Is fully automated trading a fantasy? Your opinions here

[Spectre2007]

lol. 12.5 x 60 x 60 x 6 = 270,000 dollars

21,600 round turn trades ..lets say 20,000 trades, Whats the typical volume in the ES on a given day about 800,000. You can calculate based on open interest change how many new participants entered. Most don't hold for the day.

How many computer programs do you think are operating in es-mini currently. So your computer program would on average be entering 10,000 trade entering and 10,000 exiting.. And most are flipping 50-100 lots. The programs are only marginally better. The wins outnumber the losses only a small fraction.

Out of 10,000 trade you lost 4,000 but won on 6,000, your only up 2,000 ticks, so about 25,000 dollars. Transaction costs lets say, are 20 cents. 5000 dollars. about 20,000 in profit per day trading 1 contract if your program is just marginally good at deciphering price instability. if your costs are twice, your profit goes down to only 10,000.

 

[ybfjax]

I couldn't edit my previous post, so I'll copy/paste it here:

And that's 21,600 RT trades. And of course minus commissions. So at 50% you actually lose to the house (brokerage, exchange, etc.) If you are playing it this way, it would appear that unless you can do at least 75-85% or better with the scalping method, it's useless.

Commissions would be around $3-4 RT retail with that volume with the right broker? Or maybe $2 RT for a professional? $4 x 21,600 = $86,400; $3 x 21,600 = $64,800; $2 x 21,600 = $43,200.

*********************************
Author's note....I did the original calculations using the $2,700 per 10% profit. The correct 10% profit was $27,000

But actually, the numbers work out to be the same, as the amount of trades would be 21,600 I have already edited the above to reflect this extra zero.

********************************
Especially if you are trading multiple contracts, you had better not do less than those rates consistently.

After looking at the new numbers with the added zero kind of starting balance would this require? With $100k, you'd better win the first day you flip the switch.

Even with $1 mil starting balance and assume you could get $2 commissions with that volume, if you did 50% good on day 1, you're still 4.32% drawndown.

With the scalping method, the small guy is essentially screwed unless he has a near-perfect system running.

I guess if your transaction costs are 20-40cents per trade, it maybe doable. But between the exchange, platform, brokerage, datafeed, etc, I don't see the final costs per trade (all inclusive) being that cheap. Perhaps I'm wrong about this part of the calculation.

 

[Spectre2007]

http://www.cme.com/clearing/clr/fees/equtyclrng1867.html

membership has its benefits.

 

[Spectre2007]

if you need help designing a basic scalping system pm me or email me.

 

[ybfjax]

Quote from Spectre2007:

http://www.cme.com/clearing/clr/fees/equtyclrng1867.html

membership has its benefits.

It seems like there are several different types of memberships. Assuming an individual could get their hands on the one with the $50 cap for the e-minis, this only takes care of the exchange fee.

We still have a brokerage, datafeed, platform, etc. Perhaps the platform and datafeed would be included with the brokerage. But the brokerage would still charge $1.5-3 rt so that they can eat too?

Unless I'm still missing something here.

 

[TraDaToR]

Really fascinating posts , Spectre2007.

I have a few questions :

- I don't fully understand what FIFO Dominance is. Is it related to the size of orders placed at different levels on bid and offer?

- What kind of data feed you use for this kind of scalping systems? What are the data providers that permits to use the book as the base of your system?

THANKS.

 

[callmeput]

Is someone out there profiting by producing imbalances in bid/offer to fool the machines?

 

[Spectre2007]

Charley Carey, trader and Chairman of the CBOT: “Whether you are a one-lot scalper or a million-bushel trader or a thousand-contract trader in financials, without discipline you won’t be around. You have to be flexible enough to reinvent yourself when things stop working, and you have to have a good understanding of game theory—you have to understand whether you are trading by an algorithm, whether you are trading by a fifo system. You must understand the technology that is supporting the contract you are trading. That is a must. That’s all part of the game theory of understanding. You are competing against not just individuals, but black boxes, so you have to understand how the black box is built—how’s it thinking about the market. In other words what’s been put in there that creates the automatic reaction that you are either going to have to beat, get around or learn how to manage. Where are you going to fit in to be a successful trader? That’s all part of that game theory; if you don’t have that, it’s going to be tough to succeed.”

 

[Spectre2007]

E-mini futures trade on the CME’s Globex® electronic trading system. These futures trades are matched on a First-In-First-Out (FIFO) price algorithm. That means no one has an advantage over you in the execution of your order. All orders for these futures products trade in this one centralized market. You do not need to worry about where to route your order for the best fill.

E-mini futures’ market depths and volumes are among the highest of any traded instruments. They are clearly the product of choice for active daytraders. Average daily volumes for the S&P 500 E-mini have been in excess of 500,000 contracts with record volume of 1,130,000 contracts traded in a day. All this trading takes place at the single location of the CME Globex matching engine. You do not need to search for the greatest source of liquidity to execute your trade.

http://www.terranovaonline.com/TNO_Help/Emini_Futures.asp

this link is just posted because it offered details, and I wanted to give them credit for my cut and paste.

http://www.cme.com/files/PriceBanding02.pdf
http://www.cme.com/trading/get/adv/printerFriendly/16782.html


A large money pool can inundate a price ladder with limit orders at every tick 2 points around the current spread. By doing this he gains the FIFO advantage. Having his limit order filled first.

 

[Spectre2007]

Rank 2006 Rank 2005 Fund Name Assets
1 3 Goldman Sachs $21.0 billion
2 2 Bridgewater $20.9 billion
3 7 D.E. Shaw $19.9 billion
4 1 Farallon $16.4 billion
5 16 ESL $15.5 billion
6 15 Barclays Global $14.3 billion
7 9 Och-Ziff $14.3 billion
8 5 Man Investments $12.7 billion
9 11 Tudor Investment $12.7 billion
10 7 Caxton Associates $12.5 billion

http://en.wikipedia.org/wiki/Renaissance_Technologies

Renaissance:

Renaissance uses computer-based models to predict price changes in easily-traded financial instruments. These models are based on analyzing as much data as can be gathered, then looking for non-random movements to make predictions. Renaissance represents a validation of the quantitative trading model and trades with such high-frequency that it (the Nova fund, specifically) accounts for over 10% of all the trades occurring on NASDAQ some days.

800,000 volume per session/20,000 seconds = 40 per second.

Sure you can disrupt the price equilibrium, using a large order bluff. The money pool extraction is limited. Based on the liquidity per second. The computer programs have to split about 11 million dollars among themselves for a trading session using the basic scalping system. If 10 computer programs are dominating. Its a million a piece per trading session.

One million x 250 trading sessions = 250 million annually. 250 million assuming your 100% on your price analysis. Lets say half are scratches, that leaves 125 million. 75 million in wins and 50 million in losses = 25 million. Half goes to transactional costs. And your left with 10 million. 10 million isnt much unless you can extract more from the pool of scratches or minimize the pool of losses. So the money pool extraction probably isnt worth it for funds exceeding 1 billion. For a 1 billion dollar fund, it might add 1 percent in revenue which can be with minimal risk.