Recently a trader wrote in about using the approach that I've long championing, which is to keep trading rules simple or relatively so. Specifically, this is an options trader that I quoted extensively in my 8/18 Trader's Corner
and I'll excerpt just a bit of what he wrote a second time here. (If you want to look at the 8/18 article just go to the OIN Home page and click on the 'Trader's Corner' tab at top and scroll down to this recent column of mine.)
A subscriber's e-mail to Linda Piazza included these excerpts:
"Leigh Stevens once wrote an article extolling the virtues of simplicity and cautioning against making market analysis more complicated that it needs to be. I bought his book and proceeded to learn what I could about simplicity, and discovered the value of such notions. As a result, I long ago gave up pursuing complicated strategies and intricate hedging techniques in favor of simpler approaches..."
... I now only look at what market prices are doing and have done in the very recent past. I have learned that market prices go up, until they don't, and go down, until they don't. Sometimes they trade within narrow ranges, until they don't. It does not matter one whit why prices move as they do (and as an individual trader, I am foreclosed from knowing); all I need to know is the current direction of price movement which is a much easier proposition."
"Traders can learn current market direction with a couple of exponential moving averages on the 30 minute chart (indexes preferred) set on the close for 12 and 26 periods. When the 12-period is above the 26-period EMA, be long calls and stay long until the EMAs cross back at the end of any 30 minute bar; for downturns, be long in puts when the reverse is true. For additional confirmation, use the CCI (21 period)."
The person quoted above went on to say that using the 30 minute chart for the CBOE Mini Nasdaq 100 contract (MNX) and getting into MNX calls on the morning of 7/13 and holding until the time of his e-mail (around 7/30) had proven to be a very profitable trade. The moving average UPSIDE crossover is shown on the NDX chart here:
The closing 30-minute bar where the 12-period ('length' set to 12) crossed above the 26-period Exponential Moving Average (EMA) on the 30-min. NDX chart is above the green up arrow on the left. And the trader was writing when the MNX mini was trading around 162 (NDX: 1620). A very profitable trade! We'd all love to have more of those!!
Since the writer claims that the input of this (exponential) moving average crossover indicator is pretty much all that's needed to get into and out of profitable trades, I decided to back test this indicator as a mechanical trade strategy or trading 'system', using my TradeStation (TS) software and intraday data as far back as I have; I have 1000 trading days stored of 1-minute data. There is a bit of knowledge required in setting up the appropriate strategy or system for back testing, but I have to say also that the TS software makes it relatively easy.
My software also allows me to set stop loss exit points if any trade loses a certain amount. I didn't get into that. Nor did I use the TS 'optimize' feature to see if some OTHER set of moving averages would have produced better results for the past 1000 days I tested (back until November 2006).
1000 trading days (3.8 years) in general is not a very long period to test the results of a trading system, but for one using 30-minute intraday charts and generating a fair number of relatively short-term trading 'signals', the test period for which I have intraday data is a fair amount of time to test this strategy/trading system.
Moreover, the period covered looks like this on the long-term weekly NDX char below and you can see that there are a number of sizable price swings both up and down.
In the interest of disclosure, I've used TradeStation for a long time and first got into it because it offered so much functionality in the way of (trading) system testing and development This when I was the European technical analysis product manager for the Dow Jones version of TradeStation, which was licensed from the software developer. I have not had a connection to TradeStation, the company, for over a decade.
I thought the moving average crossover method described was interesting and I wanted to verify objectively how many NDX points were gained or lost by use of the particular moving average crossover over the past 1000 trading days; approximately 3.8 years.
The system I tested does NOT account for commissions paid or execution 'slippage'; i.e., the call or put was bought when the underlying index price was NOT equal to the 30-minute closing bar. These other factors are somewhat more complex for me to get into. I can put into the trading system, an 'average' commission amount and a percent figure for 'slippage'. However, my starting point is simply how many trades were generated, how many winning or losing trades there were, and what was the net gain or loss in terms of index points for the number of trades done.
