a summary diagram of the different components to be aware of when it comes to probability in trading. Please note that this outline is merely based on personal experience and observations and is certainly non-comprehensive. I simply hope that this guideline will provide some food for thoughts (click to enlarge):
Nothing is never certain in a game of poker. As Warren Buffett once apparently said, “If after ten minutes at the poker table you do not know who the patsy is – you are the patsy”. However, like in a game of chess (or even better, a game of Go), we can analyze the different possible moves from our opponents ranging from the most probable to the least probable ones. That way, we can become prepared and able to react properly according to the scenario that is unfolding in front of us and the constraints that we have.
Therefore, we must take into consideration several dimensions and dynamics that keep changing incessantly in real time. Black Swans are also very important and interesting to analyze. Through repeated observation of price action and price reactions to different possible situations, it is likely that we detect numerous elements that are not bound to randomness or coincidences.
If I had to sum up everything in a single sentence, it would be “Once in position, a trader should become mostly a risk manager.”