Thursday 15 September 2011

Simple Algo

This is my basic algo's stats. By basic, I mean that there are no bells or whistles and that the profit target that the algo is based on is 1.5 ES points. It trades from 8am London to 9pm London times.


I can use this basic algo to auto trade or hybrid trade. The profit target is my first logical scale out point and with hybrid trading I exit part here and manage the rest of the exits manually. The important thing is that the profitability and win rate are here to begin with and allow me to take the rest of the trade to a more optimal profitability. For autotrading I re-optimise to find the best balance between profitability and draw down, but make sure I keep the robustness with walk forward testing.


If you are a trader looking for just, say, 2 to 4 ES points a day then creating a high win rate algo such as this may be an idea. The stats cover about 6 months of data and the required daily earnings can be tuned by the right number of contracts. The tight target brings the high win rate. This algo can be further improved by tightening the times of day that it trades.


Designing an algo means I start with a trade premise and this one is my basic pullback in the direction of the trend. I can make it more or less aggressive but it will impact my win rate but may be more profitable as an auto trader. The process is to always have in mind what the end result needs to be. 


Starting with a blank piece of paper, I set out what I want to end up with in terms of metrics. The idea is not to necessarily catch all the trades that may exist. Only those that meet your risk and other criteria. The idea is to make money and tailoring the algo to your precise requirements is a good way of achieving that. It's not a matter of opening an optimisation and looking for the largest amount of profit the squiggly lines can produce in back testing. It's what it can make in real time with real money that counts.


Once I have a good basic algo, I can build on it and bolt on scaling out, break evens, and other actions to maker it more suitable for auto trading, but always from a trading point of view - market action.








Some of my early students are successfully implementing hybrid algos on their own - no real teaching by me, although I do pass on tips if asked. They use the basic stuff I showed them and use the algo for instant trade recognition and trade execution. The benefits they are reporting is being able to stick to their TPs and not missing a trade. Some of my other students are coming to the workshop. All the new people at the workshop get sent the Theory DVDs from last year's training to get them up to speed before the November workshop.


All you guys who are waiting for the new credit card button to pay, we are almost there. The bureaucracy is unbelievable. Anyone who can't wait, can bank transfer if they want. Just email Candace at candace@electroniclocal.com for instructions.



2 comments:

  1. Hello EL,

    If I understand your stats correctly, your basic model trades a little over once per day on average (139 trades divided by 120 trading days in 6 months), and produces about $300 per month in profits trading one contract ($1800 divided by 6 months). Am I reading this correctly?

    ReplyDelete
  2. Anon 16:55, Not qyite. The basic model can be tuned to trade many ways. It recognises the pullback to the trend. I can make it only trade the first pullback after the change in trend. The profitability depends on the scale outs. What you saw was the model just trading one contract to that first scale out point. If you can enter just after that change of trend then it can be traded all the way to the next change by scaling out and then reloading as continuation trades. For me, an algo has to have a basis to start building it and what I showed was that beginning.

    ReplyDelete