| Author |
Message |
     davkell
Member
Username: davkell
Post Number: 356
Registered: 07-2004Rating: N/A
Votes: 0
|
| | Thursday, February 02, 2006 - 03:29 pm: |  
|
Why is it, that my system seems to work 9 time out of 10, whenever I backtest it, over varieties of stocks etc, but then when I actually trade for real, with absolutely ZERO difference in the setup, the trade goes AWOL!
Is someone looking over my damn shoulder, and whenever I put real money down, they take the trade by the balls and drag it to oblivion!!!!!
Psyched Out!
Cheers.
"Trade Your Way To Financial Freedom" - Van K Tharp
"Manage the downside; the upside will take care of itself" - Donald Trump
|
david_louisson
Member
Username: david_louisson
Post Number: 194
Registered: 02-2004Rating: N/A
Votes: 0
|
| | Thursday, February 02, 2006 - 06:44 pm: |
|
Hi Dave
I have been paper trading since early December, and (closed trades only) have a win rate of 43% (19 wins, 25 losses). My ave win size is 1.50 times my ave loss size, so my profit factor is 1.12 – hence I'm ahead on balance.
What I do is run a post-mortem on each individual trade, and try to work out how I could have done better (increased gain on winning trades, turned losses into wins, etc). I'm swing trading (imaginary) UK sector index CFDs – most trades last 2-3 days from entry to exit, longer if the trend persists.
Here are the types of analysis (thought processes) that I run:
* Did I enter too early (i.e. need one more price bar) or too late (one less) to establish the current move? Earlier = more favorable price, works better in a "ranging" market; later gives greater confirmation of trend = works better in a trending market?
* How many trades were won/lost trading with/against the market, and/or the longer term trends?
* Did I set my stoploss too tight (got stopped out prematurely, from what would have otherwise become a winning trade) or too loose (giving back too much profit at the end of the trade)?
* Would I have been better taking full or partial profits along the way? (this avoids any problem with overnight gaps)
* Which indicators in my setup are giving the highest number of wins (if I was to exclude each individual condition, what effect would this have had)?
* Am I sizing my positions correctly – risking approx equivalent % of total capital on each trade? (or is one loss wiping out the profit from several prior wins)
* Should I be hedging (using simultaneous long and short positions) more, or less? (Helps to combat market risk).
* What effect does trading long on only the strongest sectors (relative to the market) and short on only the worst performing sectors have? In other words, does this work significantly better, or is diversification more profitable, to mitigate risk and ensure that I catch all of the big moves?
* Was it bad luck (prices severely affected by unexpected news, causing a spike or gap), or some kind of market manipulation going on?
* What is being lost in transaction costs? In profit forfeited in waiting for the trend to be established and/or forfeited waiting for the reversal to be confirmed (i.e. retracement to trailing stop)?
* What effects do varying all of these parameters have on win rate? Ave win size to ave loss size? I think its important to set goals for each of these – mine are 60% and 1.6 respectively – must be realistic in terms of system parameters, time frame, etc.
If I was trading stocks for real, I could also analyze effects such as:
* Slippage – where are my orders getting filled
* Variations in the prices quoted by the bookmaker, and the OHLC from my data supplier (currently a huge issue, but paper trading ignores this)
* Other problems relating to the on-line trading platform
The main problems I currently seem to have is missed breakouts (I have instead been trying to trade pullbacks), and stoplosses that are too tight (at least 5 losers that would have become winners), and what I perceive as being abnormally high price randomness (not much I can do about that one).
Depending on your methodology, your analysis will of course differ, but these ideas should hopefully help you get started. If you have a large enough database of wins and losses, patterns should emerge, which will allow you to adjust your system parameters.
A large track record, across different stocks, time periods, and market conditions, engenders confidence. The greater your history, the better your "feel" for what is more likely, or less likely, to happen. Even the wizards encounter long runs of losses. Experience will tell you whether this is a "freak" happening, or likely to be a common occurrence.
There are undoubtedly a hundred other ways of going about this, but this is the best approach that I can think of.
I am finding that short term trading is an extremely frustrating and tough game. Just getting a system of short term entries and exits that delivers positive expectancy (greater winnings than losses) across different market conditions is hellishly difficult. I'm still not there yet. And they say that, compared to the psychology/discipline angle, this is supposed to be the easier part!! I've been seriously tempted to chuck it all away on so many occasions. But I console myself that the Market Wizards have apparently done it, some after several false starts and blow-outs, so I will keep persevering. Perhaps I am being naively conceited, but I back myself to eventually crack this problem, albeit in my own way. I have nothing to lose, and potentially everything to gain. As Linda Raschke says, you only need one pattern that delivers positive expectancy, and the discipline to trade it consistently, to make a fortune.
Good luck!
David
|
chart_rider
Member
Username: chart_rider
Post Number: 154
Registered: 01-2005Rating: N/A
Votes: 0
|
| | Thursday, February 02, 2006 - 07:56 pm: |
|
Davkell
A couple of points came to my mind, mentioned in part above also, but just to highlight:
Make sure you extract details such as largest drawdown and longest losing streak from the backtest. Sometimes a system with good overall return is not easily tradeable due to excessive drawdowns along the way, thus requiring some adjustments in the rules.
Check the XAO conditions in times when the backtest shows good returns and in times of bad returns, then compare those times with the current situation. This may shed some light on what is happening. I incorporate the XAO into backtesting routines and find it has significant effect on the outcome.
Most importantly, the backtesting results are generated from a large sample. It's therefore necessary to have faith in the long term success of the system and be careful not to lose this faith after a short trading run.
CR
|
mustang
Member
Username: mustang
Post Number: 13
Registered: 05-2005Rating: N/A
Votes: 0
|
| | Friday, February 03, 2006 - 10:36 am: |
|
Hi Guys,
In "Trading in the Zone" Mark Douglas strongly suggests making at least 20 live trades in order to evaluate your system. I'm 6 trades into an EOD trend following system and having fun with it. I'm keeping position sizes small and tracking each trade and the portfolio as well.
David mentioned Market Wizards and my favourite interview would have to be with Tom Basso. Is advice, to treat each trade just like any one of the next thousand you'll enter, really hit home for me and fits in well with what Mark Douglas says too. I read it whenever my faith in my plan is being tested.
Cheers,
Mustang
|
|
