I think if you continue this long enough. a year or two, you’ll see a bias based on prevailing trend over that time period, assuming there is one. Otherwise the results should come out break even. minus the accumulating spread.
You’ll [I]probably [/I]find a much higher correlation of win rate against shorter term trend timeframes rather than the long term bias.
If you think about this as extremes, if your trades lasted 1 minute, either because you used very small stops and targets, or you just closed them after a minute regardless of win or loss, then there very little bias derived from the trend in a montly chart, it’ll be 50/50. (Lets ignore the obvious impact of the devestating effects of spread in trading at such a high frequency)
The trends occurring in faster timeframes have much more of an impact than long term drift. If a markets moving up in some timeframe, and you take a bunch of buy and sell trades at random over that time period, and the duration of those trades is long enough to take bite sized chunks of that trend then over a statistically significant sample size, the win rate from buy trades, is going to be greater than the win rate from the sell trades.
Of course you still have the problem of defining what a trend is, but establishing what timeframe you should be using for trend determination for a given system, and then fine tuning system parameters to lock onto those trends really is a key stage in system design.
Trading random entries (without fiddling with them or imposing artificial constratints) is an enormously useful first step in establishing this framework. You might argue that its a little more complex because the nature of price action differs between long and short and impulse and corrective waves, but that stuff isnt really necessary if you just want to make money from playing with probabilities and tossing coins.
The reason that I said you’ll [I]probably[/I] find a higher correlation is of course that your given system may well be synchronised with the longer term trend.
People tend to make fun of random entries, but I’ve always found working on these ideas remarkably useful. For example, if I determine a bunch of performance metrics such as win wate, or maximum adverse excursion for a random entry, and a particular exit strategy, then unless a proposed entry strategy based on technical analysis, or price action can significantly improve those metrics, then the entry is invalid. Replacing the component parts of a system with parameters selected at random can be extremely enlightening.
you make some good points there. I was only thinking about daily trades.
Simbafx, you are actually right on target with the findings so far. We’ve gotten stopped out on the last two trades, BUT, both trades were positive for enough to make it work at one point.
Had I employed my standard strategy of moving the stop up to cost + 1 on both, or just closed them out manually, when we were up, we would be positive overall. Thinking about it in hindsight, we would have been up a decent amount considering I have it set for $0.50 a pip.
As far as frequency of trades, my usual trading style is short term, larger lot sizes. I don’t worry about paying the spread, I look for my basic setups, and pull the trigger. In, and out. I have realistic expectations of getting into an occasional unrealized loss, but I don’t average up or down. If it goes the wrong way, I look for the areas of retrace, and happily pull the trigger on a small loss, or b/e
So far this test has given me a few ideas, and after they have proven themselves further I may implement them into a more solid trade idea.
I will also be a lot more on this next week. Was short of time this week, and I spaced it a couple of nights.
I’ll post some pics of the trades on the charts later:)
Cheers!
Tails tonight…
Dang it. I dunno here.
Looks like I’ll just let it run for a bit.
We sold in at 1.3460, I’m just gonna look for 10 pips tonight.
Big round number of 1.3450 is gonna be tough on us.