System Trading
Now Reading
Learning To Read Part 2 – The end of the trend, that was your friend, until it bends

Learning To Read Part 2 – The end of the trend, that was your friend, until it bends

by ScottMarch 11, 2017

The textbooks would have us believe that markets are smooth and orderly. We have a nice downtrend and then when it ends we have a nice uptrend. In theory all we have to do is wait for the right time, buy, profit.



Reality is messy and choppy, and most of the time we are unsure if the old trend is in play, if we have transitioned into a trading range, or are trying to reverse. Each new bar of price action gives us critical information, until we can rule out each of these options. Eventually we get betting odds, and we bet.

Today I’d like to explore the tell tale signs that a new move has the potential to become a trend, and not just a counter trend rally. Then we will look at the signs a trend is ending, accelerating, reversing or transitioning into a trading range. You can use these principles

  • For discretionary trading
  • To get better exits on your systematic trades
  • To identify objective conditions in a trade with open profit where you would want to tighten your stop
  • To identify objective risk/reward points for system building
  • To stop you being a sucker and betting on trend reversal when it’s not going to happen
  • To identify those rare points where you want to be “all in” on trend reversal
  • How, exactly, to trade those “lottery ticket in your pocket” type trades where you get a good entry, for example AAPL in 2003, and you want to ride it all the way to the beach. For those times when banking 10R would be embarrassingly bad.

As we go more advanced with the tape reading you will see repeatable and identifiable conditions which become some of the highest probability outcomes in trading.


Oh SNAP! You thought you’d get it all for free. But I’m not that nice! I owe Mole a million favours (and so do you)… I don’t owe you a thing.

Here’s the deal. If you think the value I’m going to provide over the next 11 days until Mole gets back is useful to you, then sign up and give Mole a few dollars for all he does around here.

I’ll do my very best to answer all your questions until Mole gets back, and help you avoid some of the dead ends and pitfalls I wasted years of my life on.

Anyone who doesn’t feel they got value from my posts when Mole takes over, you can have a free 1 hour skype call with me (which I charge $200 for) to talk about any aspect of system building or trading you want to discuss. I can’t be fairer than that.

So buy now, or no more free stuff.


It's not too late - learn how to consistently trade without worrying about the news, the clickbait, the daily drama and misinformation. If you are interested in becoming a subscriber then don't waste time and sign up here. The Zero indicator service also offers access to all Gold posts, so you actually get double the bang for your buck.

Please login or subscribe here to see the remainder of this post.

About The Author

  • Fozz

    Clever there the way you drew us in, but the grab for the wallet was a bit obvious.

  • Scott Phillips

    Don’t care if its obvious or not. Pay or go somewhere with better content. I couldn’t care less

  • Sir Mole III
  • Sir Mole III

    I once again had to fix the featured image as it was too small.

    I happen to be a huge fan of Hokusai by the way. You guys probably recall the first one Scott had posted:

    What’s rarely discussed is that Hokusai often introduced the concept of self similarity into his art. So perhaps without consciously knowing he preceded Mandelbrot by 200 years 🙂

  • Sir Mole III
  • Fozz

    Interesting, two fish on one hook!

    Strange people who write blogs and then bitch ‘cos someone reads it and comments on the apparent abrupt change regarding a forth coming sequence of articles. If you don’t want readers then make it all pay to view. But better first to front up with your trading record and real name. The elephant in the room question is – which is most important subs or readers? And if Mole can make good money trading what is his real motive here?

    “Anyone who doesn’t feel they got value from my posts when Mole takes over, you can have a free 1 hour skype call with me (which I charge $200 for) to talk about any aspect of system building or trading you want to discuss. I can’t be fairer than that.”
    Yes you can…. Why would someone want to pay $200 more if they already felt dissatisfied with what they got? Double down I guess.

    A better option would probably be Peter Brandt’s Factor Service. For about the same price of Lord Mole’s cheapest offerings you get an in depth weekly update on all his existing positions and pending trade ideas and updates inter-week as new trades develop. This is from a man who has genuine humility and an audited track record with over 40% annual returns spanning 20 plus years who is not afraid to reveal his real name.

