Analysis on candles

N

nooby_mcnoob

I saw someone doing this online so I figured I'd test out my use of a desktop wiki on it and thought it was interesting enough to post here if anyone else is doing the same analysis and had the same result (that there is no edge).

P(crossed line2, given crossed line1) - median ~68%

We know that there is a pretty large variance between currencies (p=0.65..0.75) for the probability with the highest median which is P(crossed candle body midpoint | crossed candle body bottom). That is:

upload_2019-7-3_11-11-42.png


This means that ~30% of the time, you will have price cross the previous day's candle body bottom, and of those times, 68% of the time you will cross the midpoint. Additionally, there is a 50% chance that when price crosses the previous day's candle body bottom, that it crosses the candle body top.

So what we want to know is: on the days where we cross the previous day's candle body bottom, what is the distance between the previous day's candle body bottom and the candle body midpoint, and is there enough to make a profit?

In fact, the median return on those days is 0.07% which translates to an expected value of 0.04% given the above probabilities. Interactive Broker's commission per side is 0.01% if you trade less than $1B/month. Assuming 0 slippage, that is an expected profit of 0.02%.

So are there enough trades to make this worthwhile? In 3 years, it looks like the number of trades across all currencies that fall into this category would be ~300. With an expected profit of 0.02%, this works out to a (1+0.02%)^300 = 6% return over 3 years.

Result: No edge.
 
Last edited:
nooby_mcnoob said:
I saw someone doing this online so I figured I'd test out my use of a desktop wiki on it and thought it was interesting enough to post here if anyone else is doing the same analysis and had the same result (that there is no edge).

P(crossed line2, given crossed line1) - median ~68%

We know that there is a pretty large variance between currencies (p=0.65..0.75) for the probability with the highest median which is P(crossed candle body midpoint | crossed candle body bottom). That is:

View attachment 205178

This means that ~30% of the time, you will have price cross the previous day's candle body bottom, and of those times, 68% of the time you will cross the midpoint. Additionally, there is a 50% chance that when price crosses the previous day's candle body bottom, that it crosses the candle body top.

So what we want to know is: on the days where we cross the previous day's candle body bottom, what is the distance between the previous day's candle body bottom and the candle body midpoint, and is there enough to make a profit?

In fact, the median return on those days is 0.07% which translates to an expected value of 0.04% given the above probabilities. Interactive Broker's commission per side is 0.01% if you trade less than $1B/month. Assuming 0 slippage, that is an expected profit of 0.02%.

So are there enough trades to make this worthwhile? In 3 years, it looks like the number of trades across all currencies that fall into this category would be ~300. With an expected profit of 0.02%, this works out to a (1+0.02%)^300 = 6% return over 3 years.

Result: No edge.
More...

the same is true for any kind of technical analysis without looking at the background context... nice analysis by the way..

that's why I always say trading is about unfolding stories... play out a narrative as my pro boys play it then life is good lol.
 
dozu888 said:
the same is true for any kind of technical analysis without looking at the background context... nice analysis by the way..

that's why I always say trading is about unfolding stories... play out a narrative as my pro boys play it then life is good lol.
More...

Hey that means a lot to me coming from you!
 
Yup, curiously most price pattern stuff I backtested (in cases it didn't appear just random or otherwise unprofitable) yielded a really small positive return like 4-6% p.a, i.e. probably just below where professionals don't care about it. Amplifying it with leverage isn't worthwhile since Sharpe ratio is abysmal too and you just end up losing too much in down turns.

Nowadays I basically consider price without additional data/context to be noise. I do think most people who claim to trade pure PA actually have a lot of other context built in their strategy that help them.
 
nooby_mcnoob said:
Hey that means a lot to me coming from you!
More...

ha thx... good luck to you :)

Snuskpelle said:
Yup, curiously most price pattern stuff I backtested (in cases it didn't appear just random or otherwise unprofitable) yielded a really small positive return like 4-6% p.a, i.e. probably just below where professionals don't care about it. Amplifying it with leverage isn't worthwhile since Sharpe ratio is abysmal too and you just end up losing too much in down turns.

Nowadays I basically consider price without additional data/context to be noise. I do think most people who claim to trade pure PA actually have a lot of other context built in their strategy that help them.
More...

I'd even argue the price action itself is not that important in the sense that looking at a line chart probably gives just as much information as the candles... investor sentiment and narratives are the game.
 
Snuskpelle said:
...Nowadays I basically consider price without additional data/context to be noise. I do think most people who claim to trade pure PA actually have a lot of other context built in their strategy that help them.
More...

That's been my argument for +10 years whenever I meet someone that states they're a profitable pure TA or PA trader to imply they're using only TA/PA and nothing else.

They're using something else that they do not discuss (intentional or unintentional) in combo with the TA or PA.

wrbtrader
 
dozu888 said:
ha thx... good luck to you :)



I'd even argue the price action itself is not that important in the sense that looking at a line chart probably gives just as much information as the candles... investor sentiment and narratives are the game.
More...
Can you clarify what you mean by narrative in this context? Any examples?
 
You must log in or register to reply here.
Back
Top