The Art of Scalping

[fullautotrading]

Quote from abattia:

Please can you elaborate on this?

Thanks

Thanks to you, abattia

Let's consider 1 scalp: BUY - SELL

This can be performed either as Buy -> Sell or as Sell -> Buy, depending on price action.

An automated scalping system is programmed to perform thousands of such scalps during the week by opening
multiple scalping positions. (It's fashionable now to refer to massive automated scalping as HFT.)

Now assume at a given time you enter a scalp by BUYing.

You can have:

1. Price going up until target is reached (profit)
2. Price going down (causing drawdown)

Point 1 poses no problems.

All the "Saint Grail" is "just" about finding ways to to cope with point 2 (here is where "hedging" come into play).

There are various possibilities:

0. Most obvious to newbies (and suggested by brokers): place stops.
This is, in most cases, a 100% guaranteed losing strategy, with no possibility whatsoever to survive for long time. Mostly due to: spread, commissions, limited capital of trader, spikes, market manipulation, etc.

1. Methods which (have no stops and) use options, aiming to create "corridors" for scalping.

2. Methods which (have no stops and) use the traded instrument itself for hedging, by employing efficient methods to arrange the scalps so that they also provides protection to the scalps already open (also waiting for possible reversal).

3. Other methods (... suggest!)

Point 1 poses some difficulty as to implementation due to the the presence of options, not easy to consider within simulative schemes. Pro: intuitively appealing. Cons: difficult to put in practice, option time decay highly discouraging, difficult to evaluate.

For point 2 the strategist is bounded by his own fantasy, and only intensive simulation and forward testing can guide in the search for the most effective methods. Pro: all is managed automatically within the same strategy and implementation, easy evaluation, smaller drawdown wrt to stops usage. Cons: requires reliable infrastructures, restriction to highly liquid instruments and not too volatile (eg: zb, es, ym, zn, ...)

In any case, to have chances of success with these approaches (and, probably, any trading approach), in my opinion, good capital is always necessary (1-2 M).


Tom


(PS. About all the hype on "speed" of the HFT my personal idea is that it's just marketing talk.
A bad strategy will remain such even if you could execute orders in nanoseconds. Similarly a robust strategy will remain such even if orders are delayed by 1 minute. Obvious illusory advantages are rapidly wiped out by market and regulations.)

 

[Cheese]

HFT is not something that ET members do. Whilst HFT is controversial, there is at least an equal argument that HFT adds valuable liquidity. I side with this view.

So far as markets in commodities and index futures are concerned, funds, investment bankers, institutions in general and commercial hedgers do not get in the way of the small independent traders. They help to provide the essential liquidity which makes markets freely liquid for very small players to use. It is pathetic how often posters at ET blame big players for market movements which frustrate them. These large players have not the slightest interest in small retail traders and what they are doing, still less does it matter to the bigger players whether or not very small traders shoot themselves in the foot every day.

 

[schizo]

Quote from fullautotrading:

3. Other methods (... suggest!)

You can certainly hedge by utilizing a similar instruments. For example, 3 lots of YM for every lot of ES, etc. It's quite easy to implement. The key is to knowing when the optimal time is to unload.

 

[schizo]

Quote from Cheese:

HFT is not something that ET members do. Whilst HFT is controversial, there is at least an equal argument that HFT adds valuable liquidity. I side with this view.

So far as markets in commodities and index futures are concerned, funds, investment bankers, institutions in general and commercial hedgers do not get in the way of the small independent traders. They help to provide the essential liquidity which makes markets freely liquid for very small players to use. It is pathetic how often posters at ET blame big players for market movements which frustrate them. These large players have not the slightest interest in small retail traders and what they are doing, still less does it matter to the bigger players whether or not very small traders shoot themselves in the foot every day.

Good point but lets keep the discussion under wraps. We're not interested in the merits of HFT per se but how best to utilize it in our interest.

 

[fullautotrading]

Quote from schizo:

You can certainly hedge by utilizing a similar instruments. For example, 3 lots of YM for every lot of ES, etc. It's quite easy to implement. The key is to knowing when the optimal time is to unload.

You are right schizo.

