Australian (ASX) Stock Market Forum

# Van Tharp: I need an explanation please thanks for the help

#### ilovemycat

i have read his book super trader over and over again but im really having a hard time understanding how does he calculate the numbers in the book

after looking at this table several times

1 \$678 0.86R

2 \$3,456 4.40R

3 (\$567) –0.72R

4 \$342 0.44R

5 \$1,234 1.57R

6 \$888 1.13R

7 (\$1,333) –1.70R

8 (\$454) –0.58R
page 159 of super trader original edition

and this paragraph

The expectancy tells you that on the average you’ll make

The standard deviation tells you how much variability you

can expect from your system’s performance. In the sample our

standard deviation was 1.86R. Typically, you can determine the

quality of your system by the ratio of the expectancy to the

standard deviation. In our small sample the ratio is 0.36, which

is excellent. After 100 or so trades, I’d expect this ratio to be

much smaller; however, if it remains above 0.25, we have an

acceptable system.

i managed to calculate the expectancy which gave me
MEAN R MULTIPLE/ EXPECTANCY = 0.675R

after looking online for a standard deviation calculator it gave me
1.8568329411739
however no matter how hard i tried i wasnt able to calculate the
ratio is 0.36 i would kindly ask you if you explain how did you arrive to this conclusion
i looked online and found this formula
SQN = ((Expectancy / Std Dev "R") * Square root of number of trades)
however when i did the math the result i got was
1.0340488626
thats very different from the 0.36 you got
im mainly trying to see if my system is a holy grail system
but i dont know how to calculate this

Ratio of Expectancy
to Standard Deviation of R / Quality of the System
0.20–0.24 Average
0.25–0.29 Good
0.30–0.50 Excellent
0.50–0.69 Superb
0.70 or better Holy Grail

so mainly what im trying to do is a system when i win 30% of the times but has a high risk reward ratio 1:3 .... the ratio does not change .... this is a hypothetical system that does not exist,... hence all the data giving out is purely theoretical

mean r multiple = 0.2 (based on me winning 3 trades 9r losing 7 trades -7r)
standard deviation= 1.9321835661586 based on and the following numbers= 3,3,3, -1,-1,-1,-1,-1,-1,-1

using a standard cdeviation calculator found online
the reason why i want such a small winrate is because im a noob when trading ... like fundamental and technical analisis are not my thing hence i need a low winrate than can still make me money ... as time goes by i do expect my winrate to increase but for now i wanna keep it as is ... i honestly do believe position sizing is crucial and before even seriously starting to trade or looking at technical analysis for entries... i looked at various position sizing models kelly criterion mm, john something... for this theoretical system i have a high risk tolerance i am willing to tolerate a very high drawdown so long as i get large profits fast... this may change once i start demo trading depending on the psychological impact the drawdown may have on me... this but thats the original idea .... my leverage is 1:500 if anyone could suggest any percentage of risk given my goals it would be appreciated. i am mainly looking for an aggressive percentage large gains and large drawdowns, a conservative percentage small gains and smalll dradoiwns and an in between percentage ... that gives me decent gains with decent drawdowns

if this is its too much i would at least REALLY REALLY appreciate an explanation as to how did you calculate thew 0.36 ratio as this seem like a very important number as well as how to calculate the Ratio of Expectancy to Standard Deviation of R

there is a user named howardbandy in this forum that seems to be very knowledgeable about van tharp strategies ... unfortunately it seems as thought he is no longer active, hopefully he will see this threat and give me some insight ....other than that if anyone in this forum could help me i would highly appreciate it

#### elbee

0.36 = 0.68 / 1.86

#### ilovemycat

thanks is that number the one used to determine the quality of your system

according to this paragraph that might not be the case

Of course, there is one other important variable in your
system: the number of trades it generates. A system with a ratio
of 0.75 that generates one trade each year is not a Holy Grail
system because it doesn’t give you enough opportunities.
However, a system with a ratio of 0.5 that generates 20 trades
per month is a Holy Grail system, partly because it gives you
more opportunities to make money.

if im right whats the ratio used for...
if im wrong why does it say in the paragraph that you may have a good ratio but not a holy grail system im trying to come up with a holy grail system according to van tharp thanks

#### ilovemycat

in other words the sqn for the first system according to my math its 1.0340488626 making it a holy grail system.... this number thought seems very high somewhat unrealistic according to even van tharp

It’s very difficult to come up with a system with a
ratio of mean R to standard deviation of R as high
as 0.7. For example, if I take a system with a ratio of
0.4 and add a 30R winner to it, the net result is that the
standard deviation of R goes up more than the mean
does, and so the ratio declines. What you need for a
Holy Grail System is a huge number of winners and a
small variation in the amounts won and lost.

and for the system i created it would be
0.2/1.9321835661586 x square root of 10=0.3273268354
making it an excellent system

if someone could verify i am indeed correct ill appreciate it

#### peter2

No, you're not correct.
The SQN for your system is 0.2/1.93 = 0.10. Making it worse than poor. Even though your stats have a positive edge (expectancy), whether it's tradable is another matter.

You mentioned that you could handle a system with a large draw down. I'm thinking that you won't when you trade with real money. An experienced trader once asked me to consider how much draw-down I could handle, then halve it. He was spot on.

Yes, the number of trades is important. The SQN number is unreliable over a small number of trades (<30). A number greater than 100 is preferred and the SQRT (100) is 10. Van Tharp later modified his SQN scale by a factor of 10 assuming that people would use a large number of trades when researching their trading systems.

-------------------

It's easy to throw around numbers like the 30% win% and you're going to get 3R wins. Well, it would be more valuable to ask yourself, how are going to get these 3R wins? What sort of system is capable of giving you these trading opportunities?

#### ilovemycat

im looking at very volatile pairs ei jpy pairs gbp aud .... to be honest im not sure if its gonna work but im willing to try im demo trading only... i do have a question btw if the SQN is 0.2/1.93 = 0.10 why does the formula say to multiply it by the square root of the number of trades
SQN = ((Expectancy / Std Dev "R") * Square root of number of trades) unless the formula is wrong or im missing something

#### peter2

The formula did initially include SQRT(# of trades). However the number is unreliable with either a low # of trades or a very large # of trades (eg 900) as the SQRT(900) = 30.

A simplification of the SQN uses the number 10 as the SQRT(# of trades). This produces SQN numbers that are easier to remember and understand.
Good system >2.5, Excellent >3.0, Holy Grail (and write an ebook quickly) >7.0

#### ilovemycat

ok well sorry for asking but even to that ... that gives me even weirder results
0.36x10= 3.6 for system one which is 5 times more the holy grail ratio
and 0.1x10=1 which is above the holy grail, is my math correct?

my question is between what and what numbers should i use the square root in the formula.... between 100 and 500 for example?

or should i just completely ignore the square root period ... you said the formula initially included the square root ... did van tharp revised the formula and decided to exclude the root?

and thank you for your help

Last edited:

#### ilovemycat

i actually contacted support and received no answer this post is the original mail i just changed the subject not to make it look like i was talking to a person ( van )

Replies
42
Views
21K
Replies
13
Views
16K
Replies
16
Views
6K
Replies
22
Views
7K
Replies
15
Views
6K