Trade dependency describes a sequence where one trade depends on the result of the previous trade, to some extent. For instance, if you have a win, the next trade might be more or less likely to win as well. If a win is more likely following a win, and a loss on a trade is more likely following a loss, then the trades exhibit positive dependency. If a win is more likely following a loss, and a loss following a win, then that is called negative trade dependency.
The first question must be how do you know this, and the answer is that you have to test existing data and look for patterns. Sometimes there is a bias, but only for a certain number of wins, after which the dependency swings around. The topic can get complicated.
As always, statistical analysis is used to determine how significant your back tested results are, and whether they show a real dependency or simply a fluke. It is not necessary to go into the statistical details here, as there are numerous references and formula available should you wish to research the topic. It is generally considered significant if there is a ‘degree of confidence’ in the dependency that is greater than 95%. Because these are statistics, nothing is certain, but that 95% suggests it is very likely that the dependency exists.
You may have heard of the Martingale method, which involves increasing your stake after a loss, and reducing it when you win. This is popular with gamblers, using techniques such as “doubling down”; though with a profitable trading system the opposite, the so-called “anti-martingale method” is usually better. When you use trade dependency, taking advantage of patterns of wins and losses, it is similar in idea to applying the Martingale method, changing your stake on the basis of history.
You may be wondering how specifically you can use this information, if you find a trade dependency. It depends what sort of dependency you find, and sometimes the dependency only lasts for a certain number of trades, for instance it may be that the average winning run is three trades, so you would expect, with 95% confidence, that the fourth trade would lose.
This suggests one way to exploit the trade dependency. Why would you take the fourth trade if you reasonably expect that it will lose? In general, you can skip indicated trades where the dependency suggests that they will lose, and by keeping a count take up the next trade that should be a winner.
Similarly, if it seems that wins follow winners, and losses come after losers, then you should skip trades after a loss until the sequence ceases and the trade you would have taken becomes a winner. An alternative approach is to know that, for instance, the losing streak tends to be four trades long, and simply place a trade on the fifth choice.
It may be that the trade dependency is only short-lived, and that the dependency switches from positive to negative dependency and back again over time. This can again be analyzed and used to make your trading more successful.