Discussion in 'Index Futures' started by aphexcoil, Oct 30, 2002.
yo coil, what's yer website
True. Swim Aphie! Surf!
APHIE: i read your posts on 10-30 at 5:21 and 5:37 pm and while i do not know statistics like you i see similarities to my trading plan . It can make money and good money if traded in 3-5 lots BUT you gotta follow through. I wish you luck dude because you are a smart person ( much smarter than me) BUT once again until you can implement it into action it can be frustrating. HOWEVER I DO NOT WANT TO BASH YOU ...... ONLY WISH YOU BEST OF LUCK..... ALSO,i would be interested in seeing your website. good luck and good trading
aphie, don't play the offended guy here.
several are only pointing out you are not doing what you said you would do.
or said otherwise, you do the exact opposite of what you planned.
this is the pattern killing you now. posting on ET or Trading.
Sure you need a minimum of statistical study. but the problem is you are doing it before trading.
great ideas on paper do not always translate into a winning strategy.
first because you may be bright, but they were many very bright people studying this for decades now. many ideas are already at work or arbed out.
So the cruel truth is this :
you need to trade, lose quite some money and then (eventually) burn in your brain what does not work.
THEN, and only then, can you study statistically your idea of what might work for you.
I know when I write this, it does sound strange and a bit silly as a course of action. But again, all the successful traders I know (OK I don't know that many, they aren't that many) had great (or stupid) ideas, then lost a bunch, then built totally different strategies because they notice the errors they made.
so they found what worked for them, not in theory.
Funny enough, some do go full discretionnary afterward OR highly strategized if not automated.
We predicted from DAY ONE what you would do and how you would feel, didn't we ?
So I would expect you would start to listen now.. well, the pattern is : you won't listen.
PS : for the others, maybe we can lighten up on the language.
tell me you were never like aphie at any point in your carreers?
You're doing some interesting work here. Please don't let the mostly well-intentioned criticism get you down. Everyone has their own path to trading success. If these analyses help give you the confidence to stick to a trading system, then do these analyses.
<b> DANGER! FUTURE-LEAK! DANGER!</b>:eek:
One big caution, though. You may be about to create a "future-leak" in applying the results of your analyses to a trading system. A future-leak occurs when knowledge about the future is used to influence the trading decisions of the past during backtesting. If you take the results of your statistical analysis of Oct 2002 volatility and use it to backtest on trading days before or during the analysis period, you will have a future-leak. To put it another way, if you can tell me the statistical properties of November 2002 data before Nov 1 2002, then use the system.
The solution is two-fold. First, doing analyses of disjoint datasets will tell you if the statistical properties that you use in trading are stationary (i.e., that do not change much from month to month). If the statistical properties of a dataset don't correlate with the statistical properties of the preceding disjoint dataset, then you have no basis for using the results of the analysis for future trading. Will November 2002 volatility dynamics have the same profitable quirks in them as did October 2002????
Second, if the system passes the first test, then use walk-forward analyses to recalculate key market properties for forthcoming trades. One does this by reanalyzing a moving window of data to derive statistical properties used to adjust the rules for upcoming trading days (e.g., October's volatility stats would be used to adjust November's stop-loss limits). This is the system that you backtest: the combined reanalyzer-trader system.
Analyze well, backtest well, build confidence, trade well,
I have tried very hard to check, recheck and check again the results and make sure I am not using any type of future information in checking out various ideas.
When I check out the expectancy of what i'm testing, I try to be very liberal with slippage and account for .50 in slippage. I will try to err on the side of being conservative.
The next thing I want to do is get as much tick by tick data from eSignal and be able to test that data since it has far more information that just 1 and 2 minute candles.
My criteria for scanning for a possible trade is trend based -- where I am looking for breakouts and bullish patterns in the candles over multiple candles.
Unfortunately (or fortunately), any system back-tested by a computer should probably be traded by a computer, which is something I'd have to program over time and set up and observe how it does in a simulated environment.
If it worked in a simulated environment for a month, I'd be willing to test it with real money. If it was profitable, at least it would totally remove me from the equation, except to monitor everything to make sure it doesn't collapse.
I couldn't disagree more. I think it's naive and stupid to jump into the markets without adequate preparation and planning. It is extremely valuable to take some time to develop and test your ideas and trading strategies, and become comfortable with a trading style BEFORE you risk your own money on it.
This is especially true for lesser capitalized beginning traders, who could easily make stupid mistakes early and blow out their accounts, never to return to the trading arena. Certainly, many traders have gone down this route and been successful, and are now inclined to support this path as the 'best' way to go. I disagree.
Keep with your testing Alphie. Once you have something that looks promising, then, and only then, commit a small amount of real capital to the strategy for 'real world' feedback. Going in this path, IMO, is the best way to become successful without a learning curve which includes a fat 'learning drawdown'.
Markets opening... gotta run.
Best of luck,
I agree on this. You can't spend too much time with the study of your trades. Real trades.
This feedback process allows you to improve your trading plan/methodology (when you have one and stick to it )
feed your data to a neural net and let it trade for you.
Separate names with a comma.