System Trading
Now Reading
Learning to Read
52

Learning to Read

by ScottMarch 8, 2017

Trying to predict what the stock market will do is for dumbasses. A proper analysis of the big name stock tippers and analysts shows that they are all significantly worse than random.

This is a spreadsheet built by Tim Knight tracking Gartman’s predictions, running at a 22% correct rate for the year. What Tim fails to mention is that his own predictions are every bit as bad.

Predicting markets sucks, and it induces a whole bunch of mental baggage where we make a prediction and all of a sudden want it to be true. We cheer it on like a horse at the track.

There is a bug in our mental wiring where we only remember the correct calls and cannot remember the losers. A prediction monkey can be right one in 4 times and all we can remember is that one big win. This is a very tempting way to trade, but that way lies ruin. In the strongest possible terms I tell traders to walk away from every method that claims to be able to predict the future with unerring accuracy (wave theory extra fuck you).

But even so, there is still a lot of information contained in the way price unfolds. Our job as traders is to interpret the constant stream of information objectively in the now moment, trading only when the odds are in our favour.

We wait, for those times when either the bears or the bulls have fucked it up, and will have to pay the price for their folly. We then join the winning side and help kick the hopeless bears (or bulls, I’m an equal opportunity savage) into a puddle of lovely red strawberry jam.

A-Clockwork-Orange

People often ask me for an explanation of reading market price action. I’m going to give a basic overview today, and we will dive deep on some specialty topics over the coming weeks. Beginners make too much out of this skill. In the final analysis it is cool, but not as relevant as you think to trading success. It’s really just a cool trick, honestly, and some of the best traders I know cannot do it at all. If you were deciding to spend time learning this or the system building material I will cover, it’s a no brainer. Systems > tape reading, every time.

There is a couple of fundamental points I need to make.

Every single bar of price action gives us fresh information in context of what should be happening. It either confirms or casts doubt upon the most likely theme, or provides clarity towards a resolution. Nothing is irrelevant, nothing at all. If it *should* be going down and it didn’t, that is important. If it could not close below a previous low, thats important. If the range was decreasing as the trend went on, that’s important.

Big money controls the market. The big banks and hedge funds win every single time. If we want to be profitable we should be doing like they are doing, and fortunately for us, big money leaves big footprints. We walk along behind, seeing the footprints in the sand, working out what the big money is doing and we do the same.

To read the tape, we slow it down, and go through every bar one at a time, reading out the story that price is telling us. We want to get away from the idea that we know what the future brings, and treat it like an interesting novel, each page brings us more information and we can’t wait to see what happens. Any preconceptions we might have just ruin the story.

Let’s break it down. The very first thing you want to do when you look at a chart is say “Is it a trend or a trading range?” And the kicker is that if you don’t know, or there is any doubt at all, its probably a trading range.

both trend

Trends have:

  • Very few overlapping bars
  • Strong kickoffs
  • At the start they have small pullbacks or pullbacks contained within a single bar, indicating that the bus leaves empty
  • Bars where the price action opens at or near the lows, and closes at or near the highs, and vice versa
  • Sequences of several up bars or several down bars in a row
  • Pullbacks in trends are choppy and the ease of movement is with trend direction

Trading Ranges have:

  • Lots of overlapping bars
  • Price action that looks like random, where you get an up bar, then a down bar, then another up bar, etc
  • Lots of doji bars, small range bars that close somewhere in the middle

Now in a trend the optimal thing to do is to buy pullbacks and hold for an extended move. In a trading range the optimal thing to do is to buy at the bottom of the range and short at the top, and not hold for very long, just scalp a quick bit of cash and jump out.

These two approaches are opposite, so it’s critical we figure out which one is in play.

Ok, got that. Trend or trading range is critical. Good. If you only take one thing home today that’s the thing.

Now, moving on. One bar at a time. Get all your stupid indicators off the chart, they don’t mean a thing anyway. They don’t have predictive value, price doesn’t “find support at the 50 EMA” its just a line on a chart that’s a derivative of price.

