Simply Put When An Algorithm is Capable of Inspecting Current Price Activity and Learning From it To Make Future Price Predictions and Trades

Machine Learning is a vast subject but I would like to keep things simple;

Let's say your trading system is set up to trade when price extends beyond a certain daily level in points or pips. For arguments sake we can use a value of 100.

Let's further say that the algorithm is capable of detecting that in a reduced volatility market there would an opportunity to trade profitable when price extended only 50 points/pips and on a reduced target profit, then you algorithmic can e said to be learning as it trades and analyses the market.

Markets never stand still, they expand and contract as interest in the instrument ebbs and flows. One minute a market can be on fire the next in can be in the doldrums. 

Extended doldrums or even a change in external factors, can reduce market activity so much,  it can send the instrument and market out of business. Example the Potatoes Futures market which is no more with us due to volatility which dropped to nothing. Why ? Because of much improved refrigeration and storage possibilities,  of the product itself, which negated the need for producers to hedge on the future exchanges.

By adapting to the type of market conditions currently in place , the trader can avoid costly mistakes, for example the 'one size fits all trading style' using an algorithm which will continue to trade in 'yesterday's market conditions' rather than adapting to today's and as a result,  end up drawing down or even worse losing money. 

Machine Learning may be applied dynamically as is the case with advanced and sophisticated algorithms which have been designed and coded to adapt in flight, or at the 'end of term' whereby the algorithm is set to run periodically in adaptive mode so that an initial parameter set may be adjusted for future trading opportunities.

Many of the worlds leading banks and hedge funds now make machine learning part of their overall investment and trading approach where they use advanced statistical analysis to modify their system according the price discovery of the most recent trading activity or period.

Is Machine Learning a solution to all traders ills ? No. As the definition of what constitues Machine Leaning and how it is applied to financial markets differs depending on the market participants individual view of the market and how he/she applies the Machine Leaning algo and defines its ability 'to learn'.

What is clear,  is that the ability to adapt an algorithm 'in flight' or an 'end of term' is a valuable weapon in constructing a trading algorithm which is able to operate across all market conditions and states.

The simplest method of adopting a machine learning alg into your trading is to apply one which constantly learns at 'end of term'. This is how I apply machine Learning in my own trading.

The results can be seen in the various pages on this site and are witness to what is possible with some application.

The End of Term learning period used,  will depend on the time frame of your trading methodology. An algo which trades on a monthly timeframe or interval should not be expected to learn off a 5 minute time frame interval.

Join My Mailing List now

Why I'm a member...

Paul Langham

Below you will find answers to some of the most frequently asked questions

Find Out More by Joining My Mailing List now

Copyright © 2021 Exact Trading. All Rights Reserved.