The "time" is actually dictated by the current position (in turn, caused by price action), as one wants to avoid too large drawdowns.

There is actually one reason why i did not stress this possibility.
Using a highly (positively) correlated instrument, in my opinion, is a way always "dominated" by using the same instrument. In the sense that it's obviously always better to use the trading instrument itself than a (correlated) "surrogate".

The reason some people miss to consider that is probably because it needs an "overlay" of "trading units" (scalpers) which only a trader who works with an experienced programmer can implement.

Some people get fooled by the (naive) idea that it seems impossible to be long and short in the same account for the same instrument, because of position "neutralization".
What it is, instead, done by strategist is to write an implementation which keeps alive multiple"traders", or "trading units", if you prefer, by keeping track of their "virtual" positions. Each "trader" will have it's own (virtual) position even though the final actual position in the account is equal to the algebrical sum of the (virtual) position of all traders.

Then, when it comes to issuing orders, a "pool" order is built, by collecting the "requests" of all traders and then issuing order equal to the algebric sum of all requests.

Strategy overlaying and having all under control in one instruments helps coordinating the multiple trading units, to attain the best hedging action.

Tom

 

[dealmaker]

It is an art and its not for every one and each sclaper will have their own variations, that being said like any strategy it takes time to develop and make artistic, otherwise its called a loss.

 

[NY0BScalper]

Scalping to me's about watching the liquidity and catching the erroneous orders, the guys slow to cancel, the resting limits, and putting your money on the break through support/resistance or leaning on it holding.

 

[schizo]

Quote from dealmaker:

It is an art and its not for every one and each sclaper will have their own variations, that being said like any strategy it takes time to develop and make artistic, otherwise its called a loss.

Note that this thread is not created for intellectual masturbation. Please refrain from empty rhetorics and be as specific as you can when illustrating your point.

Per art vs. science: Since a scalper is not so concerned about capturing the entire trend, one's method of execution must be flexible enough to make decisions on the fly. However, the overall system should be robust enough to to be able to replicate itself time after time.

 

[schizo]

Quote from NY0BScalper:

Scalping to me's about watching the liquidity and catching the erroneous orders, the guys slow to cancel, the resting limits, and putting your money on the break through support/resistance or leaning on it holding.

I believe HFT (call it black box or algo or what have you) totally disrupted the market dynamics. As far as I'm concerned, the order flow is not as smooth as it used to be. Whereas the price action cum tape reading was once essential in gauging the market sentiment, now you must contend with price dislocation on a tick basis. According to the textbook definition, the overarching goal of HFTs is to provide (sic) liquidity without leaving footprints. But after all, they are no more than just another algo on steroid and it's my view that they can be detected upon close inspection. More to be dissected on this elusive subject in the coming days.

 

[tyrant]

Quote from fullautotrading:

You are right schizo.

The "time" is actually dictated by the current position (in turn, caused by price action), as one wants to avoid too large drawdowns.

There is actually one reason why i did not stress this possibility.
Using a highly (positively) correlated instrument, in my opinion, is a way always "dominated" by using the same instrument. In the sense that it's obviously always better to use the trading instrument itself than a (correlated) "surrogate".

The reason some people miss to consider that is probably because it needs an "overlay" of "trading units" (scalpers) which only a trader who works with an experienced programmer can implement.

Some people get fooled by the (naive) idea that it seems impossible to be long and short in the same account for the same instrument, because of position "neutralization".
What it is, instead, done by strategist is to write an implementation which keeps alive multiple"traders", or "trading units", if you prefer, by keeping track of their "virtual" positions. Each "trader" will have it's own (virtual) position even though the final actual position in the account is equal to the algebrical sum of the (virtual) position of all traders.

Then, when it comes to issuing orders, a "pool" order is built, by collecting the "requests" of all traders and then issuing order equal to the algebric sum of all requests.

Strategy overlaying and having all under control in one instruments helps coordinating the multiple trading units, to attain the best hedging action.

Tom

So essentially the "final system" is a collection of individual strategies implemented on the same instrument? Or, is there a main system with multiple "hedging" systems? Would be great if you can give a simple example.

Thanks