August 8, 2011

Recently some asked the questions: Are there non-conventional equities trading algorithms/strategies? What are the algorithmic trading system challenges? Does trend following really work?

I think my best answer to these questions is with an even bigger question. I opted to set up a relatively high-frequency trading strategy and then find reasonable solutions under the constraints hoping to address indirectly the challenges being faced.

But first, a lot of assumptions are contained in the above, like what is a reasonable solution and what are the constraints. The shorter the holding period the more stock prices exhibit random-like movements. The higher the rate of trade, the less time to decide to get in or out of a trade at a profit or a loss. And reasonable to one might not be someone else's cup of tea. Therefore, look at the following just as a guideline for some of the problems that you may also encounter in the process of designing your own trading strategy.

I'll start with the trading strategy rules.

1. Buy 500 shares on every dime cross coming from the dime below. You can also short the dime cross from above. No technicals or fundamentals required; just price.

2. Use mostly iceberg orders to get in or out of a trade except for the stop loss in which case you may zap the other side for a fast exit. You may cancel any non-executed entry order at any time. Looking for at least a dime in profit but ready to accept less if holding opinion changes. Better to get out with a penny than to take a loss.

3. May use any technicals available to manage a profitable trade in an attempt to extend profits; but can not bypass, extend or cancel its corresponding stop loss unless the trade has been disposed of. When in a position, you can only get out with a profit, which will also cancel the stop, or with the stop loss itself being executed.

Knowing that you will be playing mostly against machines with some that can adapt and outguess your every move. Not showing your hand becomes important if not critical. You would like to go undetected as much as possible; like flying under the radar. You do not have to take every possible trade. As for commissions consider between $5 and $10 round trip (so select a deep discount broker).

Now the real question: How would you go about implementing such a strategy?

Whatever trading strategy I design, it must always meet some very basic requirements beyond being profitable. Having a positive expectancy at times is not sufficient. You need market exposure; being in the market less than 5% of the time, committing less than 10% of your capital or continuously averaging down dumpers might not be the best ways to, over the long term, outperform the market averages. In the first case, you need 20 times the expected long-term Buy & Hold performance just to get even and committing 10% of capital will require 1,000 % to reach the same point.

Based on the above-defined trading rules, it appears as if I designed a 50/50 game where the expected long-term outcome should tend to a zero gain. Actually, a lot less since all the commissions will have to be paid while playing the game. So the real and maybe only challenge might be to convert this zero-sum game in such a way as to give you an edge that can be maintained over time. Notwithstanding this primary objective, other constraints will also need to be addressed.

1. Trades must be marketable: trading 500 shares at a time is most certainly not a problem. This is not like trying to flip 50,000 shares on every trade every day to make a dime kind of strategy. I can select stocks that have significant average daily volume with sufficient average daily range to satisfy the requirements of this trading strategy. These trades are not intended to be a long-term investment anyway.

2. The method must be sustainable: all that is required are price fluctuations of 20 cents or more and this will be there for any foreseeable future. There will be no lack of opportunities considering the low-level entry requirement. There is no edge defined or even needed to execute this particular strategy. Finding stocks that move by more than 0.50 cents a day should not present a problem either...

3. All trades must be executable: certainly no problem at this level of trade; nonetheless, the stock selection should be limited to stocks having high enough average daily volume and price range, they are in sufficient numbers to meet availability requirement. There needs to be someone on the other side of your trade; and you also wish to be undetectable, if possible, when executing trades.

4. The trading strategy must be feasible: on this count, it all depends because here you find some real constraints: available capital, available time, opportunity count, duration of trade and trader dexterity. Not counting programming skills and a good understanding of the game to be played.

With limited capital, only a small fraction of the total opportunity count might be exploited and executing everything by hand would most certainly be another limiting factor. Trading by hand more than 5 positions at a time might be pushing the trader's attention span's limit when, for instance, momentum kicks in.

But wait, what is the opportunity count for this trading strategy anyway? Let's see: the crossing of a dime from under. There are over 800 listed stocks that, on a daily basis, have a daily range of at least $1.10. That is already some 8,000 possible trades a day. Now if on these 800 stocks you are looking for every 20 cents move or more this will increase the count by a factor of 10, 20 or more. So technically, you are looking roughly at an opportunity count of 80,000 to more probably over 150,000 trades a day. And this without accounting for the reverse side. Since the watchlist is composed of high volume, high momentum stocks, the number of 20 cent moves might be even higher. Really enough to keep someone very busy!

