General Topics
Technical Topics
Brokerage Firms
Community Lounge
Site Support

# statistics, standard deviation

Discussion in 'Trading' started by Gordon Gekko, Jun 10, 2003.

1. ### Gordon Gekko

i am not a statistics expert at all, but statistics interest me...

there is something i'm thinking about....

say i have 10 numbers, but 2 are way different than the rest. for example, say 8 of the numbers are less than 4, but 2 are above 8:

2
3
11
2
3
3
2
1
9
1

what if i don't want to account for the 2 extreme numbers? or what if in another set of 10 numbers, the same thing happens, but 3 numbers are above 8 instead of 2?

this is what i'm getting at....

i suppose i could just eliminate the biggest 2 or 3 numbers every time. but then i was thinking if i could use standard deviation to help me here....

anyone know what i'm saying?!?

2. ### EricP

Some people would 'throw out' all data that is more than 3 standard deviations from the mean (i.e. average). Statistically, these are known as 'outliers' and may or may not be valid for analysis purposes. Beware, though, if your numbers are trade profits and losses, though occasional large profits and losses may not be 'outliers' and instead may reflect the normal expectation for occasional extreme moves (overnight gaps, etc.).

Good luck,
-Eric

3. ### TheCaracal

Best trade of the year was SS MMM @ 137, was almost a 3.0 STDV, meaning spends 2% of the time at that price relative to mean price past 20 days.

best,
david

On the open?

nitro

5. ### TheCaracal

Yes by 9:50. Sold off 4 points intraday, then traded down to 120.5, subsequently, which was a 2.0 STDV entry long side, went to 124 even in about 11 minutes!

Best,
david

6. ### Gordon Gekko

can someone tell me if either of these 2 methods is a valid way of doing things?

i put these 10 numbers into excel:

2
3
11
2
3
3
2
1
9
1

the average is 3.7

the standard deviation, using STDEVP in excel is 3.257299

now, if i wanted to get rid of those 2 big numbers, do any of these 2 methods make sense?

1) throw out numbers greater than 3.7 + 3.257299

2) throw out numbers greater than (3.257299)*2

7. ### Ken_DTU

as a former statistician (really!), for Ford, one thing I love is numbers, and data analysis, pattern distributions...

in looking at trading, I'm always looking For the outliers, in mfg terms we'd say it's "out of control" vs "in control" situations... eg buyers overcoming sellers, the market out of equilibrium, (eg trin far away from 1 as possible) which is what we want..

so being able to see the data plots, and for example, see which sectors are moving strongest on the open, accompanied by a TRIN of >1.5 (for shorts) or <.7 (for longs), is a daily approach I use, along w/2d high/low price breakouts, during the open ..

understanding std deviations, eg you want to find trading opportunities that are Not in the 1 sd 68% (chop) range, if you think of it in those terms ... best to look for situations as a trader, where buyers or sellers are clearly in control, and an imbalance exists..

having a background in spending years working to improve mfg process control with the use of charts and data, has been very helpful to me in trading, fwiw ..

stat books are Real dry tho lol ..

ken