StockFetcher Forums · General Discussion · Guru_Traders Uncertainty Principle applied to Backtesting<< 1 2 >>Post Follow-up
guru_trader
485 posts
msg #40739
Ignore guru_trader
1/27/2006 5:25:49 PM

I read a thread the other day that got me thinking about backtesting. Shelupinin had just opened up a trading account with $20,000 and had lost several thousand of it within the first few weeks. (See thread entitled, "psychological problems after loose".) He asked why he would lose so much money on a trading system that backtested so well. Many great, and accurate, reasons have been given, but I'd like to add another.

I was thinking about the concept of backtesting last night and realized that no backtest can ever be accurate for the simple reason that a backtester's trades never actually influenced the movement of stocks over that period of time. It's just a side-by-side simulation.

It's much like "Guru_Trader's Uncertainty Principle (GTUP)" ... where, at any point in time, it is uncertain how your position in a system (market) will affect that system (market)! The system (market) will always be influenced by your participation (trading) in ways you may not be able to predict. The best you can do is calculate probabilities and odds.

Applied to backtesting, we can't judge the accuracy of our backtesting results unless our stocks are actually IN THAT SYSTEM!

It may be that, as our relative percentage ownership in the market decreases, we will most probably increase the accuracy of our backtesting. In other words, the fewer stocks we own in the market, the smaller our influence on the market (up to a certain point) and, the more accurate our backtesting will be ... at least, in theory.

This leads to another important corollary, backtesters who never put their money in the market will always have different percentage ROIs than those who actually have their money in the market!


judgetrade
106 posts
msg #40742
Ignore judgetrade
1/27/2006 6:22:08 PM

Backtesting is a start of the game, not more.
I would not trade without it, but I know that reality will be different.
I am happy if I make 25% - 50% of my the simulated profits in backtesting.

If you trade low volume stocks, your impact trading them will hurt your strategy. But if you trade high volume, you will not influence a lot, if you do not have to move millions in an out.

Take Momentum strategies for example: There is empircal research, that the excess returns of these strats get overruled by price impacts only if you trade with a portfolio bigger then 6 billion. Until then have fun with them!

As I said, with low volume stocks, the impact is bigger.
Nevertheless: You have to find out how robust your strategy is to real trading and you might have to adapt the strategy, even the simulated results will be worse.





judgetrade
106 posts
msg #40743
Ignore judgetrade
1/27/2006 6:35:06 PM

By the way: You can loose your shirt on strategies that backtestet monster gains on paper. Its easy to come up with a capital curve that seem to explode up.

It took me years to develop strategies that make money in the real world (I wished I would have seen this board before!) Beeing sucessfull in trading can not be learned over the weekend, it takes years (at least for most of the traders, read market wizards, only 2 out of 10 had a jump start).

A good start is: Find out what the science comunity (the real one, not the EMH dreamers) says about certain strategies. I do not trade anything that have not be confirmed by them.

Its a skill game, and competition is huge. Play poker with a 100 People, the same 2-3 persons will have the money at the end of the day. Trading is similar.
Most lose.

A lot of stuff works though:

- Low valuated stocks (PE, Cash Flow) combined with a scalable business model
- Mid Term Momentum (3-12 Months)
- Short Term relative weakness (go long on it, RSI (2) below 1 combined with panic selling situations, good volume and low priced stocks and fixed percentage bet size is a wonderfull start for small portfolios under 100.000.
- Earnings suprise drift (www.pitbull.com has a good application of this strategy) trading
- Insdider Buying
- Mid Term Industry Momentum
- etc., etc.

See you...







nikoschopen
2,824 posts
msg #40746
Ignore nikoschopen
1/27/2006 7:40:26 PM

It's called "curve-fitting", and it works as a frame of refrence for backtesters whose grand ambition (or illusion depending on how you look at it) is to churn out copies of its former suceess into an alien landscape.


rtucker
318 posts
msg #40747
Ignore rtucker
modified
1/27/2006 8:46:09 PM

d

nikoschopen
2,824 posts
msg #40748
Ignore nikoschopen
1/27/2006 10:02:31 PM

rtucker,

The theory behind RSI(2) < 1 is to bottom fish those stocks that are oversold, overdone, if not overtrashed -- big time. In such times, bottom fishers usually swoop right in to bring them back from their miserable holes. This works quite nicely in a bullish period, such as the current environment we're now in. But not so in a bear market, when the baby is often thrown out with the bath water.


judgetrade
106 posts
msg #40752
Ignore judgetrade
1/28/2006 5:00:42 AM

A trading system contains money management. Tight stop losses kill you under 3$!!! Stop losses is one of the things that backtest great on small priced stocks, but will kill you in reality.

I do not trade at all with stop losses, I choose my betsize very small and I ususally have 90% Cash in my rsi (2) portfolio! Only like this you can get a smoth equity curve.

So if you have 20.000, trade not more then 1 (take the strongest signal [biggest panic, biggest drop the last 2 days!] or a screen that will give you not more then 1 or two a week) signal a day and do not put more then 2.500 in one trade.
Use a broker that charges not more then 9 bucks a trade otherwise commision will kill you.

Trade stocks below 1, because they will have an even better win ratio.


Try this:

set{ae5, atr(5) / ema(5) } and average volume(10) is above 50000 and close below 2 and ae5 above 0.14 and rsi(2) below 10 and weekly rsi(2) below 33
price is between 0.1 and 1
and average price (2) * average volume (2) > 200000
and average price (100) * average volume (100) > 100000
and roc (1) one day ago < -5%
and roc (1) < -10%









judgetrade
106 posts
msg #40753
Ignore judgetrade
1/28/2006 5:02:40 AM

For less but better signals put rsi (2) below 1.


markcrisp
187 posts
msg #40755
Ignore markcrisp
1/28/2006 5:23:37 AM

I agree IF you are trading BIG. But ifyou ares small traderyourbuy/sell will have no effect on price.

I wasoncetoldyouback testforone reason:

1) If it is proftiable you MIGHT be onto something. Research further. But do not expect those results or anything like it.

If it is not profitable dump it and move on.

That's all. You gotta learn in this game.


markcrisp
187 posts
msg #40756
Ignore markcrisp
1/28/2006 5:36:55 AM

The RSI(2) Method.

Is this a day trading method or do you hold overnight?

What kind of profits do you get per trade? 10%,20%,30%+


StockFetcher Forums · General Discussion · Guru_Traders Uncertainty Principle applied to Backtesting<< 1 2 >>Post Follow-up

*** Disclaimer *** StockFetcher.com does not endorse or suggest any of the securities which are returned in any of the searches or filters. They are provided purely for informational and research purposes. StockFetcher.com does not recommend particular securities. StockFetcher.com, Vestyl Software, L.L.C. and involved content providers shall not be liable for any errors or delays in the content, or for any actions taken based on the content.


Copyright 2016 - Vestyl Software L.L.C.Terms of Service | License | Questions or comments? Contact Us
EOD Data sources: DDFPlus & CSI Data Quotes delayed during active market hours. Delay times are at least 15 mins for NASDAQ, 20 mins for NYSE and Amex. Delayed intraday data provided by DDFPlus