    Compare with the Mole’s narcissistic self aggrandizing style and anyone can see where real value lies.

    The Zero is not an edge, just another indicator that works on occasions. If it was good Mole would call real time trades for a while and prove it. Watch the subs would flood in then. Posting after the fact shows nothing.

    Scalpius was badly handled and Mole is hyper sensitive about it because he knows he messed it up. You said so your self Scott.

    So what to pay for here that looks attractive?

    Less time on the meme generator might make you look smarter and more open to subs and readers Mole, who ever you are.
    ???? ????? – ??? ????? since ES is going all Russian.

  • Sir Mole III

    Fuzz has been blocked/banned for posting insults. Sorry, pal – I actually read Cyrillic and understand enough Russian to pick up a few words. Anyway, if you don’t like what we do here then just stay away. Simple.

  • ZigZag

    Well, I’m already a subscriber. Perhaps it would be better to have a comments section only for members away from the interwebs’ never-ending supply of trolls.

  • Sir Mole III

    So that nobody can claim I’m above responding to negative criticism. From the top but sans the insults:

    Abrupt change – there has been none. I have run a hybrid site for over six years now and have been posting a veritable mountain of free information during that time (approaching 4000 posts over eight years).

    Making money trading vs. blogging: Blogging is a side income for me plus it keeps me honest in that I put my thoughts and analysis out there for everyone to see. The open dialog in the comment section also makes it more fun as life as a trader can be rather lonely and myopic at times. I could post it all for free but that would be foolish. If you don’t see the value then that’s fine. But many people do and there’s nothing wrong with levering one’s skill for profits. Nick Radge does it, Peter Brandt does it (as you pointed out), and many others worse and better than I apparently have no compunction to share their experiences, their setups, their market analysis, etc. in exchange for money. You are either very inexperienced, naive, or are simply gaslighting in order to bolster an emotionally loaded position.

    Peter Brandt’s service – sure, if you prefer an update once a week vs. me at least once per day then go right ahead and sign up.

    My “narcissistic self aggrandizing style” – clearly the fact that ‘Mole’ is an online persona representing a caricature on a site called Evil Speculator has somehow evaded you during all those years. Did you have a beef with GS Elevator as well?

    The Zero – most definitely has an edge and just because I do not point out every single divergence live has more to do with my time zone (Europe) than with laziness. I have posted countless videos where you can actually see me label a divergence in real time. I am also regularly posting Zero charts – almost every day I believe – on my twitter feed. Once again for free. So you get to trade those even though you’re not a sub. How selfish of me expecting you to sign up to follow it 7.5×5.

    Scalpius badly handled – the system has an edge and continues to have an edge, and has recently been revised extensively to avoid drawdowns > 7%. If you go on C2 you will literally find thousands of systems which work great for six months and then completely suck. It’s not unique to Scott and I, or our systems. There are never any guarantees when it comes to system or discretionary trading, read the fine print they made you sign on your trading account and the one I post with every system subscription product. HOWEVER, both Scott and I have worked our asses off to minimize draw downs and to produce systems more conducive for retail traders. I am confident that future results (all of which are posted openly and live by the way) will reflect our efforts.

    What to pay here for that looks attractive – You are seriously asking that? All it takes to answer this question is to peruse the last month of analysis, setups, and educational content. If you then still believe that there is no value worth the ridiculously low sub fees I charge then you are strongly encouraged to find better value elsewhere.

  • Scott Phillips

    I like Peter Brandt, I’ve read literally everything he’s ever posted, and several interviews with him. He seems like a nice guy and the real deal, I’m sure his factor service is worth it.

    What convinced me he is legit is that his win rate is less than 50/50 and he doesn’t think technical analysis has predictive ability.

    This is the type of stuff he posts

    I’m almost certain that every person here can recognise an ascending triangle on a chart, or draw a trendline or resistance line. Being honest, if you can’t do this kindergarten stuff this isn’t the place for you.

    Did anyone reading Peter’s post learn something? Did anyone reading MY posts learn something? Sure you can go do Van Tharp’s system building courses, he charges 50K for it. Sure you can do Al Brooks tape reading course (which is actually quite good, but not as good as mine), he charges $400 for it.

