Any system that has 50 losses in a row is no system. It is theoretically possible to get 50 in a row with truly random samples which is not the case for a well tested and robust system.
After how many losses do you decide you are not following your system?
- Thread starter metatrader54
- Start date
If you get 50 losses in a raw on a 50% system.
Buy a lottery ticket because you’re a lucky guy.
As it should happen once every
The probability of that happening is 1 in a quadrillion, literally
To me a first principle of the market is that market regimes will change so you first have to be testing for regime change/structural breaks.
To me any run of 3+ loss needs to be scrutinized that we don't have faulty assumptions in the strategy. All these statistical concepts from ergodic stationary process mentioned above mostly don't apply.
A 12.5% chance of getting 3 heads in a row on a fair coin flip has basically nothing to do with a trading system. So much complete nonsense statistical ideas online from just not caring about the assumption of an IID process. Samples from a trading system will most likely have some dependency so any run is a much bigger deal than if you assume IID.
Obviously, the thing above about 50 losses in a row is preposterous. Anyone with half a brain would figure out well before 50 heads in a row that you are almost certainly flipping an unfair coin so assuming IID is stupid.
To me any run of 3+ loss needs to be scrutinized that we don't have faulty assumptions in the strategy. All these statistical concepts from ergodic stationary process mentioned above mostly don't apply.
A 12.5% chance of getting 3 heads in a row on a fair coin flip has basically nothing to do with a trading system. So much complete nonsense statistical ideas online from just not caring about the assumption of an IID process. Samples from a trading system will most likely have some dependency so any run is a much bigger deal than if you assume IID.
Obviously, the thing above about 50 losses in a row is preposterous. Anyone with half a brain would figure out well before 50 heads in a row that you are almost certainly flipping an unfair coin so assuming IID is stupid.
Last edited:
tip of the hat to a BlueWaterSailor post for this one...
"do your own research
do lots of testing at small scale until you're satisfied that it works for you.
understand the larger context in which these trades are supposed to be executed to be successful"
https://elitetrader.com/et/threads/...solid-trading-plan.340340/page-2#post-5014315
'
Last edited:
after the strategy works less than 50%. calculate always in precentage
After two losses in a row - cut your position sizing in half. When you can manage two winning days in a row - go back to your previous position size.
Works like a charm. Note how my first statement relates to trade size and my second statement relates to daily P&L.
If your trading system winners versus losers percentage over a protracted period of time is less than random chance (50/50) - scrap it.
Works like a charm. Note how my first statement relates to trade size and my second statement relates to daily P&L.
If your trading system winners versus losers percentage over a protracted period of time is less than random chance (50/50) - scrap it.