Neke: A great way to gauge all of the criteria Lucias mentioned is with your sharpe ratio. It will be a one-stop figure for roughly determining how much leverage you can use without blowing up - it will also give you a hint on how viable a system is (a terrible ratio, such as <0.5, is hardly something you can profit from). I wish you (neke) would listen to the troves of good info you've recevied. It seems like all you want to do is keep trading your laughable garbage system kicking and screaming, without taking the time to devote yourself to new systems (they never last forever - YOU need to compensate for this!) and becoming stronger through knowledge and practice you acquire on these priceless boards.
Focusing on the PF is not the most important factor, but it is an important way for me to judge the quality of my (discretionary) picks. Being reckless and taking marginal and even decidedly negative trades (mostly on the fade) has been quite harmful to my account. Not only could it lead to losses, but it tends to lead to aggravated losses through negative psychological follow-thru (revenge trades, doubling down on losers). Over a large sample of trades the effect to my account is significant. I want to cut down the chance of that happening by being more selective. I understand the sample size is smaller, and therefore possible things might not go the intended way - but that is a risk I can live with. I have risk level on a per-trade basis. My expectation is that on low conviction (LC) trades I risk 1 unit. For MC trades, 2 units. For HC trades 4 units. The size of a unit will be determined and ramped up as confidence/account grows. Right now I would say 1 unit is equal to about 1% of account. Enforcing this for my discretionary trades requires some work on my system: I am still trying to see how to get the computer help me enforce this.
Multicharts or Wealth Lab Pro are the only software packages that will help you "enforce this." Your profit factors are low because your timeframe is short, but unless it's above 1.3, the losses from lack of edge will eat away most of your profits. The sum of the net profit column by the sum of the drawdown column is probably below 0.1. Your APD stat would also be low in light of your profit factors being below profitable enough levels for decent high frequency trading results. If you use Multicharts, this calculation would be especially easy, but very difficult to get data on the trades you're making or will simulate. Both packages have portfolio simulators, which is superior to every other product for this reason.
I have been thinking about this lately - eliminating LC trades and putting more effort to identifying a few extra HC and MC trades. This is how I look at it. A single HC trade (with average 2.50 PF) is worth about 15 LC trades (with a mean PF of say 1.10) based on similar sizing. But given I am willing to risk 4 times for an HC trade vs an LC trade, one HC trade might end up being worth 45-60 LC trades. That means the 320 LC trades are at best equal to say 6 or 7 HC trades. That is too little for too much activity being involved with them. Instead I should conserve my energy (each position requires identification/initiation/monitoring/exit) and devote them to looking for HC/MC opportunities. So yes, at the end of the year I might have 30 HC and maybe 120 MC trades, which would be worth more than the earllier scenario.
Whatever statistical analysis you are using to classify HC/MC/LC trades is curve fitting, and part of your consistency problem because they are almost certainly based on extremum parameters in your model. The curve fitting process is done by way of your method of classification, and can only be rightly processed with software to do so that can also be backtested. The validity of your model must be in question in your head, but whether you want to believe it or not the foundation of your strategy is indeed a curve fitted result of data mining, not just the curve fitting itself. The only way you can escape this problem is by backtesting, not statistical analysis because you're doing the trade classifications incorrectly since the trades you perceive as either HC, MC, or LC are results of a massive curve fit and whether you have backtests showing your returns they probably are not that good. If you were to share your models, a lot of us would know what you're working with, and only performance summary statistics are necessary. Without knowing win percentages of each category it is impossible to know how much the percentage profits on each category is, and if you know this, it would help your credibility a lot to disclose this.
Neke - There are lots of ways to make money in trading. With your experience you should be able to hunker down and write up some solid systems with no curve fit data in no time. You're not a noob - you should be able to know and perform the necessary steps. Instead you want to be stubborn: not listen to the mountains of golden advice you receive, not do whats necessary, keep using your system that I would smile and throw away if it came up in my results, and cry about it on the way... buddy wake up!
Weekly Update for week 3/50 ended 2/4/2011 Disgusting week, down 3.8K (2%). Was having an OK week, until Friday. A short position cost me a bunch and reversed my gain for the week. Took out 6K for the month ended 1/31/2012. Code: Opening Balance: 198,139 Net loss for the week (3,770) Cath Withdrawal (6,000) ------------------------------------------------ Net Balance: 188,369 Since Inception of Thread 01/18/2012 - 02/04/2012 Opening Balance: 203,729 Net loss (9,360) (Down 4.7%) Cash Withdrawal (6,000) ------------------------------------------------ Net Balance 188,369
Seem to be running scared. A $4,000 loss means $6,000 withdrawal? Without more information, there's no way to tell what went wrong, or if you didn't notice the uptrend since mid-December to reverse and go short. There are some individual stocks that might have worked on, but until you quit ignoring your readers, the march to $100,000 is on.