I take it this is a multiple entry, cost averaging system. Otherwise your entry timing needs to be bang on and your entry criteria is vague.
Apologies if I missed something, there’s a lot in there and kids have been distracting me
Same thread by same name on FF, not sure what he is selling but it’s coming…
2 Likes
Yes, you need to cost average. This is why the 2 core rules of the risk management is STAY SMALL and trade often.
You can only do the latter if you are already doing the former.
Additionally, the smaller the spread, the larger your VAR will be. So your initial entries should not be your larger ones.
2 Likes
I have an EA that trades this. It’s basically the kof strategy simplified from @AmericanTrader thread. And I have some bad news. It doesn’t make much money.
Tell me your settings for when to make first entry, when to add and what size. I’ll plug them into the EA this week and run a back test. The win rate was really high, but every now and again, a stubborn trend wipes out almost all of the profits.
It is possible to still make money on these trends with mean reversion, but it requires a lot of time to keep taking profits on the pullbacks and re-entering. Needs very good judgement of price action. I don’t know how to program that
3 Likes
thanks - that’s what i was wondering
this part was already fairly clear 
well - that’s “cost averaging” for you, isn’t it? 
thank you for sharing, this sounds easy… and well, it is. i trade similar but i don’t use MA instead i use a Bollinger band (which is also an MA if you take it serious) and i only enter markets when the price is above/belowe the middle band because, as you said, price will mostly return at the median/fair value.
i also use the ADX to get a confirmation because this strategy only works in ranging markets (guess what, we are talking about mean reversion). i only trade when the market is NOT trending and have similar winning rate i think it was about 71 - 76% winning trades.
I’m wondering myself if MA/fair value would also work when market is trending🤔 would love to read further details.
happy trading my friend😊
2 Likes
you could use Bollinger band for starting point and sell when price is near uppper band, it will mostly return to the middle band.
1 Like
@this_is_ando It’s a good question to ask, “will it work in trending”, but when it comes to currency markets, there really aren’t trends.
This is the distribution of essentially every currency pair:
It’s very close to normal except one consistent divergence, kurtosis. But besides this, the skew is almost always 0, the same as the total expectancy of the distribution as a whole.
In a normal distribution, Kurtosis remains at 3, in currency prices, it’s consistently over 3, even when you extend the duration.
It’s consistently higher, meaning there are large moves that occasionally occur in both directions. The total skew of these moves is still 0 though.
This data poses a good question for currency traders to think about:
What’s the difference between what most would call a “trend” and a “tail move”?
The answer is in the skew… Most typically confuse the two.
So we will have moves that look trendy, but they are essentially tail moves. This is why one of the most fundamental rules is to stay small so you can trade often and not be blown out by tail moves, because I can guarantee you they will come.
4 Likes
I was going to ask if somebody was maybe using BBs for entry signals.
1 Like
What is your graph showing? An unlabelled chart is hard to understand what it’s telling me.
Not sure what you mean by there’s no such thing as a trend. The USD has been rising vs most of the major currencies for over 10 years.
I’m not saying this can’t work, but your criteria for entry are non existent, and going with the trend, short or longer term is almost certainly going to be more profitable. Especially with cost averaging
It’s a distribution of daily price returns (in pips) for EUR/USD.
It will help to understand what I am saying if you learn more about probability distributions, specifically, the law of long leads, or arcsine law.
Even though a distribution is perfectly fair or normal (flipping a coin to create price paths), the least likely return value it is to have is 0. 1/2 of its values will spend most of its time on one side of the starting point or the other.
All of these moves look like “trends” going up or down. They are not though, they are long leads.
So the question is, what’s the difference between a trend and a long lead from a sample set?
1 Like
What time period is that over? I don’t need to study statistics to see that over 15 years EURUSD has posted lower highs and lower lows continually. If you’re telling me that 15 years isn’t a trend, then we’ll have to agree to disagree. And no matter what you call it, you shouldn’t try to catch a falling knife
1 Like
lol when did I say to catch falling knifes?
