Hello Tommor,
Mean reversion strategy
would it work better on daily trends by closing below the 50 ema on an uptrend or vise versa? Or are you using it on everything(Range, Consolidation, Non trending pairs)
I am looking back over some pairs and I see a lot of losses if you bought right at the close
Also when you say 1% below price for Stop loss do you mean price - 1% as in image I have posted?
Cheers
Hi @fishmoyne1 - yes, itās more than possible that some simple additional filtering would raise the win rate. It seems so logical it must be probable?!?!
Yes, a later close should increase average winner.
In both cases I am trialling on maximum simplicity for a while and then will see how to finesse the strategy rules.
I will certainly be looking back for the criteria you highlight.
PS, yes, my 1 percent entry to stop distance is 1 percent of price.
Iām wondering what the overall R is, if the take-profit is at the first profitable close and the stop-loss is at least 1% of the price?
By overall R, do you mean overal risk or overall return? Orā¦?
Iām just testing out this strategy, targeting the major, minor and Australasian forex pairs plus some stock indices and all I can say is so far is that it has never gone into drawdown. It appears that the average winner will be much smaller than the average loser, but that the win ratio will be very high, Iām thinking after a few months itās going to deliver something like an 80% win rate.
I am also still testing through what to do when you have an open losing position in one direction and the chart signals a new entry in the opposite direction.
Too early to say anything except this simple system has potential.
Sorry for the ambiguity, @tommor, I meant ratio of reward to risk (or vice versa).
Thanks for your helpful reply. Itās very interesting and itās one Iād like to try. Good luck with it, and hoping to see more statistics later, when you have them?
I think the risk:reward ratio of each individual trade is going to be low enough to give many traders a heart attack - it will never be as high as 1:1.
But the aggregate, i.e.total capital lost to total profit gained will definitely be OK, maybe 1:2 or 1:3.
Or at least, as far as I can tellā¦
Thanks, that wouldnāt put me off.
Iād just be careful with position sizing, because with a win rate of 80% youāre still going to hit 6 losses among 9 or 10 trades at some point, and you need to allow for that?
It would also be interesting to test the effect of taking profits at the second or third profitable close rather than the first?
Thereās a lot to test, here, but all very interesting.
All correct. Of course, taking profit at the first profitable close might strangle the trade but itās convenient for me as a simple parameter for the first test. Equally, the mean reversion entry signal could be taken as the start of a much longer-term open-ended position at an early point in a new trend. Lots of possibilities.
Another thought about the high win rate. With the mean reversion entry pattern being so reliable, I have already actually used it on a medium-term long position on the Nasdaq to highlight an exit point ahead of a pull-back. Choosing an exit point on a winning trade is always difficult and I am confident aleady that a mean reversion signal to go short will be more reliable than an imminent resistance level.
Terribly difficult! To be honest Iāve always found it about the most difficult thing there is, in trading.
I follow the logic. Interesting stuff - thanks.
Another thought on mean reversion D1 trades.
A simple tactic to deal with situations where price has not immediately reacted to an āextremeā close and produced a profitable exit opportunity - close position after 3rd close following entry.
Donāt wait for the stop-loss to be hit. Most winning trades are exited on the 1st or 2nd close after entry.
Iāve actually been wondering whether a ātimed closeā - in profit or loss - after a set number of bars would work well with this.
It would be possible to ādivideā the position, too (e.g. close a third after 3 bars, a third after 7 bars, the last third after 10 bars, or whatever???).
First, there is no way to predict the direction of the next candle. Because the reason for the move in any direction is most often subjective, made on the spot. Also, if you research a lot of numbers/directions, the answer will be 50/50. This is due to the law of large numbers. You can get the result you want if you use little data. If you use a large amount of data, you will find the direction of a candle is 50/50. No point in wasting time on this research, the answer is known.
In fact, the proportion of candles which show a higher close in an uptrend is about 55%. Not as high as many people suppose, but enough to give a margin over a random entry in a market which is not trending.
Yes, I remember in one of Van Tharpās books he gives 56% as the average figure measured over various data sets of tick data from 20-25 years.
Brooks also makes it about 56-57% over huge data sets of 5-minute bars, in his second book.
As you say, itās not as high as most people think (and would like) but itās certainly not 50/50!
Thanks @TruncatedUsernam .
Your post has reminded me to update here.
A further thought on the simple D1 mean reversion strategy I am trialling.
As the entry signal is inherently counter-trend it seems logical to avoid those which signal entry against particularly strong trends. I am tracking the strengths of trends on those pairs Iām getting into now - maybe results will show best win rate by taking trades where there is no trend, but maybe winners could be smaller.
We shall see.
Lol, interesting: in one of the places I worked, we actually once got a trainee/intern to check this out over 1,000 bars (10 sets of 100 consecutive bars randomly selected from trending periods, which we defined for the purpose by taking one of Chandeās definitions: SMA-20 above SMA-50 with both rising ,or vice versa).
We did a bit better than you guys are referring to: 585 bars out of 1,000 were directionally with-the-trend.
1,000 bars is nothing, really, but however you define ātrendā it will certainly never be 50/50 with any significant set of data.
It makes sense initially, that the (percieved) probability of a green candle is bigger in an uptrend, and a red candle in a downtrend. In the former, there is optimism, momentum and all that. People want to buy, and they are willing to pay higher prices. So in that moment, in that frenzy if you will, the probabiity is most likely higher for a green candle. However, over time, and time plays a role here as it does in all probability cases, this will mean reverse. For every buyer, there is a seller (in regulated stock anyway), and at some point, the temptation to sell might match or overcome the interest in buying. And we have a reversal. So in the big picture, long term, the probability is at least close to 50/50.
Maybe the probability of a forex price moveās direction over the long-term is indeed 50%, assuming that the long-term includes both trending periods and non-trending periods.
But within a trend, the probability of continuation is about 55%. So the big picture is too big.