Zoom that chart all the way in. You only want to be looking at a few bars at a time.

Each bar you want to be looking at

  • Did the bar break a previous bar’s high or low?
  • How close to the top and the bottom of the bar did the bar open and close. An open at the lows close at the highs indicates buyers controlled both the open and the close (ie sellers are fucked)
  • Did the bar close above the previous bar’s high or low?
  • Was the range increasing or decreasing?
  • Is there a microgap (open above the previous close)?
  • How much overlap is there with the previous bar (overlap is two sided price action)
  • What will bears be hoping? Is the bear case still viable?
  • What will bulls be hoping for? Is the bull case still viable?
  • What SHOULD have been happening this bar? Did that actually happen, or did something weird happen indicating that something else is in play?

See how much more detail I go into on every single bar than most of you? Every single bar you want to do this much analysis. I literally read it out in my head like a bedtime story.

Let’s observe the breakout in the ES futures from the trading range, bar by bar.

breakout

bears fucked

Look at the most recent bar above, closely. See the tiny gap, at the open above the previous close. That right there is a telling clue, that the buying pressure was so strong, that traders could not and would not wait to get long, they just bought the bid right at the open, they were so bullish that they didn’t care they were bidding the price up. Small thing, big clue!

At this point most of the bears would have been forced to cover. The few remaining shorts would be hoping against hope for a failed breakout, but that is an extremely low probability outcome. The bears have only a limited amount of ammunition before they become weak for a time. If they expend selling pressure here, they will be unable to mount an effective defence later on. So we should eagerly watch for any signs of selling pressure. A failed attempt to drive the market down here would be extremely bullish. And a successful attempt would be a low probability outcome, which usually results in a big move. We just don’t know yet, we only know that the odds strongly favour a bull move at this point, and only the dumbest of traders would remain short.

You don’t read the tape to try and predict the future. At any given point you kind of know the odds of it going up or down. The best betting odds you get are rarely better than 65:35. A significant amount of the time the odds of it going up or down are truly 50/50.

But here’s the catch. Low probability outcomes cause MASSIVE MOVES.

This is so important I’m going to repeat it, a different way. When something looks really really certain to happen, and it doesn’t, then almost certainly the market is sending a hidden signal it wants to do the opposite. When traders get positioned short on a very obvious short setup, and that short setup fails, they will be forced to cover, driving prices higher.

Let’s take a look at the worst time in the GFC to illustrate this. What I want you to see is the last red bar 5th from the right. That bar, the market tried to go up, failed, and closed at the lows of the month, indicating the sellers were in control. It also closed below the previous 10 year lows, with an increased range, indicating things are accelerating to the downside and there is no buying pressure in sight. At this point it’s reasonable to assume that the market is completely and utterly fucked.

gfc1

Now at the furthest bar on the right hand side of the chart we have a 4 bar counter trend rally. Each bar is decreasing in range, indicating the rally is gradually running out of steam. The last bar has about 60% overlap with the previous bar, indicating two sided price action, and the last bar closed below the previous high, indicating the market could not support fresh highs. Also it’s a shooting star, which is a nice way of saying the market tried and failed to go up in a single bar. All in all we have everything looking bearish as fuck, from every perspective you can think of.

Or is it?

Let’s put our thinking caps on and go back to the lows. The market SHOULD have gone down after the last red bar. Why didn’t it? Our spidey sense should be tingling there. It’s a clue, a big clue, and we have to remember that every single bar of price action gives us fresh information. Nothing is useless.

Even so, the market objectively would be a 70% chance of at least attempting to go down and retest the lows, the whole world would be getting short again, preparing for another leg down.

