Development Journal

The following is a random journal of my systems development. I don't discuss every aspect of development, but when something strikes me as noteworthy, I will journal it here.

To learn more about my systems, please click here.

Commentary for April 29, 2016

Volatility is one of those words everyone knows, but not everyone understands.

"...Volatility refers to the amount of uncertainty or risk about the size of changes in a security's value. A higher volatility means that a security's value can potentially be spread out over a larger range of values. This means that the price of the security can change dramatically over a short time period in either direction. A lower volatility means that a security's value does not fluctuate dramatically, but changes in value at a steady pace over a period of time." Source, investopedia.

Using the above definition, if an instrument gradually moves higher and higher with small pullbacks, the volatility is lower than if it were to move higher and higher with large pullbacks. Volatility does not relate just to trending markets though. A rangebound market can also have high or low volatility. If an instrument trades within a tight range volatility is lower, where as if the range increases, so too does volatility.

Think of volatility as bi-directional price movement. The greater the movement in both directions, the greater the volatility and thus uncertainty about future price action.

So now the big question, how do you forecast volatility? There are plenty of volatility studies and other resources such as ATR, skew or simple technical analysis on the vix. And I've been spending a lot of time trying to understand this very subject as it relates to my own trading strategies. Why? Because I use a mean reversion strategy to scalp intraday. When markets trend and volatility comes in, my strategies underperform. The same is true in rangebound markets with high volatility.

So my strategies prefer rangebound markets with reduced volatility or trending markets with high volatility. But to an extent. Rangebound price action with high volatility are equally dangerous for my systems. Why? Because my stops are too tight. The result, price may move through my stop by a few ticks, only to reverse and then proceed to take out, what would have been my limit, were the trade still on.

So I have to find the right amount of volatility. Using Bollinger Bands as an example, if they are too wide, there is too much movement. Too narrow, what I like to call a pinch point, and it becomes a coin toss as to which direction price will break out.

I believe I have identified that sweet spot for my systems. That level of volatility that is high enough to not represent a pinch point, and low enough to not blow through my stops. Which then raises another question.

Why not have two strategies. Using the same mean reversion approach, but with different risk reward values. When volatility is higher, open my stops for example. Which is where I am now from a development standpoint. My goal is to have as many systems as I can manage and develop. But that doesn't mean simply diversifying through completely different strategies or instruments.

It may be as simple as trading less instruments and strategies, but with tweaks to those strategies to better accommodate the market in which it trades.

This leaves one bigger concern. But something I believe is near impossible to fully identify. When a strategy begins to go in favor and out of favor. Those transitions are difficult. Once it has happened, and my strategy is no longer in sync for example, I have my equity curve to help me know when to turn the system off and eventually back on. And that may be the most a trader can ask for.

Commentary for March 9, 2016

I have been spending a lot of time (i.e. months) trying to find one additional conditional statement (trade rule) to add to my script (program) to avoid trading during conditions that do not support my mean reversion scalping systems.

It's vital that you understand why and when your system performs well and when it does not. The example I like to use is with trend following and mean reversion. They each work in opposing markets. A mean reversion strategy for example will not perform well during markets with strong trends.

A few months back I was able to identify one very simple rule based on the overnight range to determine when my systems would trade for that day or not. The premise, which backtesting proved correct, was if there is too much movement overnight, my stops are too tight and therefore I will have a higher probability of stopping out before price mean reverts. And that worked well for the latter part of my backtest samples. But the earlier samples (I test over three years back) were not impacted at all by this additional logic.

The reason being overnight moves the past year have been far greater than the previous three. So I kept searching for another rule to apply. And then it dawned on me. Are there some environments where the opposite is true. Rather than too much movement making my stops to too tight, are there some instances where there is not enough movement and my limits are too wide?

In other words, is the market trading in too tight of a range that mean reversion essentially  is not sufficient for the size I am targeting. Well guess what? The answer is yes.

So after a very long search and trying countless tests, I was able to identify one very simple rule, which is now incorporated into my systems. The result is a 20% drop in trade frequency, an increase in win rates and a reduction in drawdowns.

I'm alway weary about adding new statements but with over three years of historical testing, I feel confident it is the right move. But I will monitor closely over the coming months.