Wednesday, January 04, 2006

Developing a trading system

Everybody and his cousin tells you: You need to have an edge, you need to have a system to trade the markets.
But, if you are like me, sorry, but that sounds so great, so hollow. A trading system! If there were one widely available, we all would be millionaires a long time ago.
Have you met really successful traders? Me, I can count them on one hand and have some fingers left!

A tradeable trading system isn't something available round the corner in the next traders bookshop.

Maybe you noticed the one word I added in the last sentence: TRADABLE

The problem is, there a quite a few trading systems available and I can prove, they are successful in backtesting and in today's real markets.

the problem is, you need already a lot of money to trade them, because they all have periods of nasty drawdowns and with a small account you can't afford such long, statistically irrelevant, drawdowns.

The truth is: You need money to make money

And there are just a few traders, who have the luck, determination and persistence to make the step from small account to big account.

Yes you need luck, because no persistance, no nothing helps you, if you start trading a trading system in one of it's nasty drawdown periods. And add worse to misery, you don't have the knowledge to know when to deviate from the trading system, if you just start out.

There are ways, ... not to shorten the learning curve, because I no longer believe in that nonsense. There is no shortcut to trading success and if someone tries to tell you otherwise: Run like hell or you will pay a far bigger price than you are prepared to pay.

I will post the way, which made sure I still have an account to trade today, even if I made huge mistakes in my career, even if not so long ago I blew 50% of my account in one day and 75% in a week.

Right now, we are looking for a tradable trading system:
I won't give you one right now, maybe, if there is interest, we can try to develop one together. But I will tell you what to do, if you think you found one.

A tradingsystem MUST be backtested. Why?
Because, if something did not even work in the past, you would be foolish to assume, it works now. It might, but are you really this arrogant, that an idea you just had will make you money, if testing it on past data shows you, you lose?

That means the test for the existence of a tradesignal has to be made on Bar Close, otherwise you can not backtest it.
(There are a few exceptions I know to circumvent this statement (especially if you test for the existence of certain bar patterns), but generally you need Bar Close tradesignals, as all indicator values will be calculated just once and not in a range like an OHLC bar.)

The reason you use Bar Close tradesignals is simple: If you use intrabar tradesignals, in backtesting you will see just the working tradesignals, while all the tradesignals, which are present intrabar, but not at bar close will not be seen.
And guess what: Each and every one of these disappearing tradesignals will have been a loser. Why? Because if not, the tradesignal would have been present at Bar Close as well.


I hope you see an Alert at the bars where you thought you would have gone long or short.
Now scroll back your chart in time and do your first backtest in visual only. Look where you got tradesignals based on your newly defined rules.

Usually you end with too many tradesignals or actually none at all. And that means back to the drawing board. ask yourself: Why. What is missing, what additional parameter is needed.
Do you have every ingredient identified, do you need to add some salt or pepper to the soup, a little filter here, a little tweaking there.
Sure that's curve-fitting, but that's making sure your tradesystem does work at least with the chartdata you currently work with. The reality test comes later.

You don't need a 100% system, you need a tradeable tradesystem. So look for about 70% winners and 30% losers. Make sure the stops necessary for the winning trades to become profitable are managable and within your risk:reward parameters.
Unless you design an alltime in the market system, don't introduce reversal trades early in the system design. It just adds extremly to the complexity of the whole system design.

Don't be discouraged, if your first efforts to develop a trading system are met with failure. If it were an easy and simple task, everyone would be a millionaire already.
Trading systems based on the widely available indicators usually perform good either in trending or in ranging markets....but not in both types of markets. Knowing in advance when to use which system is the big question. And if you have the answer to that question in realtime, you will do very good in the markets.