The Distinction Between Human And Algo-Trading

Tyler Durden's picture

Submitted by The World Complex

One more time--the distinction between human- and algo-trading

The markets do not act like they once did. The trading in certain stocks is operating on time-scales so small that they cannot be in response to human thought. Not only are certain individuals able to access key information before others and so respond to news releases faster than the speed of light, but certain entities have free range to post and cancel orders on a microsecond basis, and queue-jump by shaving off (or adding on) tiny fractions of a penny from their orders.

Stocks traded by humans tend to make significant moves on a timescale of minutes to days. Even when there is a news event that radically changes the apparent value of a company, if there are only humans in the market, the move takes time to occur. Below are a couple of charts for Detour Gold (I currently have no position in this stock)

Normally, when looked at on a ms timescale, the graph is not really distinguishable from a straight line.

The little squares occur because all the price-changes I saw in the course of the day were a penny. On this scale it scarcely matters which axis is the current price and which is the lagged-price.

Once the algos get involved, the millisecond phase space plots get a lot more interesting. Some of them are works of art! Below, some plots for Century Casinos (I have no position in this one, either). Data here.



Algos playing tug-o-war.

Nice to look at, but maybe not so nice to trade against.

Remember the adage about playing poker: If you don't know who the sucker is . . .

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.
A Lunatic's picture

A few of those charts look a lot like my 'running around like a chicken with his head chopped off' investment history.......

michael_engineer's picture

The below creates closed loop shapes that would put those algorithms to shame.

With it being new technology and new on the app market in the past 2 weeks could even give it a better chance at significant ROI than many current investment plays.

It is fun looking and it may be possible to leverage that unique logic engine into several cross promoting technologies.

With the addition of a "sharing" user base it could have interesting potential.

markmotive's picture

Who cares if they are human. Human investment managers don't know what the f--- they are doing anyway. Most of the time you're better off owning a cheap index ETF anyway.

The Myth of the Successful Money Manager: Infographic

Wahooo's picture

Can someone explain to me what these charts are showing?

ultimate warrior's picture

Oh squiggly line in my eye fluid.

I see you lurking there on the periphery of my vision.

But when I try to look at you, you scurry away.

Are you shy squiggly line?

Why only when I ignore you, do you return to the center of my eye?

Oh squiggly line, it's alright, you are forgiven.


Running On Bingo Fuel's picture

Just trade the index for the last half hour of the day. Buy. Do this till January, bet big, retire.


GrinandBearit's picture

Looks like a chart my 3 year old nephew might draw.

Dr. Engali's picture

Man I new that hit of acid I took at a Dead concert would come back and haunt me some day.

Harlequin001's picture

If you bought it through silk road it might be sooner than you think...

Goldilocks's picture

'Game theory'
1955 Stanford Research Institute, Bank of America, and General Electric --- ERMA and MICR (search  ‘history of computers’)


Silent Weapons For Quiet Wars Document - Full Read (1:03:33)

Stuck on Zero's picture

Maybe someone can answer this question for me.  If I buy and sell stocks over a period of one year my tax form is a big mess.  50 trades and it's difficult to fill out the form.  If I were to high-frequency trade 50,000,000 times in a year what would my tax forms look like?

Second question: Suppose I machine traded 100,000,000,000,000 shares in one year.  Could I submit all the records to the IRS and permanently wipe out their accounting systems?


0b1knob's picture

"Taxes and the law are for little people."  The TBTF banks.

Urban Redneck's picture

I can't tell you how it works TODAY, but there was a year when I traded for myself (and not through a corp acct) and back then- the process was to file a declaration with the IRS, BEFORE the tax year BEGINS in which you want to trade professionally, and you can elect to calculate taxes due on a closing balance method, so even if I went full retard and executed 1000 trades in a day my schedule D just had 1 line for the annual net change with a supporting supporting statement attached that had 200 and some trading days worth of opening balance, closing balance, net gain/loss. The actual underlying math gets a complicated (because you have to exclude interest and dividends and report them elsewhere) and there can be additional adjustments depending on how you actually filled out the IRS election form (which primarily deals with a cash basis vs accrual accounting election, and that impacts all your other tax forms).

In short - your schedule d will look nice, but your taxes will still be a fucking mess (and you can no longer feign ignorance when dealing with the IRS). Also, I am not sure it is even possible to file the more sophisticated forms electronically (which also prevents system overload), so when the IRS gets the Fedex BOX - if there isn't some multi-million dollar check attached, they might just audit your ass rather than doing the manual data entry.

And I'm ABSOLUTELY SURE the IRS processes haven't gotten any better or easier in the last decade.

When it comes to the IRS- don't ask, don't tell, don't make things more complicated than they need to be.

Trampy's picture

I can't tell you how it works TODAY, but there was a year when I traded for myself (and not through a corp acct) and back then-

If all you want is a simple Schedule D, trade futures instead of equities.  Cap gains are spilt 40/60 short/long and yr broker does all the math giving you those two numbers for the entire year, options and futures both.

