In my (endless) series on Markets, I came up with the "gambler's fallacy" retort :
for those that don't know, gamblers often think that after a serie of black or white, the chances of the output switching color is higher ("the ball ended 20 times on red, so it will necessarily fall on black very soon")
but actually statistics say that it doesn't matter how many times the same color gets out, there is still 50/50 chances of any color for the next output.
Now, the question is : is it also the case when you have a system that has a positive expectancy ?
like for instance 60% winrate ? is the 'gambler's fallacy" then also true ?
or do you have more chance to have a winning trade after a couple of bad ones ?
an example I could make is : let's say we got a system, and in the backtest, over 28 pairs and 10 years, we ended up with a max consec losses of only 3.
now if we monitor trades virtually and we wait for 3 losses in a row, we can almost be certain that the next one will be a winning one, am i wrong to think that ?
what do you think?
Jeff
Re: Gambler's fallacy
2ionone wrote: Sun Oct 16, 2022 11:54 pm In my (endless) series on Markets, I came up with the "gambler's fallacy" retort :
for those that don't know, gamblers often think that after a serie of black or white, the chances of the output switching color is higher ("the ball ended 20 times on red, so it will necessarily fall on black very soon")
but actually statistics say that it doesn't matter how many times the same color gets out, there is still 50/50 chances of any color for the next output.
Now, the question is : is it also the case when you have a system that has a positive expectancy ?
like for instance 60% winrate ? is the 'gambler's fallacy" then also true ?
or do you have more chance to have a winning trade after a couple of bad ones ?
an example I could make is : let's say we got a system, and in the backtest, over 28 pairs and 10 years, we ended up with a max consec losses of only 3.
now if we monitor trades virtually and we wait for 3 losses in a row, we can almost be certain that the next one will be a winning one, am i wrong to think that ?
what do you think?
Jeff
Hi Jeff,
hmmm...60% winrate
100 trades
40 loss streak then
60 win streak
still makes it a 60% win rate..but can you afford 40 losses before finally winning?
I would think its more than percentages at the end of the day.
Re: Gambler's fallacy
3I think the answer is as simple as one would expect: yes, the next trade also has a 50/50 outcome, despite the fact that you have 600 winners and 400 losers in a 60/40 situation in the long run (what traders tend to call an edge).ionone wrote: Sun Oct 16, 2022 11:54 pm Now, the question is : is it also the case when you have a system that has a positive expectancy ?
like for instance 60% winrate ? is the 'gambler's fallacy" then also true ?
or do you have more chance to have a winning trade after a couple of bad ones ?
[Although, of course, one could argue that trading isn't really gambling etc and that the gambler's fallacy doesn't really apply]
The distribution needs a high number of cases to work itself out statistically. For the Gaussian distribution (for example) you should have at least 1000.
Re: Gambler's fallacy
4Not on point but related.ionone wrote: Sun Oct 16, 2022 11:54 pm
now if we monitor trades virtually and we wait for 3 losses in a row, we can almost be certain that the next one will be a winning one, am i wrong to think that ?
what do you think?
Jeff
Re: Gambler's fallacy
5Thought of this post when reading this from 'The Big Short' Michael Lewis.ionone wrote: Sun Oct 16, 2022 11:54 pm In my (endless) series on Markets, I came up with the "gambler's fallacy" retort :
for those that don't know, gamblers often think that after a serie of black or white, the chances of the output switching color is higher ("the ball ended 20 times on red, so it will necessarily fall on black very soon")
but actually statistics say that it doesn't matter how many times the same color gets out, there is still 50/50 chances of any color for the next output.
Now, the question is : is it also the case when you have a system that has a positive expectancy ?
Jeff
''Above the roulette tables, screens listed the results of the most recent 20 spins of the wheel, marvel at the improbability, and feel in their bones that the tiny silver ball was now more likely to land on red. That was the reason the casino bothered to list the wheel's most recent spins: to help gamblers delude themselves. To give people the false confidence they needed to lay their chips on a roulette table.''
(reminder, the casino has the edge)