    I have some knowledge which is not common at all, I’m offering to share it. I’ve shared a lot of it for free over the years on this blog, to everyone’s benefit. I post my trading statements from time to time, and generally don’t play the guru game.

    It’s literally astonishing that I take time out of my holiday to write several thousand words on the hardest parts of trading (building systems), and I get criticised for asking for some of you to support the blog where you spend time each day.

    Not for me, mind you, for Mole. Go fuck yourself whoever the fuck you are anonymous troll

  • Sir Mole III

    And much valued one. No, I don’t believe in closed comment systems. Too much of a danger for them to turn into an echo chamber. In comparison with other sites the quality to troll ratio is huge. I’m talking one troll per quarter perhaps. No complaints and both Scott and I are big boys – we can handle it.

    FYI – I would have let his comment stand but will not condone outright insults. Funny is okay, not PC is okay, cynical is okay, slightly crass is okay. Insults are not. His loss as he wasted his time writing it.

  • Scott Phillips

    I bought a small Hokusai original print in an antique shop here in Hokkaido last week. I really can recommend Japan as a country, everything about it is very civilised.

  • ZigZag

    Hmm, quality to troll ratio. Perhaps that is a marketable indicator.

  • Scott Phillips


  • Sir Mole III

    that must have cost you a fortune.

  • Sir Mole III

    I’ve been long Evil Speculator for eight years 🙂

  • Scott Phillips

    Surprisingly, no!

  • Ladywandering

    What happened to the new post??

  • Mulv

    I spent some time with excel on the fakeout setup homework… the details:

    1. new 10 day true high
    2. close below previous 10 day true high
    3. sell the close

    I then compared the next 1-5 days returns with the previous 1-5 day returns. Sadly, only the third day’s correlation is somewhat negative (what one would hope for from a sell setup in a market making new highs), but day 2 and day 4 are mildly positive, so I’m guessing there are a couple outliers that drive that result — not an edge.

    I also tested this setup:
    1. new 10 day true high
    2. close below previous 10 day true high
    3. sell a break below the low of fakeout bar (on the next day)

    I compared the next 1-5 days returns versus the previous 1-5 day returns lagged by an additional session (I did this to eliminate the close – entry return in the comparison, plus the correlation came back at +96% for the previous 1 day return, so I knew something wasn’t right). Also, nothing seems trade-able for this setup. my data set is continuous E-mini over the last five years (Bloomberg conventional roll-over and back-adjusted by roll percentage). I’m going to work on the buy fakeouts over the next day or so and then check it all again as my eyes are glazing over with making sure the dates are lined up properly.

    Here’s the spreadsheet:

    Please let me know if anything obviously wrong.

    In a shock to no one, no consistent edge selling the E-mini in the last five years.

  • ZigZag

    It’s “like it never even happened”

  • Scott Phillips

    Truly excellent stuff 🙂 I’m in the airport in Tokyo, but I’d like to use this in the post. I suspect it will be an edge if we make a condition (the usual one) that the low of the setup bar is broken. I’m looking forward to doing the work and won’t mind at all if it’s a dry hole 🙂

  • Scott Phillips

    This is A LOT OF WORK!

    If you get a chance shoot me an email on scott@scalpius dot net and I’ll go through some of the finer points with you. With your permission I’ll use this in the post.

  • Sir Mole III

    It will be posted tomorrow as most of the crew hasn’t seen the current one yet.

  • Sir Mole III

    “I’m going to work on the buy fakeouts over the next day or so and then check it all again as my eyes are glazing over with making sure the dates are lined up properly.”

    Yeah, I’m actually thinking of putting together a tool in NInjaTrader that outputs CSV files automatically. Some of that code is already written as I experimented with SVMs in the past two years. Which meant I needed a parser that exports those vectors.

    Per your work – mean reversion (as it is understood in trading) is actually not frequently observed on daily timeframes. And clearly the last eight years had a distinct bullish bias. You may have better luck testing a long setup with inverse rules. But of course we should expect that one to fall apart if the long term trend eventually changes.