Professional traders NEED TO BE PROFITABLE or they're not a going concern.  Filing that way to simplify your Sched D is silly.  But if you have lots of trading profits and you want to take deductions for that expensive trading software and a roomful of computers and expensive reference books ... and Andrew Maguire subscriptions .... to write off from your trading profits, go right ahead, because that's the only time it would make any sense.

Urban Redneck's picture

The purpose is not to simplify the schedule D, it is to allow the individual trader to use Mark-to-Market accounting for their portfolio.

But I do agree that if you are NOT a professional with sizable income and expenses (or window dressing) then it's not a wise method to pursue.

BTW, if "your broker" is doing the math, you're leaving money on the table.

GoinFawr's picture


1 Ans. To the IRS they would look like they need to be reassessed @ least three times during the course of the next year (or more depending on what the |\|5@ thinks of your political views)


2 Ans. No. Even WW trois can't wipeout cockroaches completely. The resulting reassesments would be the same as in question 1, only 200,000 times more expensive for you.

algol_dog's picture

There's tax software available now for daytraders and the like, that automates the process. It is fairly simple to use as long as your broker keeps accurate records of your trades (not guaranteed). Also, you can send a CD for your records to the IRS. I have for years now.

Tegrat's picture

I am not an accountant, maybe others can provide more insight, but this is how I do it. Caveat Emptor.


I get the broker tax statement every year and put the entry/exits from the html or document provided by the broker into a spreadsheet. Easy to do if you know basic parsing/programming and/or copy paste. Figure out the grand total of cost and sales and enter one giant number for cost and sales in the schecule D (all my trades are by definition are short term). I then send the IRS the hard copies of the broker statements to prove my numbers were close, let them bastards deal with wash sales and their own BS rules. If they do all the work of calculation to the letter of the law, you may be entitled to more money  or the amount you owe is so small it's not worth the man hour(s) of effort to get to the exact numbers.  I am nice and send a CD-R with the XLS version of the spreadsheet + hardcopies and files on it. The one year i just entered the totals they correspondence audited me and i had to send them and updated Schedule D. 





monkeyshine's picture

Professional traders file unnder a different schedule with the IRS.  You can make this election too but its not as easy you really have to be a professional trader making 1000's of trades (or more since some cases have been decided against small family firms) and make trades pretty much every day of the year. At the end of the year you mark everything to market (as if everything you hold was sold) and write the gains and losses all as ordinary income.  You can also write off your computers, your "office", your DSL line and any research costs.  There are no long term gains allowed (you pay ordinary income on all of it) but you can avoid the wash rule and you can write off all losses as ordinary losses in one year (not the $3000 a year limit "capital losses" we regular folk have to abide).  You would need a different corporation to put your "investments" for long term gains, assuming you were seeking them. 

Iam Yue2's picture

"The markets do not act like they once did. The trading in certain stocks is operating on time-scales so small that they cannot be in response to human thought."

No just in response to NSA intercepts.

Intellikon's picture


The algorithmic arms race is heating up.  The days of the mouse-click intraday trader must be numbered... soon everyone will need a computer to make any buy/sell decision...  and at that point of course markets will collapse into oblivion.

luckylogger's picture

The alternative of course is patience and learning how to play with them--- sure they cause wild spikes that dont make anysence but have you limit orders in to exicute at the extreams + 1 or 2 ATR's and once a month or so a guy gets into a really good trade. We gotta learn how to deal with them and take advantage of thier shenagens. Just my 2c.

Intellikon's picture

If you're a click-trader the algorithms have you beat already... algorithms are sophisticated... they have all the trading school knowledge and indicators and behavior analytics built into their program, and they know what you are going to do next.. they know before you know..  these things use the same methods that Google uses to choose an ad to display for you, or Amazon uses to show you other books you might like. 

> We gotta learn how to deal with them and take advantage of thier shenagens.

This is exactly how algorithms view us.

Godisanhftbot's picture

 suggest not doing what you had planned, do something else.



Godisanhftbot's picture

 suggest not doing what you had planned, do something else.



Trampy's picture

The alternative of course is patience and learning how to play with them

Exactly.  Instead of bellyaching that it's "not fair" learn to deal with it.  Jesse Livermore described how to make a lot of money trading manipuated markets.  I don't see how you can do anything but blow up an account by placing "blind" entries at far extremes and waiting to  get hit.  ... because not all the algos are trading the same direction.  Do you go long just because the market spiked?  That makes no sense unless you have so much money that you can Martingale it, but if you do, then why wait?   There can be some very hairy whipsaws on a major print like NFP or CB rate announcements.