And this is why and where trade automation is required; there is no other way out even with limited capital. And with trade automation will comes another set of constraints.

In this trading strategy, it was not the market that determined the limits, it was you as a trader coming short on capital, hardware, fast fingers and attention span. To compensate, a really fast machine with a high-speed link to the market is a minimum requirement. When playing against high-performance computers, sophisticated programs that can detect and analyze your every move, you should think of designing your own “stealth” trading mode. You play at the second or the minute bar while your opponent may answer within 5 microseconds to any bid or ask on any exchange and thereby front-run or torpedo whatever you may want to do. Leaving a limit order on the books just becomes an invitation to lunch, just as issuing a market order for that matter, and guess who is on the menu.

With all this effort, will this thing be profitable? Well, definitely YES. And the main reason is: we are not good enough at predicting future stock prices, but then so is the other guy. Because you are using a variant to a limit buy-stop order (iceberg) you have a chance of going undetected, not showing your hand. Have you noticed how many times when your price target exit has been reached, that prices continue going up afterward without you?

Randomly picking your trades from your availability set would be another way to remain camouflaged. Your opportunity count is high enough that you can pick and chose what may represent to you a higher probability trade. Having tight trailing stops in place, you will be limiting your losses, while using trade extenders will give you the ability to push profits a little higher at times. Technically, it will be in the design of your trade extenders (trying to hold winning positions for more) that you will find your “edge”.

The real reason you will win, in my opinion, is simply by default. You structured your own game within the game, doing your little thing extracting a dime here and there and cashing in on a high number of trades. I estimate that one can have an expectancy of about $20 to $30 per trade. It is not much, but if it is done 1,000 or 10,000 times a day, the numbers really add up. From my observations, people design trading strategies that try to adapt to market conditions indirectly forcing them to use technical tools to evaluate where the market is; and then basing their decisions on these evaluations. Whereas the proposed Dime Cross trading strategy is just designed as a game of chance where the trader extracts what he can from the price gyrations knowing that he/she has a positive expectancy. The same logic would apply to a 20 cent cross; which would reduce the opportunity count but be more profitable, and where commissions would represent a lesser percentage of gross profits.

This Dime Cross trading strategy should be a pure gambler's delight. There is no requirement to know the future. Your entry targets are very easy to set: the next dime up or down. And because you are using actively traded stocks, after entry you are more likely to get your next dime than to lose it. Losing would require buying within less than a dime of the next intermediary top. My analysis on this is that we are very poor at predicting prices; and in a method such as this, it can become a real asset.

It is in the trade extenders where some effort is required. Only technicals can help. Only trend-following indicators can come to your rescue. It is the design of these trade extenders that will make this strategy go from an already winning strategy to a highly profitable one. My own initial designs of these extenders are very promising. And they are all of the trend-following types.

Under the microscope, this method is a very short-term trend following system. Show me my 10 cent move and I'll consider getting in; and once in, I'm looking to extend my stay if profits continue to increase. If not, I'll take my stop no matter what happens afterward. There will always be more than plenty opportunities to pick from. It is a low-risk trading method where you only put a dime on the table at a time. In a way, you have designed a casino slot machine with a positive expectancy where your “edge”, the size and number of trades is your way to the jackpot.

The challenges encountered are not what might have been expected as it turns out that it is better to concentrate one's efforts on trade extender algorithms as that is what will produce a disproportionate amount of the total gains. It is not a question of being right or wrong, it is just a matter of the market as a whole having a very hard time keeping price fluctuations within a dime for very long on high volume stocks. You are using the very nature of the market itself to extract your own profits. The trade extenders, defined according to your vision of the game and your programming skills, will determine in the end how much you really make.

Right from my archives, here is my little stock trading story. I use it to analyze the feasibility of a trading strategy, study its behavior and possibilities. I think it might be of some use in designing your own. You can use the numbers generated by your current trading program and see how your edge can pan out. The story itself provides the needed understanding to recreate the mathematical formulas.

The Trading Story

Trading Story 

(click to enlarge) 

For those wishing to save some time. You may download the spreadsheet. It's nothing fancy but it works. (Note: spreadsheet no longer available 12/04/2018).

I usually design long-term trading strategies where, in the end, the entry itself is not that important when looking at the overall performance. The Dime Cross is my excursion in the very short term trading arena. But there too, one can design trading algorithms in accord with the very nature of the game itself.

Hope it can help.


Created on ... August 8, 2011,    © Guy R. Fleury. All rights reserved.