    One of the aspects of finding an edge that has not been covered in the previous articles is the cyclical nature of markets. So let’s say you test on a time series and you find a positive or negative correlation you believe can lead to developing an edge. How do you know that you are not just testing a time period that happens to be positively biased? You could test other ranges, perhaps during distinctly different markets. If that holds up then you may indeed be up to something.

    But what if you do NOT see a correlation? Is it because there is no edge to be had? The scatter plot Scott showed you guys will show you the frequency but not the standard deviation inside the 2-d plot. What we may also need is the concept of time, which would look something like the pic below. Because it is quite possible that correlations come in cycles and then either fall apart or perhaps even invert.

    IF that is the case then it will be necessary to determine the cause of this. I concede that this may prove to be problematic. Scott and I have often found ourselves in that very situation and the best we could up with was to make that determination implicitly by measuring results, which obviously should not be random. By observing the P&L curve we are able to establish a threshold at which we decide to rest the system.

    Now, the astute reader may have noticed that we are actually measuring the P&L curve which is the result of a full fledged trading system including entry rules, campaign management, exit rules, etc. Which is heavily filtered from the correlation we are seeking to exploit. So a better approach may be to actually measure a Markov chain of the correlation instead of our P&L. Because if the correlation remains to be positive but our system is not then it may be necessary to adjust the parameters of our system instead of turning it off.

    As of today neither Scott or I know of a single trader or trading firm employing this approach. So you heard it here first 😉

  • Sir Mole III

    Apologies for featuring my own comment but I wanted to make sure everyone sees it.

  • Sir Mole III

    Please read my comment above. I believe scatter plots as employed may be deceiving as they are based on a global data space. Correlations however may come and go – as a matter of fact I pose that MOST correlations (i.e. inefficiencies) either have a finite lifespan or come in cycles.

  • Sir Mole III

    A good way of explaining what I am talking about is via a handful of sand and a piece of paper. Start sprinkling the sand in a distinct diagonal pattern for about 10 seconds but then suddenly start sprinkling in the exact inverse pattern. What you’re getting after about 20 seconds is a cross like pattern which suggests a lack of correlation between the parameters you are attempting to measure.

    With a 3d scatter plot however you would clearly see a shift from one phase to the next.

  • i Bergamot

    ????? ???????

  • Mulv

    Instead of Markov chain analysis, could similar be accomplished by plotting correlation through time and adjusting parameters by the slope of the line, assuming cyclical behavior? And by parameters, you primarily mean bet size?

  • Sir Mole III

    The parameters can be anything but yes, in most cases we would be measuring a vector composed of two or more closing prices. And you are right – it could be a Markov chain or it could just be a series of vectors. Which is actually needed for a scatter plot, my mistake.

  • Sir Mole III

    How’s my crew? Apparently my absence has been barely noticed as Scott is doing an awesome job. Maybe I’ll just stay on vacation 😉

  • ridingwaves

    fed musings week…rates going up, FIRE sector might be safest place to hang…

  • ZigZag

    I spent a bunch of time manually gathering samples for my /tf project, and literally the moment I finished I got a message from TD-tos that ICE was cutting off free /TF data beginning Apr 1. Still have a lot of work to do, and its a good setup, butI’m not sure I want to trade in an instrument where the last of the unprofitable retail traders are going to throw in the towel. Volumes already suck, but I hear that the /tf is going back with the other indexes in the 3rd quarter.

  • sunseeker

    well said 🙂

  • BobbyLow

    It’s Quiet in here today. The Markets appear to be quiet too.

    I’m still holding my short Silver position which has been an excellent trade so far. I’ll be out of it by the COB tomorrow if not stopped out before. I don’t want to be in anything Metals related on Fed Day Weds. I also re-entered a long position in NG via UGAZ ths morning. Fundamentals might suck on Nutty Gas but it’s been on a pretty good run on the long side over the last 9 trading days. If this move runs out of “gas”, I’ll get stopped and so be it. I’m also not going to touch the miners one way or another until after the Fed does its thing.

