Перейти к содержимому

- - - - -

Why Emotions Affect Even Computerized Trading

Написано EneaDoku , 07 Март 2017 · 579 Просмотров

Опрос: Test Yourself Before you read on (2 пользователей проголосовало)

What better describes you methodology?

  1. Systematic (1 голосов [50%])

    Процент голосов: 50%

  2. Discretionary (1 голосов [50%])

    Процент голосов: 50%

Голосовать Гости не могут голосовать
Interestingly, the need for emotional comfort will even have a detrimental impact on systematic trading (i.e., computerized, rule-driven trading), an area of trading one might reasonably have assumed would be free of emotionally based decisions. Typically, when people approach systematic trading, they will test their system rules and then discover that there are many past instances when following the system rules would have led to uncomfortably large equity drawdowns—an observation that will be true even if the system is profitable over the long run. The natural instinct is to revise the system rules or add additional rules that mitigate these poorly performing past periods. This process can be repeated multiple times, making the simulated equity curve smoother and smoother with each iteration. In effect, the natural inclination is to optimize system rules for past price behavior. The resulting final optimized system will generate an equity curve that looks like a money machine. Such a highly optimized system will be much more comfortable to trade because, after all, look how well it would have done in the past.

Beware of testing too many hypotheses; the more you torture the data, the more
likely they are to confess, but confession obtained under duress may not be admissible
in the court of scientific opinion
Stephen M. Stigler, "Testing Hypothesis or Fitting Models? Another Look at Mass Extinctions, "
in Matthew H. Nitecki and Antoni Hoffman, eds., Neutral Models in Biology, OxfordUniversity Press, Oxford, 1987, p. 148.

The irony, however, is that the more a system has been optimized to improve its past performance, the less likely it is to perform well in the future. The rub is that the system’s impressive simulated results are achieved with the hindsight knowledge of past prices. Future prices will be different, so the more the system rules are tweaked to fit historical prices, the less likely the system will work on future prices. Once again, the human instinct to seek emotional comfort has negative consequences in trading—even in computerized trading

2 Questions for you.
Answer in the comment section below, remember you can learn by reading but you will learn more by interacting with other people. I will make sure to reply to all of them.
  • What are some of your own experiences in parameter optimisation?
  • Do you know what curve-fitting is?

  • 0

What are some of your own experiences in parameter optimisation?



Making optimisation by some parameters.


Hard optimisation will create system, which  very well at short period, but awful at long term. But there is possibility, that profit at this period be enought to making medium result at long term profitable.  In this way optimised most of parameters, or even all.


Other possibility - crude optimisation. This method have 2 goals : making long term results profitable, and get profit often. For this type - optimised some parameters.


I know 2 ways for achive acceptable crude optimisation:


1. Making tests at previous long term history. Method of "blind search"


2. Classified all marked moves. Then, optimise system for profit result at half of them or most of them.



Do you know what curve-fitting is?



It is situation, when results of optimisations  dont appear in future. It can be hard optimisation or incorrect testing.

    • 1

Trackbacks для записи [ Trackback URL ]

Для данной записи нет trackbacks.

Последние записи

Последние комментарии