Also this is a “strong trend” for EU since 03?
You seem to have a comprehension problem. You should re-read what I wrote.
Your lack of understanding regarding probabilities and statistics will be dangerous to you. It’s akin to using 10 years of data to estimate the size of a 1 and 1000-year flood…
“The flood levels for the last 10 years never exceeded x feet. The trend is obvious that floods don’t go higher than x feet so we can build the levy there…”
If you slice from the peak, it looks “trendy” but that is a single sample path. The law of long leads proves that even in a perfectly normal distribution with no skew, 1/2 of the sample paths are above the starting point (like an uptrend) and 1/2 are below. When you combine them though the skew is 0. There were no trends. Only long leads.
1 Like
So here are this week’s trades using your strategy (i haven’t traded today yet and prob. i won’t trade friday). But I’m looking forward to run further tests.
as you see I’m still having problems being confindent enough to let winning trades run / setting higher TPs (it’s demo account tho but we learend to act like demo is real money :D)
In fact it’s similar to a “strategy” i posted a few months ago where I suggested/wondered if trading in the opossite direction would gain returns - it worked, while some guys just thought it’s coincidence or “unexpected market behavior”. Trades go ALWAYS in the opposite direction
next steps are:
better RR-Ratio
setting higher TPs
testing also on higher timeframes
profit: 91,70 Euro
most of loosing trades would have turned into winners if i had set stop loss just a few pips wider (2 - 5 pips, haven’t took the spread in account when setting SL 
)
1 Like
Just bumped on this thread. The author is quite a statistician who just do not have the necessary teaching skills to explain all the probability jargon to their students.
I would suggest explaining in “english” so that everyone can get your point, rather than losing your audience in a jungle of statistics jaw breaking language.
Your strategy makes me remember the WSS (World Simplest System) on FF. I have used it several times with some nice result.
1 Like
I used to be a big fan of mean reversion (and still am to an extent) back in the day. One of my first strategies that I tried was using mean reversion to buy/sell FTSE vs DOW based on a correlation coefficient of about 0.9 at the time.
Anyway, with that in mind I thought I’d quickly knock up this strategy in ProRealTime to back test it. There’s probably some merit in there but I don’t have time to play around with it too much at the minute.
This notion of ‘fair value’ and ‘seasoning’ is a little vague and I’m not really a fan of vagueness in strategies, you should be able to write any strategy down and it feels too much like a hunch or a feeling for me.
BUT… 55% win rate is pretty good, some very good daily profits there, it’s certainly something to work with, it’s just as per the written rules is loss making over the longer term.
I’ve had a bit more time today to play around with this and I’ve managed to adjust a few things, (add trailing stops, remove some of the time constraints etc etc). Backtest results look very good, too good, I used a 1 pip spread but haven’t used tick by tick. Although this shouldn’t make too much difference with something like this, only reliant upon a closed candle.
I feel I’ve overlooked something in my settings or code, the results are too good but I just can’t seem to see what I’ve missed.
I’ve a bit of space in my demo account at the moment for trying new strategies out so I’ve added this bot and will report back with findings. Hopefully this will allow me to see what I’ve missed too!
Ah ok, I think I’ve seen something, so, the backtest was using a weird time period for some reason, I’ve adjusted this to run from 2018 to present day, I’m currently just simming a walk forward test too.
This will probably only work in a ranging market such as where we currently are, if you’d started this back in 2018 you’d likely have given up by 2019 and certainly by 2020 it’d be in the bin
EDIT**
Got it, was a bug in the way I’d coded a trailing stop I was trying which was then erroring out when I tried to run it in a demo account. Having rectified this, below is the very best I can achieve with this strategy. It might make some money at some points in time and it generally won’t lose a lot which is great, but it’s not something I’d want to be deploying full time.