  • Yoda

    before you guys all die of boredom watching the tape whilst waiting for the FED, I have a triple bottom action set-up for you:

  • BobbyLow

    Now that’s a 3X we all could like. 🙂

  • Yoda

    intraday play only, otherwise they rip you off 😉

  • Ronebadger

    shiny AND golden… Long GDX…pun not intended

  • Tomcat

    Would you recommend going LOOOOOOONG or Short here?

  • Yoda

    Definitively “balls to the wall” long after dropping off shorts.

  • Ronebadger
  • Yoda


  • Mulv

    I computed a rolling correlation over a 10 trade “period” for one of the strategies (10 day fakeout low, buy on close). 10 trades may not be sufficient to make trade decisions, but I figured that it would exhibit cyclicality if there were any… and there does appear to be a cycle for 1-5 day returns after the setup. Unfortunately, the outputs look like very sharp turning sine waves oscillating around zero correlation. Scott may have some kung fu to make it work, but it looks like a dry hole to me.

  • Sir Mole III

    I’m afraid we’ll need quite a bit more data there Mulv. Another reason why I’m thinking of building a parser that exports for Excel.

  • Sir Mole III

    Hello? Anyone remember me?

  • Sir Mole III

    I was posting actually some interesting input below (also featured) – no comments?

  • ZigZag

    We’re trying to let you have a vacation. Bored yet?

  • Mulv

    You certainly know better than I do, Mole. Using 5 years of data, the fakeout setups for E-Mini on the daily presented a maximum of 159 trades, so the sample size is limited to start with. Sufficient data could be another challenge to finding and characterizing correlations.

  • Sir Mole III

    Oh – so you focused on the trades, not the actual entry condition? Or are you saying that the fakeout entry condition only occurred 159 times?

    Also, any particular reason you are using daily data? Are you able to trade intraday as well?

  • BobbyLow

    Unfortunately, my basic math skills are OK but not good enough to contribute to more sophisticated levels of mathematical conversation and that’s on me. 🙂

  • Sir Mole III

    Out of my mind! 😉

    J/K – I’m kicking back a little and am catching up on my reading. Right now I’m deep in Ernest Chan’s Algorithmic Trading book. Quite math heavy but I’m flying along by the seat of my pants.

  • Sir Mole III

    My bad – I tried to explain it with the sprinkling sand example. Does that make sense to you?

  • BobbyLow

    No problem Mole and yes the “sprinking sand example” is a good visual and makes sense.

  • Mulv

    I put four iterations together…

    Fakeout sell – new 10 day high, but failed to close above prev 10 day high, sell close (159 entries)
    Fakeout sell – same setup as above, but delay entry to next day if fakeout bar’s low is broken (90 entries)
    Fakeout buy – new 10 day low, but failed to close below prev 10 day low, buy close (91 entries)
    Fakeout buy – same setup as above, but delay entry to next day if fakeout bar’s high is broken (54 entries)

    I ran the correlations (and scatterplots) for future returns after entry (1-5 days) against past returns (1-5 days / 2-6 days for the intraday entry). I also ran rolling 10 trade correlations for the first Fakeout buy setup in the above manner to see how stable the relationships were and I found that they oscillated around zero correlation pretty sharply.

    I used the daily time frame because I think that was Scott’s suggestion.

  • Yoda

    Do you ever stop working?

  • Darkthirty

    It certainly takes yuuge cajones to post on the markets in these times. Been here since about 2008. Confidence comes across as arrogance to people that don’t understand the investment. I say fuck em………..

  • captainboom

    Getting data out of a platform so that I can crunch my own numbers would be wonderful. Especially if it works on different time frames.

  • captainboom

    Nothing like a little light reading to help a guy relax.

  • Sir Mole III

    In my comment earlier this morning I mentioned that mean reversion is rare in artificial price series (i.e. market symbols as we know them). And they are even more rare on daily series where you’ll find the half-life time may be measured in weeks or months. What you are measuring right now has a low probability of succeeding until you measure the half-life time of your data series. There’s a matlab function for that which Ernest Chan posted but just visual examination can often give you a pretty good approximation.

