# Standard deviation has to be..... standard

Hello everyone! I'm relatively new to Amibroker and I love it. Unlike Matlab, it provides an excellent interface -- specifically for finance -- right out of the box. However, I'm perplexed by one omission: Standard Deviation. It's a metric that should be included everywhere -- in backtest mode, explore mode, walk-forward optimization. Everywhere. I see some discussion on this board about how to add a custom field for standard deviation, but honestly there should be nothing custom about it. It's as fundamental to the principles of finance as percent return. I defy anyone on this forum to pick up a textbook on the basics of finance and not find standard deviation throughout the text. The standard backtest shows the Sharpe ratio of my trades, the CAR, and I set the risk free rate, so I can solve for standard deviation using a little algebra. But why isn't it a standard column/row? I feel I should be inundated with standard deviation in any financial software package, especially one a incredibly robust as Amibroker. I've searched the forum and nobody has pointed this out, which leads me to believe I'm missing something huge. Is there anyone else out there that feels the same way?

It is standard, read the manual

So here is the standard system test report: https://www.amibroker.com/guide/w_report.html
Please show me where the "Standard Deviation" statistic is, notwithstanding its mention in the Sharpe Ratio section. Thanks.

Any metric can be added via Custom Backtest procedure (as well as CBT examples in KB).

Yes, output of standard deviation to report is here for example

As for "should not be custom"...
First of all, AB is programming software and stdev not being rocket science (so easily to be added by user).
Secondly, you may do request feature addition at feedback center.

1 Like

Thank you, fxshrat. I've previously seen the example that you posted. It just feels like a bought a new car and I can't find the steering wheel and people are telling me that I can add any steering wheel I want through some basic code. In other words, it's so fundamental that it cannot be excluded at the outset. Perhaps I'm in the minority.

3 Likes

The real world is not a Gaussian distribution so it is not fundamental to trading. You may need to check your biases at the front door.

For what it is worth what you are asking IS already included.
Actually something better, standard error of linear regression estimate which
measures how far your actual equity is from perfect growth (standard error of the linear regression of equity line)

Standard Error - Standard error measures chopiness of equity line. The lower the better.

2 Likes

Ah yes, the real world. So I would agree that standard deviation may not be especially relevant to trading; however, it is quite relevant to portfolio management. If, for example, I develop several equity curves from distinct trading strategies and I want to combine these strategies into an efficient portfolio, it might be an important important measure. I think Harry Markowitz would agree with this assessment. But what's even more important than Harry's opinion, if you've ever been meetings with institutional investors, you will invariably get three questions, in no particular order:

1. What are your portfolio's returns?
2. What is your portfolio's standard deviation of returns?
3. What is your portfolio's Sharpe ratio?
If you don't have answers to each of these questions, you will never get an allocation. Not ever.

I don't disagree, but who cares about institutional investors and their mediocrity? AmiBroker is not well suited to portfolio management out of the box without complex coding, which is actually one of my bones of contention. The most advanced portolio managmenet tools I've seen for retail is in crypto which is sadly more of a side effect of crypto being innately technological with bucket shop APIs widely available, and even then it still leaves much to be desired because you're dealing with inexperienced 20-somethings rediscovering the wheel for the first time. For example, naive risk parity is not to be found (oh look, that uses standard deviation!).