probability basics in trading | Forex Factory

Options

Permalinks
Print Thread

Similar Threads

Introduction and trading basics 27 replies

Setting up the Basics for Trader101 style trading. 28 replies

FXTrader1979's Lessons on the Trading Basics 14 replies

Trading Basics: Concepts and Psychology 0 replies

Currency Trading: Understanding the Basics of Currency Trading 0 replies

probability basics in trading

  • #1
  • First Post: Sep 25, 2017 1:08am Sep 25, 2017 1:08am
I've been wondering.

Lets assume we have done sufficient statistical back testing (a decade of data) and discovered a mechanical system with a 60% win rate.

Monday we have a loss which was a 40% probability.
Tuesday we had another loss.. Is this still a 40% probability?

What is the probability of having 5 losses in a row (Monday to Friday)

-surely the probability of a loss gets smaller every consecutive loss day..?
  • #2
  • Sep 25, 2017 1:34am Sep 25, 2017 1:34am
A 60% win rate over a decade of data sounds pretty reasonable to estimate a trade's probability of working at a given point. However, you have to keep in mind what that entails.

That simply means that 60% of the trades over a decade of data were profitable. That doesn't mean that those trades were evenly distributed across those 10 years. Consider these extreme scenarios:

  1. All the profitable trades were done during the first 6 years of data, the last 4 years were pure losses.
  2. Say you took 1 trade a day. After 6 days of winning trades, the next 4 will be losses, the next 6 days are winners, the next 4 days are losers, etc.

The takeaway is that you don't know how exactly the wins/losses will manifest. Consider looking at additional different data sets. Such as: win-rate on a yearly basis, monthly basis, quarterly basis, weekly basis, daily basis, set-up basis, etc.

Quote
What is the probability of having 5 losses in a row (Monday to Friday)

0.40^5 = 0.01024 (1.0%)

Quote
-surely the probability of a loss gets smaller every consecutive loss day..?

Theoretically, yes. But like most models, the modeled behavior follows the model until it doesn't. Anything can happen in the market!
An eternal student.
  • #3
  • Sep 25, 2017 3:28am Sep 25, 2017 3:28am
If the distribution is purely random, than the probability of a given "next" loss does not depend on the results before.

The probability of loosing 5 trades in a row should work out to 0.4^5 =0.01024 or roughly one percent. That means on average every hundredth trade you will encounter a series of 5 losses.
  • #4
  • Edited Sep 26, 2017 2:50am Sep 25, 2017 11:18am | Edited Sep 26, 2017 2:50am
the probabilities are always 50/50.
  • #5
  • Sep 25, 2017 11:44am Sep 25, 2017 11:44am
the probabilities always 50-50...on academic work.

in trading, where the participants arent passive part, like a coin you toss, the real probabilities even if shorter period, but definitely not 50-50.
i commodities this 50-50 gets even more skewed, cos virtually zero price is non existent.
if oil cost 50$/barrel, the real life probability to go to 0$ or 100$ just isnt the same, even though academically it is 50-50...
there is always, always another trade!!
  • #6
  • Edited Sep 26, 2017 2:50am Sep 25, 2017 12:00pm | Edited Sep 26, 2017 2:50am
Quoting profitfarmer
the probabilities always 50-50...on academic work. in trading, where the participants arent passive part, like a coin you toss, the real probabilities even if shorter period, but definitely not 50-50. i commodities this 50-50 gets even more skewed, cos virtually zero price is non existent. if oil cost 50$/barrel, the real life probability to go to 0$ or 100$ just isnt the same, even though academically it is 50-50...
you are talking about the most likely not probabilities (common mistake)
  • #7
  • Sep 25, 2017 3:11pm Sep 25, 2017 3:11pm
If I had a special coin which was slightly heavier on the heads side, resulting in a 60% chance of landing tails side up (based off the statistics from 10 000 tosses).
The probabilities off tossing the coin 5 times and it landing heads side up every time is not 50/50. Obviously.
  • #8
  • Sep 25, 2017 4:06pm Sep 25, 2017 4:06pm
If you toss the coin in an empty room, with the light off, in complete darkness, the coin remains in a state of both heads and tails until someone turns on the light to observe it.
  • #9
  • Edited 11:11pm Sep 25, 2017 10:06pm | Edited 11:11pm
Quoting KoiFish
The probabilities off tossing the coin 5 times and it landing heads side up every time is not 50/50. Obviously.
Statistically. what I mean is that the actual probabilities are 50/50. right now not in the past

and if we count the transaction costs the probabilities in our favor are even less...

we need faith, faith, faith lol
  • #10
  • Sep 27, 2017 1:58am Sep 27, 2017 1:58am
It depends on your sample size over a decade. If you measured 10,000 trades, ~ 4-5 trades a day, then losing 5 in a row does not change your win/loss percentage. Your question is too hypothetical to answer.
You cannot be extraordinary by being normal
  • #11
  • Last Post: Sep 27, 2017 11:32am Sep 27, 2017 11:32am
Quoting BubaYaga
I've been wondering. Lets assume we have done sufficient statistical back testing (a decade of data) and discovered a mechanical system with a 60% win rate. Monday we have a loss which was a 40% probability. Tuesday we had another loss.. Is this still a 40% probability? What is the probability of having 5 losses in a row (Monday to Friday) -surely the probability of a loss gets smaller every consecutive loss day..?

Assuming that trading days are independent events, the chance of losing in Friday after losing from Monday to Thursday (providing that they are known outcomes) is still 40%. What you talk about is well-known misconception called "player mistake".


Probability of losing 5 days in a row is (2/5)^5=0.01024 i.e 1 percent (intersection of outcomes means multiplying odds)

Probability of starting to trade on Monday and getting first win in Thursday is (q^n)*P=0.4^3*0.6=0.0384 (geometrical distribution)

Cheers
0 traders viewing now