What happens when a high probability outcome does not happen? Low probability outcomes cause MASSIVE MOVES, memba?

spforcedtocover membz

spforcedtocover trading range kicking

As for the markets today. The big news is Crude broke out of it’s months long trading range. The easy meat is done, expect a bit of chop today and don’t rush in. But no longs in CL, the market has spoken, and you don’t wipe away one of the longest trading ranges in history with a single day move. As for the indexes, the market tried and failed to go up today, when it was reasonable to assume another assault on the old highs is a probability. The conclusion is that bears are a little stronger or bulls are a little weaker than they appear, but price action is still very two sided. This is a market balanced on a knife edge, and we don’t have betting odds. The highest probability is for a long setup appearing after a more complex and hard to figure out correction.

es

crude

We will resume the system building stuff tomorrow. Scott


About The Author
Scott
  • Yoda

    22% success rate? Gartman is certainly an edge for contrarian trades.

  • Scott Phillips

    No doubt about that!

  • Scott Phillips

    The problem is when we start thinking we can do better. Honestly, if I was in the prediction game, I’d be no better than 35%, and that’s the truth.

  • http://evilspeculator.com Sir Mole III

    So the lesson folks is – trade the opposite of whatever Scott thinks and you’d be 65% correct. SOLID EDGE.

  • Scott Phillips

    Very solid actually! I’m as susceptible to bias, groupthink and delusion as the worst traders here.

    Very rarely do I see discretionary traders who even keep records. Why do you think that is? Everybody likes to be right, nobody wants to be wrong.

    Anyone who claims not to be not susceptible produce their records, because I’m calling bullshit :-)

  • http://evilspeculator.com Sir Mole III

    Okay, in the context of scatter plots (previous post). Here’s something that will blow your mind:

    http://www.edvancer.in/wp-content/uploads/2015/10/9898052ab66950074582766bdfa511d9.png

    I’ve been bending my mind over this when I started to dig deeper into support vector machines. The underlying idea here is that classification or regression model may often look like noise and perhaps *you are just looking at it the wrong way*. An SVM allows you to project your data into higher dimensional feature space, which sometimes leads to a better separation (as in the simple 3d example above).

    So if you think about this further then the question that can be asked is this: Am I looking at the historical time series the right way? We usually will choose a 2-dimensional model: up and down and perhaps sideways – the x/y Cartesian world we think in as traders. But price or its various derivatives (think indicators or combinations thereof) may exhibit behavior that is completely invisible to us. Why? Because we are looking at it the wrong way and perhaps by projecting it into multi-dimensional. There may be systems out there which we’ll never even think about because we simply just don’t see the patterns that create them.

    I’ve been trying to crack this nut off and on. A lot of people are looking at big data and machine learning these days but what they are throwing into those algos is the same old 2d Cartesian price data. I am rather convinced that key to using machine learning tools is not to make them chew on noisy 2d price series. Instead use these tools to arrive at better representations of historical time series.

    This is one only of many esoteric topics I never talk about on the blog because quite frankly it would go completely over everyone’s head.

  • http://evilspeculator.com Sir Mole III

    Funny thing is that I started to post live updates of my campaigns here (on the right sidepanel – still there). It was a lot of work and doubled my time producing my posts each day. Unfortunately after a few months I found that nobody was really looking at them, which really threw me off as I spent several weeks implementing that functionality.

    So yes, if you write about this here and ask people they will all profess that it’s the only way. But from a blogging perspective (not trading to be clear) all that stuff gets ignored. Most people want to just see entries entries entries and more juicy setups. Educational stuff is appreciated but only in measure.

    I have found a pretty good middle way in the past few years which is to mix both updates with new campaigns in my posts. It lets people absorb good campaign management by osmosis. BUT they will not seek it out

  • Yoda

    I for one like your balanced approach

  • Scott Phillips

    Yeah but you ACTUALLY KEEP RECORDS. Just because you don’t make them public doesn’t mean you don’t. If you didn’t you would be literally fooling yourself.

  • Scott Phillips

    Trippy stuff!

  • Scott Phillips

    And my dissatisfaction with the blogging game is why I don’t do it :-)

    People want entries setups setups, predictions setups.

    But we both know that you spend 10% of your day on that, and the rest on system design, record keeping and monitoring, and campaign management.

