How does one decide Moving average and its cross-over parameters to use

Question

For the purpose of trading options of a volatile index, I want to understand how to decide between the following for a particular indicator, say, RSI:

Assuming that the RSI changes from 0 to 100 in a day, I want to detect more than normal changes in RSI at any value. How do I decide between the following:

1. Whether to use RSI Moving Average simple or exponential or some other
2. What period to use for Moving Average, most docs say 14
3. Whether to use Moving Average Cross overs
4. What values to use for 2 cross over periods

Similarly, one of the other indicators, moves around 0 in fractions as well as hundreds during the day. I want to detect a big change there too. How does one go about using the values for above?

I can see various values being recommended all over the web but the logic to select those values is not explained. The logic is needed to choose different values based on the volatility of the indicators.

Answer 1

The plethora of day trading strategies stories all over the web are like condoms that come in three sizes: Small, Medium and Liars. Most are trying to sell you the dream that some combination of indicators will enable you to print money via day trading. Something like 90% of day traders fail so your best chance of making a million dollars from day trading might be to start with two million dollars.

Re some of your questions: The success of moving average crossover systems is dependent on selecting the right periodicity. The longer the MA, the less noise and the fewer the number of whipsaws. In return for that benefit, trades will be late in and late out (lag). If you shorten the MA, you'll have more timely entries and exits but you'll incur a lot more whipsaws. The only way to know which MA works best is hindsight.

To see this statistically, obtain several years of data. Optimize the indicator on say the first quarter of your historical data to determine the ideal periodicity. Now apply that ideal MA length on the remaining quarters of your data. Odds are, it won't do well. Then try that same indicator and ideal number on another security. Oops, not so good.

Every security has a period where one MA does quite nicely yet it doesn't on another period of data. Only knowing the future will enable you to successfully trade with MA-s and that's obviously not going to happen.

As for the Welles Wilder's RSI, it's an improvement on Rate of Change and Stochastics indicators because it removes the “take away” effect of early data. Because it is a ratio, it eliminates the problem of needing large amounts of historical data but because a ratio is used in its calculation, it is more volatile and erratic.

As mentioned above, shorter RSI periods result in a larger number of whipsaws. Longer RSI periods result in more reliable signals but they're not as profitable because of delayed entry and exit. Don't get caught up in believing that such signals are accurate. Anything that is overbought can become far more overbought and anything that is oversold can become far more oversold.

The short answer is such technical analysis indicators provide information like support and resistance, trend, and current momentum but they are a just reflection of past price and/or volume movement and they predict absolutely nothing going forward. It's like looking in the rear view mirror to see where you have been.