Maximum Profit Strategy and Directional Efficiency by Valerii Salov

I want to build an automated expert advisor.
I am trying to understand these articles http://www.futuresmag.com/Issues/201...-analysis.aspx http://www.futuresmag.com/Issues/200...l-profits.aspx http://www.futuresmag.com/Issues/201...is.aspx?page=4 written by Valerii Salov ,who is a scientist and he worked for major investment banks as head quant/developer.Although I have read more complex papers and articles by different scientist,I am having difficulty grasping the articles.Does anyone know exactly what maximum profit strategy and directional efficiency is?

Good find, better than the usual claptrap. This is basically a non-directional reactive strategy. Simplest way of thinking of MPS as its described is:
- Imagine price is at 100.
- Long at 105. Target 120
- Stop and reverse short at 95. Target 80

Repeat for as many times as it takes before price hits target.

Everytime price travels through 105-95 you incur transaction costs. Working out the size of the bands and the target is a product of filters he suggest (like volume, obviously not available in OTC instruments) and duration of trade. One other he introduces is directional efficiency, which is number of ticks to get from a-b expressed as a ratio.

He's translated automated delta hedging techniques and applied them to instruments without the burden of writing or buying the derivatives. The strength is in the number of instruments he's scanning across but in FX, correlation will work against you.

Without true vol and the increased correlations, you may want to look at other filters/parameters like time of day and if you're feeling brave, using IV to work out whether it's cheap or expensive to trade.

Thanks Triphop ,

Hmm ..I agree ,I dont think the hedging will give edge in an expert advisor.Threre are lot of literature about money managment, risk managment but they mainly focus on risk allocation, position sizing etc..

I am looking for strategies(probability theory, validity tests used in digital signal processing etc..) to sharpen the entry and exit of my trading signals.


I have only come up with the idea of calculating indicator's z score so far.If the z score of the indicator is significant, the indicator's signals should be significant, too.

The actual strategy isn't so much hedging as based on the concepts of dh, which is different. I'm sure it could work in FX, but you need to tweak it to take in account the different nature of the beast.

Z scores are fine, but maybe just scrap the indicator and take price as the primary input?

Strategy wise, there is a mountain of stuff as you've found out. The position sizing stuff is pretty tired but it can make all the difference. The probability of price moving in x or y changes relative to a point X (eg midnight open), and ensuring your size changes with this is in my experience crucial, however you trade.

Check out Mikkom's thread for a few ideas.

The geometric directional efficiency, GDE, is simple. It is suggested as a property of the market price time series. On a chart price vs. time any two points, ticks, can be connected by a straight line segment. Its length, l, will depend on the units and scales of price and time. It is easy to compute this length using Pythagoras theorem. If these points are not neighbors, then there are other points, ticks, sorted by time between them. Let us connect these intermediate points by sequential straight line segments, measure length of each, and summarize the lengths getting some L. Clearly, L >= l because the shortest distance between two points is the straight line. L will be equal to l only, if all points are on one line. This would mean that price was moving from one point to another in the most direct way. Thus, the ratio GDE = l/L, varying from 0 to 1, characterizes linear efficiency of the move. It is a successor of Fractal Efficiency, FE, - a ratio of a price increment divided by the sum of absolute, by module, price increments for intermediate prices. FE does not account time. GDE is a measure combining price and time. For coinciding points GDE is set to 1.
Studying statistical and moving (when time interval is gradually shifted to the right by time) properties of GDE one can make better decisions about changing market properties and make better trading decisions. This property is easy to adapt to automatic computer trading programs.

It is an objective property of the market. A price time series recording market transactions is a sequence of ticks (time, price, volume) ordered by time. Forllowing to market trading regulations (margins, short sells, etc.) and assuming transaction cost one can record trading actions (buy, sell, do nothing) by a sequence of integer numbers (number of contracts or shares) corresponding to transaction times. Negative numbers are sell, zero is do nothing, and positive are buy actions. This list of actions is a strategy. For any intervals of the time series using price, costs, and actions, we can now compute profit or loss, PL, for any given strategy. If we restrict the initial capital and/or the number of contracts/shares on a trading position, then we can speak about a strategy generating the maximum profit. This is the maximum profit strategy, MPS. It cannot lose money. If there are no price moves exceeding transaction costs, then the MPS is do nothing strategy.
MPS is a rich concept. It is rather new. MPS creates a sequence of optimal trades marking optimial trading elements, OTE. Studying OTE properties is a new form of market analysis. MPS is a quantitative measure of the Main Law of the Speculative Market. The last is a statement that speculative markets provide for huge and frequent profit opportunities. The MPS exactly says what is "huge" and "frequent". This is a phenomenological law and the main condition of the market existence. We do not know exactly, why the prices fluctuate. However, as long as they do and, if this are large and frequent price moves, trading will attract risk takers.

I guess you are Valerii.Thanks for the clarification.

I have read your last article in the Future Magazine.Although it will take some time for me to study and grasp it , it is a great article,I appreciate your work.

Thank you