    Like children who want to eat chocolate for every meal IMO

  • http://gerb-reloaded.blogspot.com/ Gold_Gerb

    whoa?! another post. just like the golden days.

    I scattered a plot of the S&P open-minus previous close, versus the day of the month.
    Thinking would be that during the beginning or ending of a month, the deltas would be bigger.
    You know, manipulation to create window dressing. 😉

    Nada, zilch burger. See? anyone can do it.

    https://uploads.disquscdn.com/images/82f340c43608bc4fb1daa509dca6b15c193d49831818fcea8a183b5872ca685c.png

    The graph is day of the month vertical center, with delta negative or positive. 5 years of data.
    -GG

  • Scott Phillips

    Most excellent stuff mate. The fact that it was a dry hole doesn’t diminish it’s value. You now KNOW it is a dry hole, so when some well known dumbass tells you how important it is you won’t be fooled.

  • http://gerb-reloaded.blogspot.com/ Gold_Gerb

    Thanks Scott/Mole for sharing with the leeches.

    IMHO, this takes this forum to a Whole New Level.

  • Scott Phillips

    I’ll be back on this stuff for tomorrows posts. Also, the research environment at quantopian has tools to do this stuff (and the programming required isn’t even difficult for a dumbass like me). Its worth doing the codeacademy python lessons, which would be all you need to get up and running.

  • Scott Phillips

    Gerb try this. Take that same data for day minus previous day. Measure the Y axis as the day before. And the X axis as the day after.

    Try 30th-31st on one axis, and 31st-1st on the Y axis.

    See if there is a rebalancing (mean reversion) effect at the end of the month

  • sutluc

    Thanks Scott, good stuff.

    I have system that signaled short crude seven damn weeks ago, (on a weekly time frame).

    I finally sold today on discretion.Made a decent profit, couldn’t stand to look at the range anymore and seeing the possibility of a pretty substantial retrace.

    I trade etf’s in a tax free savings account, futures not allowed in there. I actually made a few percent just on the contango of the futures while the range worked itself out.

  • Scott Phillips

    Good choice to bank today, you got the windfall, caught the easy part of the move. Congratulate yourself and move on!

  • Scott Phillips

    Can you post the spreadsheet in a google docs sheet, and I’ll fix it up for you :)

  • Scott Phillips

    Question. Are you using this for a learning model, ie for market regime classification, or to build a matrix of factors for system optimisation or something in between?

  • AcoBrasil

    Fantastic perspective. Thanks for putting in all the time on this.

  • Ronebadger

    Thanks Scott! Good stuff

  • Tomcat

    Loving these posts!!!

  • http://gerb-reloaded.blogspot.com/ Gold_Gerb

    speak of Gartman. He’s truly out of oil, cost 1.5%.
    He must be a shill for a hidden party.

    ..everybody back in the pool!
    😉

    http://lh4.ggpht.com/-TnjfsmiFrPY/UCPcrm–SgI/AAAAAAAAbno/6tBssi12PZs/china-crowded-pool-1%25255B7%25255D.jpg

  • Yoda

    LOL

  • Tomcat

    I am not sure, if anyone of you follow David Tepper, but I couldnt agree more with what he said about going against this market, paraphrasing here…that as long as the punch bowl is still in play, shorting this market is foolish…

  • http://gerb-reloaded.blogspot.com/ Gold_Gerb
  • Tomcat

    Nat gas draw of 68BCF, short term bullish

  • BobbyLow

    I’ve been long UGAZ at $15.27 and had my finger on the sell button (just in case) when the report came out. So far so good, but I’ll be out by the COB tomorrow if I don’t get stopped before.

  • Tomcat

    Congrats Bobby. I took profits too early in hindsight. IMO this “rally” will be short lived, mid-term I am still bearish and expressing my view with short late 2017 contract.

  • sutluc

    Yes, that’s what I’ve done. System makes position changes near the end of the last regular trading day of the week so I should still be in…

    But I suffer from “do-itis”, the urge to be doing something. Watching a position that was supposed to be trend following in a long term range wears on me and tests me mentally.