The results from the trading system called the "EMAverage Crossover" strategy is as follows and assumes that just ONE trade was done per buy or sell signal:
1st Trade: 12/13/06 (Long-Buy call); exit 12/18 (+ 21.3 points);
Last Trade: 8/26/09: (Short-Buy Put); OPEN position
Completed trades: 289; Total Net profit: 78.8 points
Number of winning trades: 94 (32.5%); # of losing trades: 195;
Average winning trade: 38.9 NDX points; avg. loser: 18.4 points
A couple of things stand out from the results noted above:
1.) The average point gain on the winners was just over twice that of average loss. This suggests that building in a stop loss component that cut the number of losing trades could be a way to increase the overall profitability of this strategy.
2.) Given the large number of trades generated, an estimated commission would have to built in to further check the results of using this trade strategy. Also, building in an average 'slippage' estimate would be something to consider.
I'll carry out some further testing later but the foregoing will serve as an introduction to the concept of systems testing and development.
My next charts will show how the initial strategy looks when 'inserted' or applied to a 30-minute NDX chart within the TradeStation software for the most recent NDX 30-minute chart (through 8/26) and then some back periods when prices were not in a well-defined trend; periods when a moving average strategy will get whip-sawed.
The green up arrows indicate the bar above where a buy signal was generated and the little green 'hash' mark or pointer, notes the price at the Close of the bar where the index entry was (theoretically) made; as if we were buying a stock. The red down arrows indicate the bar below where a sell (short) signal was generated due to a DOWNSIDE crossover of the two Exponential Moving Averages.
From the recent results seen below, it's apparent that there were a number of trades generated, but only one that had a subsequent substantial move.
By the way, with this type of software a 'signal' alert can be set that visually and/or audibly flashes on your screen that a trade signal has been generated. A strategy 'signal' can be sent to a pager or sent I believe via a text message.
'Whip-saw' periods can be seen on this next chart where a sell signal was rapidly followed by a buy signal, which was repeated two more times in the period shown.
One of the whippy periods is 'blown up' in my next chart and you can see the problem in using any moving average ("MAC" stands for Moving Average Crossover) system at least in a mechanical way, and without some built in method to cut losses short, OR to build in a means to stay out during 'non-trending' periods.
There are other things that can improve results also, such as 'optimization' of the moving average lengths, which I'll get into in a future article. One strategy involving the Relative Strength Index (RSI) as a 'triggering' indicator (for trade signals) that I have tested as a trading 'system' cries out for optimization of the setting I initially tested for the lower RSI extreme that is part of initiating a long side (buy calls) trade.
A different trading strategy shown next also involves 'pyramiding' by allowing additional 'signals' to trigger additional trades. This is the meaning of numbers (e.g., -1 to -4, meaning additional put purchase signals) that are seen over the RSI notation.
A sell signal (buy puts) is generated when the 21-hour RSI crosses BELOW 70, a buy (calls) signal is involved when the RSI crosses ABOVE 30. Use of 30 as the lower 'oversold' level is what I want to 'optimize' to see what might be a more promising number; 30 for oversold is in common use but optimization for profit may show another number as having the optimal results.
Without ANY tweaking of this strategy or trading system for NDX this was the indicated results of an RSI Buy-sell strategy for the past 1000 days:
61 trades generated since 12/22/06
Total Net Profit (in NDX points): 588.5
Number of winning trades: 42 (68.8% were profitable); Average winning trade: 79.8 points; avg. losing trade: 145 pts.
The average amount lost on losing trades needs to be cut by use of stop-loss protection. While 69 percent winning trades looks favorable, the average loss is too high. The use of 30 as the lower 'trigger' should be revised and optimized.
I'm going to work with the above described trading strategies, particularly the later one, to cover in my next week's Trader's Corner and then go into a further discussion on trading strategies that could be automated.
Imagine sitting at your desk and not in front of a screen all day and getting 'texted' to enter an occasional trade; or, having that trade go automatically to a broker to buy or short some number of shares of QQQQ. That will be a next step, to back test an NDX strategy system using the proxy QQQQ tracking stock, which will give better results in a P&L; the back-testing software I'm using is set up to show a net dollar profit or loss for stocks.
GOOD TRADING SUCCESS!