A real trend following model will spot the change in real time as it happens. Or upon re-optimization give you that change in real-time. Your model is late, making it a lagging trend follower at best, and more of a breakout system at worst.
As with putanski it isn't possible to have an accurate trend model on weekly/yearly data. The lag is very apparent because there's so few data points. When I optimize, I use 2 years of constant volume bar data on whichever interval of 7 gives me between 5,000 to 15,000 bars, then optimize ande re-optimize periodically on those or until there is a losing trade.
It has been working last 40 years on 25 different markets. It has been profitable on S&P last 223 years. It works on majority of stocks. It has been working on S&P 2000 - 2010. Thanks for your insight, but I will stick with my system.
Define "start". If the ES is going to go down 10 points, ultimately, and I get in after 3 of those points have already occurred, is that me getting in at the "start"? And even if I got in 7 points after the turn and captured only 2 points of the 10, if I can do that consistently, there is no end to the amount of money I can make. Do you disagree? If the OP can't do that, it's because he's got a bad system, not that the idea of attempting to do that is a bad idea. Also, why would you not wait for a confirmation (however you define it) before entering a trade? That doesn't, of necessity, lead to "chasing price". In fact, by requiring confirmation, you do less trading and by passing on potential entries, you are actually doing the opposite of "chasing price" by simply remaining on the sidelines when confirmations aren't present. All systems rely on confirmations at some level. Obviously, the best systems are the ones which combine high accuracy of confirmations with timing as close to the actual reversal as possible. In the ES example above, maybe someone else could have a system which only required a 2 point move as its confirmation. I'm highly skeptical that there is any working system which can do away with confirmations, though, although I would guess there are nearly an infinite number of systems which don't work and which also don't have confirmations built into them. If you look at the number of points in hourly fluctuation in the ES over a month, it numbers in the thousands (taking the hourly high and the hourly low and just adding the absolute values up). If you can catch even 1-2% of those fluctuations, you will absolutely rake in more money than you can spend. In that context, it doesn't make a lot of sense to try to catch the top or bottom ticks of a trend. In fact, I would state that, as a rule, the more you try to do this, the lower your accuracy will be.
So Eckhardt, Dunn, Abraham and hundreds of others are not "real trend followers"? And their systems can't work? They're gonna be devastated.
You are wasting your time with someone who cannot count. He thinks a drop from 500K down all the way to 200K is a 40% drop. He is a distrurbed joker with a vulgar name "lucky puta from Poland", as this is what luckyputanski means and his trades are random. He has no concept of trading and the markets when he is talking about 223 years or something like that in S&P 500, because the index was introduced in 1957. Thus, he has no idea of what he is talking about. The sad thing is that ET allows and tolerates this disturbed personality.
Start is within the immediate proximity of a turn. It shouldn't take more than 5 points to get confirmation, but most strategies that are supposedly "trend following" do wait a set number of points before entering. It's that literal calculation of "well, how far up has this come" that I should be thinking it's time to go long or vice versa. Having a consistent indicator of that is much more important than getting confirmation from a set number of points before entering. The universal trend trading algorithms I have can identify every type of oscillation in price, but, realisitcally, there are only four: higher lows, lower highs, breach higher highs, breach lower lows and if you understand market mechanics you should know what to do on each one without having to wait and give back what could have been more profit if you're right about calling the turn. The term "start" is would you rather be late after confirmation or early? My tendency is to prefer to be early, and typically I'll see the turn in real time so just having hundreds of hours watching futures for those oscillations makes me aware that that is the only acceptable theory of price action.
Probably. I didn't know what he meant by 223 years. Maybe 223 weeks? 60% dd is pissing away a lot of money. I didn't get the equity curve chart since 2000. Like if he's down 60%, how can that be his equity curve? Looks like there have been losing years even, not just losing quarters. Thinking that that should have been an indication something wasn't working. But you're right. I probably am wasting my time. With a journal entitled Universal Trend Trading I expected the author to have a better model than the one that's got him down $300,000.
223 years - I was mislead by this: "http://stooq.pl/q/?s=^spx&c=mx&t=l&a=lg&b=0" Probably it's some composite index, as they do with euro. SP chart for last ten years - it's only S&P. I trade 20+ symbols in reality and that's in this journal. There have been losing years in the backtest. I gave up a dream of profitable every quarter long time ago. No one with public record can do it, only anonymous geniuses on ET can. 60% - yes, it is a lot, but that's the risk I accepted and trade. Biggest drawdown during backtest was over 70% That's it. I will not answer any more mean posts, as that's me who is wasting the time.