    Best move was to move on.

  • BobbyLow

    Thanks TC and Congrats on your profits too.

    It’s been a long time since I’ve traded NG but I do remember it trading kind of wierd. I’m currently trading an hourly chart and usiing BB’s and a multiple of ATR to get me in and get me out. Not having a lot of discretion on my trades has worked out good for me.

  • ridingwaves

    moved stop on PIRS to 2.80

  • ZigZag

    I’m absolutely sure no one is peeing in that pool.

  • Darkthirty

    Agree, gotta work with what you have, but one day whack the vix isn’t gonna work…………..

  • http://gerb-reloaded.blogspot.com/ Gold_Gerb

    XLE,
    this week would be the time to take a chance on high reward, low prob long.

    https://uploads.disquscdn.com/images/97ee5bdf575a392a231ec534374601f1aed760482a4e1b9a07a0c1c2e485fdae.png

    If I was a in-the-money bear, I would wait actually, below the (blue) line to take profits, or tighten stops. look at that momentum!
    -GG

  • Darkthirty
  • Mark Shinnick

    Be sure to account for any recency bias when it come to short term supports.

  • Scott Phillips

    Or any possible basis why drawing a series of lines on a chart might motivate the numberless buyers and sellers in the world to change their positions.

  • http://evilspeculator.com Sir Mole III
  • http://evilspeculator.com Sir Mole III

    So the Mole call is back on track!

  • Ronebadger

    end of day should tell the tale

  • http://evilspeculator.com Sir Mole III

    I would only short on a short term basis, meaning 60min charts and below. In the end it all depends on your trading horizon. But going short and expecting a week long sell off is a bit foolish given the vehemence of the preceding rally.

  • http://evilspeculator.com Sir Mole III

    “couldn’t stand to look at the range anymore and seeing the possibility of a pretty substantial retrace.”

    Not to curb your enthusiasm but that phrase suggests that you still have some work do to on your campaign management. Focusing on entries only is something I am trying to get you guys away from.

  • http://evilspeculator.com Sir Mole III

    It usually does.

  • sutluc

    Agree. Always need improvement.

  • CandleStickEmUpper

    Excellent post Scott. Very useful and informative for a scalp trader like me.

  • Tomcat

    Don’t you have a vacation to “focus” on???

  • StockTalker

    Straight up BS, look at the signal

  • BobbyLow

    Scott’s post yesterday afternoon got me thinking that I haven’t been taking advantage of all the data that my back testing had provided. I’ve done plenty of back testing for calculating APE’s etc., but never realized how much information that was there and went undiscovered.

    I’ve been able to cull certain data after reviewing my back test for Natural Gas like over the long haul, it was more probable than not to be more profitable closing out my positions by the COB on Fridays. So here it is Thursday and I’m sitting on over +2R on a Long Nutty Gas position and I’m itching to close it out but my stop is too far away. Out of curiosity, I went back and looked at the trades on my back test and found that over 60% of the time, these trades closed worse on Fridays than on Thursdays. Of course the one thing about these stats is that I can make turn them around and say that if the trade was struggling or just slightly profitable then there might be a 40% chance of it getting better on Friday.

    Then again, this could all be Bull Shit. But if I’m going to get into a trade I need to have a technical reason and the same thing goes for getting out of one. So I’m booking profits and am out MOC today. :)

  • Scott Phillips

    NEW POST

  • JH

    Amazingly good stuff here, succinct and well illustrated.
    I wish I’d read this current series by Scott ten years ago when I first got interested in FX, would have saved me years of unfocused dabbling in technical analysis, treating moving average crossovers and fibonacci levels as gospel…. Eagerly anticipating the upcoming posts.

    To any beginners out there who are serious about being a trader, this is a stellar post about the basics, and the points and framework towards reading price action as illustrated here ones to be internalized. (and constantly revisited) Just so any of you younguns chancing across this comment will take my words to heart, I am currently an FX options market maker in a ‘big bank’.