Universal Trend Trading

Quote from luckyputanski:

An interesting theory, but I find it difficult to believe. If a strategy was working 40 years on dozens of markets (and then checked for fun on different markets - working there as well) and had a new equity high not so long ago (just after I started this journal), I find it hard to believe that it stopped working exactly now. It is possible, but come on, what are the odds?
More...

You might have fallen victim, like many have, to the illusion that backtesting something must mean that it will work the same way in a forward test. The past 40 years will not be like the next 40, or the next 400. There is a reason the government requires the disclaimer ",past performance is not indicative of future results" for anyone attempting to sell their BS. Think hard about that.

Basically anything on past data is curve fit... this creates a giant problem. Do you think curve fitting is reduced by using more data? Could it make your backtest even worse? The answer is, yes, possibly. The fact is you don't know. Anything is possible.

I don't know the details of your system, but personally, from the personal real money I've traded, using nothing but logical risk managment you should never have a 50% drawdown... My entries have not been near what a full time trader could do with unlimited capiltal, yet still I haven't experienced a 50% drawdown in 9 months (yes this is a short amount of time). I have traded (most likely over-leveraged) a very small account of 10k-24k.

Remember, a good trade will never go far against you and the only constant is change. I'd be intersted if the system shows 50%+ drawdowns. Is it price based? If you are a long term trader, why are you in and out of positions that show profit within days?
 
Quote from masterm1ne:

You might have fallen victim, like many have, to the illusion that backtesting something must mean that it will work the same way in a forward test if you backtest it on as much data as you can get it. The past 40 years will not be like the next 40, or the next 400.

There is a reason the government requires the disclaimer ",past performance is not indicative of future results" for anyone attempting to sell their BS. Think hard about that.

Basically anything on past data is curve fit... this creates a giant problem. Do you think curve fitting is reduced by using more data? Could it make your backtest even worse? The answer is, yes, possibly. The fact is you don't know.

I don't know the details of your system, but personally, from the personal money I've traded, using nothing but logical risk managment you should never have a 50% drawdown... My entries have not been near what a perfect trader could do with unlimited capiltal, yet still I haven't experienced a 50% drawdown in 9 months. I have traded (most likely over-leveraged) accounts of 10k-24k.

A good trade will never go far against you.

The only constant is change.
More...
9 monts? That's how long you've been trading? I've been trading 7.5 years and I know that my drawdown depends on the risk I'm taking. I made a decision to trade the way I do and I know drawdown like this can happen. Of course backtesting proves only that an idea worked in the past. But we all have to learn from the past, future hasn't happened yet. As to my system: it uses 5 parameters and it yields profit on 96% of 120000 combinations of them. I'm using parameters that gave slightly better than average result in the past, so I'm not worried about curve fitting.
 
Testing it for the past 40 years is insufficient imho since it does not include the great depression which is where we are now (or heading towards).

In other words: Past 40 years have only been a bull market of some sorts.
 
Quote from Xena:

Testing it for the past 40 years is insufficient imho since it does not include the great depression which is where we are now (or heading towards).

In other words: Past 40 years have only been a bull market of some sorts.
More...
Below equity (no commissions) from 1929 to 1940: corn, djia, nikkei, silver, sp500 (should get more data, maybe will do in the future). It's the same system I trade now, no changes at all. Doesn't look that bad.
 

Attachments

Looking at that equity curve - the same propensity to trend seen in markets appears to exist in your system.

Which implies the question, why not run a trend following signal over a trend following system's equity curve?

:D
 
Quote from Wide Tailz:

Looking at that equity curve - the same propensity to trend seen in markets appears to exist in your system.

Which implies the question, why not run a trend following signal over a trend following system's equity curve?

:D
More...
Hmmm... Equity curve trading. I heard about it, but never toyed with the idea. Below some statistics of my trading system trading its own equity curve (arithmetic value):
Average move caught per transaction: 18.41%
Transactions: 181
Profit factor: 8.74
Expectancy: 4.87

Seems like a good idea to stop trading this until things pick up. The only thing remains to confirm is to add up short trades of this system (test above includes profit made on both, short and long trades).
 
You must log in or register to reply here.
Back
Top