NON-CONVENTIONAL STYLE OF TRADING SHARES:

We don’t like intraday trading stocks, but we are short-term traders and we like to run and run in the US stock market. We like to get into positions when they are moving and then exit in two to three days. We believe this is a very efficient way to trade and combines security with very high returns.

But to do this we use a very unconventional style of negotiation. We established a very large group of markets, currently 96, we limit our commitment to each market to about $ 1,000 and then we take mechanical trading signals from a trading system that we have programmed and traded with real money for many years. We use a custom trading platform that interacts with E-signal’s live streaming data. We sit in front of a computer for six and a half hours per trading day and typically perform between 10 and 30 trades per day.

IDENTIFICATION OF VOLATILE MARKETS:

But because we do a lot of trades and are only trading for two to three days, our methods will not work in dead markets. Our methods require us to identify volatile markets.

Identifying good volatile markets can be a bit tricky. At one point I used a simple form of back testing to do this. I’d take a market, get a couple of months of tick data for that market, and then apply our trading system and look at the results. If the results looked good, I would put the market in my portfolio and if the results looked bad, I would discard the market.

The results of this method can be disappointing. A market that had made a lot of money for 8 weeks could produce a series of two or three losing trades just as I was putting real money into it and the market that I had discarded could start to make money.

What I soon realized was that this approach was actually a form of optimization that was, in effect, attempting to predict the future performance of the trading system by trying to fit a system to a given data set. It was a form of “curve fitting” and curve fitting is the worst thing you can do to identify profitable trades. This was simply not a good approach.

But what I realized from working with market data was that the critical factors in identifying profitable markets were volatility and tracking.

I then investigated trading software that allowed the user to scan a large number of markets and enter certain criteria to identify the markets that met those certain criteria. I found this trading software useful for identifying volatile markets, but the results were not as satisfactory as I expected.

The problem was that most trading programs use the range over a period of time to determine volatility. The problem was that sometimes that range occurred in a single day and the rest of the time the market was dead.

I will give you an example of a market with a lot of volatility for two days, but was nonetheless a waste of time to trade the rest of the time. On 12/16/09 there was breaking news from DCGN, deCode Genetics, and the market exploded and ranged from 6 cents to over 30 cents, quadrupling in value in a single day. That is volatility! One day this market was at the top of the market winners list and the next day it was at the top of the declining markets list, going up and down in two days. As I write this on January 10, 2010, DCGN is back to where it started before the news and is as flat as a pancake. But if you run a volatility analysis on all stocks from December 2009, DCGN is likely to top the list. And yet it was but a one-day wonder, and outside of that day, there would be no point in keeping it in a trading book.

This type of market movement is not unusual and creates problems identifying good markets to trade. Software that uses range over a period of time does not filter this type of market.

After a bit of experimentation, I found a solution to this problem that I will share here. What I did was develop a program that could scan a data stream and identify the features that normally work well with our unconventional business methods.

The markets that performed best with our trading methodology were the markets that had repeated volatile and expansive breakouts with follow-up for a day or two. After a range expansion, the market could contract for a few days, but this contraction could be followed by another expansion and then some more.

DUMMY DAY TRADING SYSTEM

To identify such markets, I programmed a fictitious daily trading system. We do not carry out daily transactions and I do NOT recommend intraday trading or this system for real transactions. But to identify good breakout markets for our methodology, I established the following simple rules for the fictitious intraday trading system:

1) The “system” uses our proprietary scheduling method to determine the number of contracts traded and limits the size of our positions to approximately $ 1,000 per position taken. In the world of stock trading, this could be considered a small position. We do this to enable us to trade a large number of markets with a small amount of money. We currently operate in 96 markets and in doing so we protect our business capital through diversification. Therefore, we will buy 1000 shares of a share that sells for 98 cents per share, but only 100 shares of one share that sells for $ 10.02 per share.

2) After the close of a given day, the DUMMY SYSTEM determines the range for that day. It then calculates 25% of that range and adds that value at the close of the market to determine a buy point for the next day. Therefore, practically any type of significant bullish movement the next day will result in the fictitious system buying the market. Typically, the fictitious system will receive a buy signal every two days and will show around ten trades for every 20 trading days or so.

3) An entry stop day is entered immediately when a position is taken. Using 15-minute bar data, this stop will exit a market if it retraces its move more than 75% from the last intraday high. You rarely get to this stop.

4) All positions are closed at the close of the trading day.

This fictitious system is really just a detection device. These are partial results of a GOOD MARKET, BIOF, which was tested with intraday data for eight weeks from 11/09/2009 to 01/08/10:

BIOF BioFuel Energy Corp. (NASDAQ) 15 min bars 11/09/09 – 01/08/10

Total net profit = $ 552

Number of operations = 17

Wins = 10 (59%)

Average profit per trade (wins and losses) = $ 32.49

These are partial results from a BAD MARKET, ARBA, that was also tested with intraday data for eight weeks from 09/11/2009 to 08/01/10:

ARBA Ariba, Inc. (Public, NASDAQ) 15-minute bars 11/09/09 – 01/08/10

Total net profit = $ 44

Number of operations = 19

Wins = 12 (63%)

Average profit per trade (wins and losses) = $ 2.32

When you look at the three-month charts of these markets, you may be inclined to believe that both markets are volatile and that they would be good markets to trade. Conventional methods of determining volatility will likely show that both markets are truly volatile. But when we apply the fictitious system to the 15-minute charts, the difference between these markets becomes apparent.

The bottom line is that BIOF is a huge market for our methods, but we are wasting our time with ARBA. The problem is that ARBA is simply not volatile enough to beat our transaction costs by trading our relatively small positions. That is why we must reject this market.

As a general rule, when I scan markets with the fictitious system, I like to see the average trade (win-lose) above $ 10. If the average trade is less than $ 10, I reject the market for use in our portfolio.

I have found this market selection method of identifying “hot markets” for short-term trading to be superior to other methods, trading or otherwise. I have found that markets that show an average trade above $ 10 using the fictitious system will generally show excellent real-time profits when trading our short-term trading methods.