I look for setups like the stair-step rally shown in the chart and join in.  One of the best technical indicators is clock time, with many good moves lasting exactly 30 minutes, which is plenty of time for a human being to catch a ride for at least 20 ticks after a range breakout pushing a new 5-day high.  Moves like that can end up making 100 ticks in a night, where the Asian session is the best for a trend trader like me.  It helps to program sound alerts that a particular instrument is moving so my puter will tell me something is going on with crude by playing a sound file so there's no need to be watching the screens all night and it'll tell me something's happening even if that chart is not open, or i'm watching TV or posting here.  Faves for trending moves these days are 6A and CL right at the China open, or 2 hours after Globex open.  Another thing besides time is that the algos tend to move prices in 5-tick "blobs" and often "rest" at round #s, so with TF for example, it'll often hesitate at 1112.5 with some jittering before moving on to 1113.0 where it'll "hover" so tightly that it often just tests 1112.8 or 1112.9 before moving up to ~1113.5, 1114.0, et seq., which allows my usual OCO bracket of 3-tick stop and 18-tick target, so it'd be enter with Buy Limit 1113.0 and then OCO Sell Stop 1112.7, Sell Limit 1114.8.  It's a rush to enter TF with just a 3-tick stop and stay in by just a single tick, but the algos are protecting me by dutifully, and RELIABLY, scarfing up contracts while my stop is just 1 tick away.  The algos are helping my trade and it's a blast to have stop so tight and have it hold.

CL does the same stair-step moves with rest-stops at each 10-cent level, and major resting points at the .00 and .50 levels, etc.

For a technical breakout right now I'm drawing charts on 6E Euro futures where it's itching to kiss 1.40.  It'd be a hell of a short squeeze with so many positioned for a top at 1.39.  I'd enter that long with a stop order because once it breaks through it could jump all those 100 ticks in an hour.  And there's also selling OTM calls on it if it gets to 1.40 where it seems a safe bet it won't go much farther, at least for soon.  Being short calls doesn't mean you need to hold till expiry--some of those trades can be closed out in a day or two and make 20 ticks on the option, which is lot less volatile than underlying.  When YEN/USD crossed 100 this summer it had some great strong moves of up to 250 ticks in a day or two, but now it's pretty much all pooped out and stuck in jittering. 

Crude was a nice play last July 3 when it sliced through 100 and kept on going.  But now it's back where it belongs and I keep an eye on it in the evenings for buy programs that'll slowly ramp the price for an hour or two, moving stair-step for either 15- or 30-minutes at a time.  And WATCH out after a 60-minute move to new 5-day high because it's usually perfect time to reverse position and fade the new top.  Yeah, my charts have "Donchian" lines showing the 1, 2, and 4-week highs and lows.  Forget the NY session.  Markets are much more sedate and predictable in the Asian session.  And then when London opens, there's often a reversal of trend.  Aside from major prints like data from China or RBA where it whipsaws, markets are mostly one-directional for at least 30 minutes at a time.  And last year soybeans were fun to trade at the China open, but this year no so much.

My new thing is learning to trade the exchange-traded calendar spreads on crude, many of which inverted recently as crude seems to be slowly transitioning back to contango.

Intelligence is what's needed to adapt to changes in the environment.

The trick is being flexible and keeping an open mind.  I'll bet almost nobody here knows that the algos on a buy program in CL and 6A usually move the price for exactly 30 minutes, eh?  And there's a big difff between buy and sell programs, with buying almost always much smoother than selling, for CL at least, and only a bit more than not for 6A.


Intellikon's picture

Sure, human click-traders identify patterns that work well and milk them until they don't work.  The success of your approach requires that the brain be able to accurately identify patterns and have well-timed execution, consistently.  Computers are really really good at all that - better than humans in a lot of respects. 

This wasn't always the case, but algorithmic research has increased and computers adapt and learn.  As a result, market dymanics have changed:  Patterns appear and disappear faster than than they used to.  Trends which used to take weeks to develop take half the time.  Pull-backs are becoming rarer, false breakouts more frequent.   This is causing click-traders to become increasingly inefficient in generating profit (brokers are already seeing this).

Something has to give way in all this.  Either click traders will become extinct and computers will rule the market, or high frequency trading will be taxed and things will slowly get back to normal... although it is likely that markets will be pushed to the brink before any regulatory rules are implemented.


ebworthen's picture

Who says Hal 9000 can't fingerpaint!?!?

"Trying to grab that penny from in front of that steamroller ain't worth it kid!"

withglee's picture

The obvious solution: Add a random amount of time when assigning the timestamp to the transaction. Then process in FIFO manner according to timestamp. Set the time constant such that all traders are treated equally (i.e. mechanical delayed to manual). If you use a random delay with a mean of seconds, a few milliseconds brings no advantage.

Of course, this is going to leave a lot purchasers of these high tech gaming facilities with egg on their faces and their cries of "foul" will be heard far and wide.

Todd Marshall
Plantersville, TX


texas sandman's picture

Full seconds are sooooooooo 20th century.  

JamesBond's picture

Oh, that's just the Etch-a-sketch algo at work




venturen's picture

can you map the trades the day before knight trading obvelerated itself? Or how about the Goldman robot that went rough. 

ToNYC's picture

Rise of auto-mechanical $hearing, duh looking back, like Mr. Gatling's gun extracted Anger and Opposition from its counter-parties automatically.

olto's picture

just makes me want to stack

so simple if i had money or credit

Godisanhftbot's picture

 You know you're screwed when you can't tell if  the op is mocking nanex or serious.