  • Sir Mole III

    Anyway, this is why I suggested shorter time periods which exhibit a) more volatility and b) are more prone to mean reversion.

  • Sir Mole III

    What can I say – I love this shit. Plus after about three days of vacation I’m starting to lose my mind.

  • Sir Mole III

    Not really…

  • Sir Mole III

    Yeah, for some reason ICE has increasingly been fucking with retail over the past few years. On Kinetick they raised their fees every single year and I guess TD was sick of fronting it after the last increase. That’s fine – if they want to run an index dominated by HFT bots then that’s their prerogative. But if that game ever dies off or shifts to different markets then they’ll be dead in the water in a jiffy.

  • Mulv

    Half life time of daily series … I have more homework. Is it fair to say that a setup in a single day is unlikely to trigger a mean reversion that is likely measured in weeks? thank you for the guidance, Mole.

  • Mulv

    I think you’re right. It seemed that if there was a trade to be had with these daily fakeout setups in the E-Mini it was continuation, not reversal, in the fakeout direction after a couple of noisy sessions.

  • Scott Phillips

    I’ll be covering it in the post, but in general mean reversion works much better in direction of the dominant trend. Obviously shorting emini isn’t going to be much of an edge these days, but I just landed in Australia and I’m interested to play with the data

    I’m very encouraged that someone took the time to do the work on a likely dry hole.

    That is the kind of thing that winning traders do that losing traders don’t

  • Scott Phillips

    In terms of mean reversion the evidence bears this out.

  • Scott Phillips

    There is a problem with just measuring results.

    You get an unlucky one tick stopout which reverses and shoots higher without you on it.

    The next trade reverses after making 1.6R and stops you out for breakeven.

    The trade after that you get a fed announcement which enters, you, grabs 2R at a target, and whipsaws you out. Nothing to do with the trend/mean reversion/whatever you sought to capture, just dumb luck.

    Even if you are taking a lot of trades, 30 trades a month or so, those 3 trades are enough to make a great system look shitty, or vice versa. If you are taking a few trades a week, even more so.

    Of course edges are not consistent between markets, nor timeframes. Nobody is suggesting they are.


    The confidence intervals on a backtest mean literally nothing , because they are a fragile interaction of 10 different things.

    Scatter plots are superior (when you have statistical significance) because you can accurately measure the confidence interval on your single hypothesis.

    Of course statistical significance is not a small number.

  • ZigZag

    Hey Scott,

    So here is my scatter chart from the /TF exercise below. I went through my 2016 charts and pulled out all the examples which met my specific target of a virtual exact double bottom (within 5 or so ticks on /tf), at support, and with a tick divergence from specific points. That really limited the samples as there are obviously tons of double bottom setups of slightly higher/lower prices. I was hoping to limit my selective bias in not “seeing” failed setups this way.

    Since the price was virtually the same I then used the change in the price after my tick signal bar (after it’s clear the price is going back up), compared with the change 15 minutes later (I also have 30 min, 60 min, 120 minutes, and closing price changes in case I want to use that data). Is it correct that the question I’m looking to answer in this exercise is – Does the setup have a positive outcome over the signal bar?

    And if so, what next? ( I can send you the excel file if that would help)

  • ZigZag

    Well people keep saying that it’s a done deal that ICE is selling it back to (the CME? or whoever has the others) in June or July, so maybe it will come back to life then.

  • ridingwaves

    Ackman’s Pershing Square sells Valeant stake, takes $3 billion loss
    Reuters March 13, 2017

    ridingwaves 7 days ago
    for Kudra…Ackman lost billions because he stayed in too long and held firm on his beliefs…it can happen to the best minds on the street….

    3 billion, check out some of the state, teacher pension amounts….ouch…

  • CandleStickEmUpper

    Awesome post, thanks Scott

  • Sharon

    Getting all the new information copied off into Word files. And, then printed out to be studied. Looking at paper trading during my full time job and then starting with real money in July when my job will be done for the rest of the year. I am so grateful for both Sir Mole III and Mr Scott and everyone else who participates here and shares their own work. Ya for 2017.