Van Tharp's SQN with over 100 trades | Psychology and Money Management


futures.io - futures trading strategies, market news, trading charts and platforms


Psychology and Money Management


This forum is where successful traders spend most of their time and where struggling traders never visit. The psychology and money management forum is where you learn to become a profitable day trader




 

Van Tharp's SQN with over 100 trades

  #6 (permalink)

Kansas City, MO/US
 
Trading Experience: Beginner
Platform: NinjaTrader
Favorite Futures: Futures
 
Barz's Avatar
 
Posts: 16 since Jul 2018
Thanks: 26 given, 9 received


brakkar View Post
Hi,
this post assumes that you know what SQN (System Quality Number) is.

in "Ultimate Guide to Position sizing" first edition 1, I quote:



How does this translates in the equation to calculate sqn?

Original sqn calculation for UNDER 100 trades is

(Expectancy / Standard Deviation) x (square root of number of trades).

Not quite sure what you mean by "translate," but I'll take a stab at it.

I have the 2nd Edition and this is discussed on pages 37-38.


Less than 100 trades:

SQN = (expectancy / std dev) * SQRT(number of trades)

More than 100 trades:

SQN = (expectancy / std dev) * SQRT(100) = (expectancy / std dev) * 10

Why would you do this?

Honestly, it's a bit of a hack. If you have less than 100 trades, then you don't have enough trades to give you a good measure of your system. Similarly, if you had 900 trades, you can get a great looking SQN for a not so great system.

So, Tharpe says you should cap your number of trades at 100 so you don't get abnormally large SQN values.

The whole point of the SQN is to compare one system to another over the same time period to determine which is better.

All other things being equal, systems with 100 or more trades in the same time period are better than ones with fewer than 100 trades. The SQRT(# trades) helps to capture that to some degree in the equation.

Quote from p. 38:
"There's some analytical value to this exercise, but remain cautious of normalizing small trade sample sizes up to 100 because the resulting SQN 100 score can be quite misleading...As a system acquires more live trade results it's a near certainty that the standard deviation also goes up, which then reduces the SQN score."

Reply With Quote
The following user says Thank You to Barz for this post: