Retail day traders and Micro Scalping?

I wanted to kick of a discussion of something I've worked on for years, which I call "micro scalping". I define micro scalping as hunting profits in a price range of from 2 to 10 pips. Usually a fairly large number of small trades are executed.

As in all trading, Analytics are absolutely required in order to "predict the future" which is what all trading aims to accomplish.

Clearly, trading or scalping for such a small range of price brings up critical issues which are often brokerage specific. Professionals require brokerages which are often called "ECN" (or similar language) and certainly not the old "dealing desk" approach which is rapidly disappearing for savvy traders.

Key factors are:
1) the tightest Bid/Ask spread possible
2) the lowest commission per transaction possible
3) proportional commissions, rather than "per trade" fixed commissions
4) speed and accuracy of Order Entry and lack of "slippage"
5) usage of limit orders or low latency automated systems

I use Dukascopy and especially LMAX to achieve the highest precision and the lowest costs, but increasingly all Forex brokerages are moving away from dealing desks and toward more "transparent" trading venues since professional individual Forex traders won't tolerate the old "rip off" model, so brokerage quality is forced to increase as savvy traders leave the uncompetitive brokerages.

In all markets, there's a considerable amount of time spent in "congestion" or "chop", and this may be 90% or more of the time. If a trader can consistently win in these "congestion" periods, or micro scalp, then at nearly any time of the day, in principle, profits can be taken. However, obviously, when looking for smaller price moves, in order to make things worthwhile, larger lot sizes may need to be taken. But larger lot sizes increase exposure and risk, so Analytics become even more important to support moving to larger position sizes.

My personal view is that traders should not only day trade or micro scalp, but should also have "diversified" trading operations which involve Swing type trading, and preferably even some sort of automated trading based upon technical analysis.

So I'm wondering if you call yourself a micro scalper day trader, by this definition of looking for profits of 10 pips or less in day trading, what is your approach to Analytics and Trading ?

HyperScalper

Not much interest. I'm attaching an image based upon advanced analytics research work I've been doing on predicting micro trend movements on LMAX by analyzing the Depth of Market (aka "The Book").

In the image, we are able to determine a couple of minutes in advance that the "high side" or the Ask/Offer side of the book is having size "pulled away" from the inside market. The Red/Blue is the Bid/Ask price measured against the axis on the right.

But the indicator is the Yellow line, measured against the leftmost axis where Zero or Null is midway down the chart. So we see the indication generally persistently elevated prior to the move.

This makes sense, in preparation for a fairly rapid lift of the market as Market Maker generally pulls Offers further away from the market, since the price will be moved up.

The move is 8-9 PIPs and there's plenty of advance indication to allow us to get in for the <10 pip move which follows. These are very sensitive measurements to get reliability.



Good Trading !


HyperScalper

What is the line that is changing from yellow to white that for the most part is above the yellow bias line ?

Trader333 said:

What is the line that is changing from yellow to white that for the most part is above the yellow bias line ?

Click to expand...

That line predicts where the near future price is likely to be moving.

Good question, this is on LMAX and is using the LMAX API which provides us with 5 Depth of Market levels on both Bid and Ask side. Now, that doesn't sound like much, but we can extract a lot of information from that.

Basically, stabilization algorithms process each Book update, up to about 12 per second are provided. This is USD/JPY so keep in mind it's pretty liquid and probably won't have "gaps" in the book where there is a price level near the market without any Bids or Offers. So it's what I'd call a fairly "dense" book, as opposed to a "sparse" book where there exist price levels near the market which do not have any Market Maker quotes present.

Having said that... So we find the VWAP of the Bids and that gives us a price which I'll call the "Virtual Bid" price, and we compare that with the "significant" Inside Bid Price. There I say "significant" because on LMAX we can have "trivial" amounts on the Inside Bid, and these are eliminated so we have a more "stable" inside Bid reference price (not using "trivial" inside values).

So we get a Virtual Bid Price (VBP), versus the Significant Inside Bid (SIB) Price and there is a PIP distance of the VBP from the SIB. We "sample" that PIP distance within a Specific sampling object which is "locked" to that SIB. Same for the Ask side, we sample the Virtual Ask Price (VAP) versus the Significant Inside Ask (SIA) price and there is a PIP distance.

In general, the "Virtual Bid/Ask" prices taken in this way, will be Away from the market's Significant Inside Prices.

By sampling these "Delta" values at specific Averagers, "locked" to precisely those prices, we get stabilization of the Delta Bid vs Delta Ask at each specific price level. (I call these PriceLevelAnalyzers because they are averagers which are accessed by a Map or Hashtable, and "mapped" to a very specific Inside Bid or Ask price level.

So these samplers exist at every possible price, near the market, and they populate at their specific price values as the market itself moves up and down. But we're trying to estimate a "delta bias" from the market and reduce a lot of the inherent noise.

In other words, we try to Factor Out the fact that Price is moving up and down, and thus "polluting" our estimate at each price level. I wrote all of this complexity, but it's aimed at extracting a very simple answer: UP, DOWN or SIDEWAYS...

BOTTOM LINE: When Market Makers are going to lift the market significantly, they will pull their Virtual Ask Delta pricing HIGHER, and AWAY from the market. This makes sense because they are "getting out of the way" in preparation for lifting the market. Similarly on the Bid side, they will lower the Virtual Bid Delta, thus getting "out of the way" prior to moving the market down by a significant notch.

The Theory of Micro Trend Prediction thus says that BEFORE the price move takes place, there is a detectable Delta Bias where Market Maker size quotes are pulled "away" from the market in the direction where the market price will be moving in the very near future.

WHEW !!! But that's actually what we're doing here, measuring a realtime Predictor of Market Maker's micro trend intentions, hopefully BEFORE the price actually moves !! To be a Successful Micro Scalper, this is the sort of thing you must be able to measure, which is why I believe only Specialist trading platforms can support Micro Scalping consistently and successfully.

Good Trading !
HyperScalper

Hi Hyperscalper

I had not wanted to comment until others had got involved - as I think this is an interesting topic and I do agree with many of your comments - although have to disagree on some of the others

First of all - I am pleased your working charts are as messy as mine ;-)

If your UJ chart was from today - I will say for me to match up - it appeared an hr later ie UK time - are you in Europe or using a different time zone ?

I am attaching my UJ chart as well - using one of my manual scalping methods on a tick chart.

I use linear regressions and time windows along with PA and PS on a tick chart for scalping 2 -7 pips

I would call micro scalping - just 1- 3 pips and anything over 7 or 10 pips for me is just short term intraday.

For micro scalping with tick charts - I would uses 2- 3 pip stop size ( soft ) and one click in and out

For all my other trades I use between 3 and 7 pip stops - to include the spread as well. On FX pairs like the UJ with spread costs under 0 5 pip - then 2 - 3 pip soft stop is OK to work off.

With the GJ or pairs with spreads over 2 or 3 pips - I need 7 pip stops and would only trade for moves of above 7 pips- ie 7 -25 pips with stops that big.

I appreciate you are using an automated system - I am totally manual - and like you I have leading indicators - that tell me a direction in advance to the move.

I rarely take over 20 trades a day - most days 8 -15 trades with only a few under 5 pips as my main targets are with RR's of over 2+ - and so 10 -15 pip moves are great and over 20 pips moves cream on the cake so to speak.

There is a lot I would like to discuss with you and will do over the weekend

Meanwhile here's my UJ chart on CTrader - although I have used Dukascopy as well

Regards

F

I will tomorrow sometime show you my charts on the UJ for yesterday covering the main session and all the main moves - all on tick - 1 min and maybe a 3 or 5 min for the bigger picture that helps with my session price structure bias

All the best

Regards


F

sound much more advanced than anything that any of our other resident experts use, Hyperscalper.
Look forward to hearing more
regards.

scholfield said:

sound much more advanced than anything that any of our other resident experts use, Hyperscalper.
Look forward to hearing more
regards.

Click to expand...

Well, you could just as well say it is a "crazy" level of computation, and perhaps way over the top. Depth of Market Analysis is one of my specialties and I've been successful in using it as a predictor. But it's generally extremely complex, and extremely fast, and way too specialized. And of course Forex lacks Time and Sales which could be a basis for easier Order Flow type analytics.

But I'm always looking for a better way to just make money with less effort. I'm always looking for "clues" as to where Market Maker is preparing to move the market, and it all keeps coming back to Depth of Market Analysis, or directional Volume analysis, always based on the generally accepted idea that Market Maker trades against both Retail Buyers and Retail Sellers, and the assumption that day trading moves are completely "Engineered" planned in advance (a topic which often sparks much debate, but which I take as an Article of Faith). For my sins, it's the path I've taken for so many years.

But, on the other hand, intelligent Swing trading can also make money, and that's the bottom line. Just some nice BOT that does the work for me, and then I could retire all of this intensive day trading stuff, maybe..... I definitely have some BOT plans and a design that I think might make life easier for me....

HyperScalper

4 Charts of the Usd / Jpy for Friday 2nd Jan 2015

Hi Hyperscalper

Before I get down to how I see the "game" - I am posting a few charts from yesterday on the UJ - as that's a pair you chose in your example.

For a start there is a 3 mins - 2/3 day overview of the pair - to see the important levels etc and to get a price structure in advance to the European Session and the levels that need to be breached for either continuation - or for a turn.

Then next a I minute chart - and this is the chart I use most of the time with important areas etc marked etc

Then a Tick chart for up to a few hours of the session

Now this is were I differ with a lot of theorists - as I count all FX charts under 60 seconds as tick charts - whether the tick is from time or from order flows.

Because FX is not centralised and the most crooked and corrupt game amongst the trading world - I discount the majority of order flows - or DOM or level 2 or what ever names you want to call it - specifically because it only from your broker and his liquidity provider - and not the whole picture.

Plus the fact - it can be falsified and set up using cancelled false orders - as well as through dark pools and manipulation caused by LP's holding orders back and then allowing them into the market at times that suit them - normally after they have shown "false sentiment"

More important to me than order flows is time of the day - and 6 key times per hour.

Remember FX is different to stocks and shares etc etc - its a different ball game and because its supposed to be the largest money market in the world - many think its too large to be manipulated. My answer to that is ..... RUBBISH - every trick in the book is used to create false trails etc etc.

OK - I can understand you have found a correlation with DOM and movements - there is one - but time is even a better mover.

Everything a scalper - whether for just 3 pips or 30 pips targets is there in the charts - but so far no chart platform manufacturer as managed to supply it all on one chart - thats why I have to use at least 3 different chart time frames as well as others when I need a big Macro picture view.

I specialise in 7 to 25 / 30 pip moves ideally all within 20 mins.

My focus is making money - not one great trade taking a day with a RR of 5 - instead - 10 average good trades with RR's of 1 to 3 are a lot better - and if I get a few wrong - its no problem - I even have the answer to that with 30% stakes and stops already in profits - so it a win/ win scenario.

I appreciate you expertise is on the computer science / programming side and mine is just expertise in Intraday moves at what I call the "coalface" - ie all charts under 3 mins in time periods - but we have common ground and I have no problem sharing my methods ( I retire within 2-3 years and the last 7 years its served me well )

I think the best leading indicators for FX intraday trading are the eyes and the brain.

The eyes need to inform the brain of what they are seeing and the brain as to filter out that 90% of what's shown is false sentiment - designed to trick the trader and get them to enter a wrong bad trade.

For me entry is key - paramount - the skill for getting the timing of entry to correspond with an interim high or low takes years of study and practice.

But - once you have it - then even if your target or direction does not go to plan - you can still exit with profit.

If I can get an entry 3 or 6 or even 8 pips before a 5 min or 15 min swing trader - even if the new trade is wrong and false and its not going my way for say 15 -25 pips - I can still get out with profit - even if its just one or two pips.

Traders needing stops of 15 or 25 or 35 pips are on the road to failure - whether its takes 3 months or 3 years - OK good money management will keep them in the game - but there winning trades RR's will not be good enough

OK - a lot more to discuss and will follow up later on

hyperscalper,

It is certainly interesting the approach you are using but I guess it can only work on LMAX as forex is not centralised. Also I presume that the vwap you are talking of takes into consideration the volume of bids at the various levels of price depth ?

Because forex is not centralised then if those trading on LMAX are setting up a 9 pip move, does this happen in a very short time frame as how do they know that the rest of the market will follow and if they don't then they are open to huge arbitrage
trading ?

check out forexperians threads..........hes happy to scalp from 5pips+ and very very effective

N

Using Depth of Market Bias analytics to predict

The attached diagram is an example of the situations in Micro Scalping which this platform targets. They are High Probability, High Value, relatively small target Wins in the range of 2.5 - 10 PIPs. That's what I define as "micro scalping".

The objective is to be "correct" 95%+ so that High Value lot sizes can confidently be used in order to make the relatively close Targets yield significant profits.

So there is a high reliance upon realtime technical analysis, which is not primarily based upon price movements. The hope is to find predictors which allow us to enter Prior To price movements taking place. To do this, we try to analyze Market Makers's "intentions" and the clues which their Depth of Market or "Book" quote patterns can reveal.

I've done this on Dukascopy and that software is stable, but my interest is in the much tighter Retail spreads, plus the Wholesale opportunities on LMAX.

In Micro Scalping or "Nibbling" the moves are extremely fast, so we cannot "Chase" market movements, but must be able to anticipate the correct movement Before it happens. Price will not reveal what the near term move will be, but custom Analytics using the streaming Depth of Market can reliably reveal the next price movement. This is the objective anyway

HyperScalper

Hi Hyperscalper

I had heard of some original HFT trader bots on FX getting over 90% success rates - even taking over 1000's of trades a day at milliseconds rates. This was soon "caught out" as others saw what was happening and then many of the advantages were lost as more of these companies with super computers and extremely super fast cable connections joined in. I believe this activity is still going on with the regulators accepting that these players are assisting with liquidity etc.

Obviously - thats on another level requiring millions being spent on systems and development to even just take part in way of trading. I am not aware of any automated systems costing under a few hundred thousands being able to adapt and make on going monies continually from the markets. The sophistication etc needed is extremely complex and costly.

Also - I reckon maybe up to 50 lots per pip - yes you might be able to play their game - but you would not be able to undertake your type of trades with large sums of over a few hundred lots a pip - the market would delay them and cause problems

I think its fine for retail traders and if you have say a quarter of a million dollar capital account and could make say 5% per day return - using only 20 -40 lots sizes per pip - then absolutely great.

Would be interested to find out how large a size you or your investors have used so far ?

Regards


F

Conspiracy Theory in Forex markets ?

Trader333 said:

hyperscalper,

It is certainly interesting the approach you are using but I guess it can only work on LMAX as forex is not centralised. Also I presume that the vwap you are talking of takes into consideration the volume of bids at the various levels of price depth ?

Because forex is not centralised then if those trading on LMAX are setting up a 9 pip move, does this happen in a very short time frame as how do they know that the rest of the market will follow and if they don't then they are open to huge arbitrage
trading ?

Click to expand...

Trader333 you are asking some extremely important questions, some of which call into question our most basic assumptions concerning the Forex market !

Yes, a simple VWAP just calculates on the "momentary DOM". However, significant work can be done "smoothing" the noise we find and stabilizing the DOM but basically that's the fundamental idea. How "close to the market" is the Virtual Bid or Ask? What is that "delta" value, and what does it mean?

My research strongly suggests that analysis on Dukascopy, with a more stable Depth of Market and now a slightly different but similar analysis on LMAX invites us to reconsider the "decentraliz(s)ed assumption" about Forex markets, which is a fascinating hypothesis to explore. So let me explore it !

Clearly many of the major "liquidity providers" participate on multiple trading venues so that is most likely the key to predictability across trading venues when the Depth of Market, or "The Book" is examined in real time. Otherwise, I agree with you that you would not expect predictability at all.

Now, it must be said that the "quality" of the Depth of Market is important. At Dukascopy, I can show persistent long term trends using DOM Analytics even though we are given ONLY 10 price levels which may represent as little as 1 PIP above and below the market. It is surprising that with such a limited "view" of the DOM, that any predictability could be extracted. However, it's extracted continuously by "banks" of analyzers as the price moves or wiggles UP and DOWN so the successive DOM "snapshot" information samples "smear or overlap" thus giving us a wider view than we might at first think we would have...

On LMAX, using the Java API in my case, we are given only 5 levels of DOM at any moment in time, representing as little as 0.5 PIPs either side of the inside market. Plus, the LMAX DOM is "noisier" than, say, the Dukascopy DOM.

I'd like to have experience with some other major DOM streams, but I am certain that predictability and correlation with near term price movements can be extracted reliably.

THIS FORCES US TO CONSIDER hypotheses which might be somewhat "conspiratorial" in nature. However, I propose what most traders "feel" or "believe" to be true, and that is that at least Intra-Day non-news-driven price movements are "engineered" by the large entities which I'll just call "Market Makers" in line with terminology in Futures and other "mature" markets.

After all, Forex is rapidly maturing in the fairness and transparency of its "venues" or "exchanges" and away from the old obviously unfair "rip off" dealing desks and all of that "legacy corruption".

NOW, on a "fair and transparent" exchange, perhaps we'll call it an ECN, there is little or no ability for Liquidity Providers to "pull" their quotes and deny a transaction they "don't like". On LMAX there is no "last look" ability and when Liquidity Providers post their quotes, they cannot refuse to honor them.

THIS IS THE KEY to having DOM predictability. IF quotes MUST be honored by Liquidity Providers, then they will NOT post Quotes which they prefer NOT to honor, and that will be apparent in the DOM data stream. (maybe, if you know how to find it).

So, without going on and on, let me try a partial summary of my hypothesis.

1) Liquidity Providers (LP) or Market Makers (MM) participate on multiple "exchange" venues
2) Increasingly, "last look" or "requote" behavior is not possible for them
3) Therefore, when they post Quotes, these must be honored,
4) With the result that they show "reluctance" to post Quoted size if an upcoming (planned) price move would result in detrimental pricing or undesirable retail transaction flows, to themselves, the major MM's.

This is what I call getting "inside their heads" through technical analysis.

Price movements are "engineered" and planned in advance by the Top Dog Market Makers, so Depth of Market Quote patterns will reflect near term price movements, IF we can detect Quote Bias on the DOM prior to the "engineered" or "planned" price movements. All of this intraday price movement (seemingly unpredictable), is designed to DEFEAT the largest percentage of Day Traders, and has little to do with "fundamentals" which operate over longer time frames.

By the way, LP participants on LMAX, the major liquidity providers, are known to be "unhappy" that they do not enjoy "last look" transaction denial ability anymore. If they post a quote and it gets hit, they can't back out. I'm talking here about Large Retail transactions which "eat" through multiple Quote Tiers, and force MM's to honor the trades they would prefer to deny or to reject.

Therefore, the VWAP delta from the market, representing their major Quote size is made more distant from the market, especially when a "planned" move in that direction is about to happen. I've observed these biases to persist for several minutes before the ultimate move takes place. That's lots of time in the micro scalping world

WHAT DO YOU THINK ABOUT THIS "CONSPIRACY THEORY" ?

Good Trading !
HyperScalper

Forexmospherian said:

Hi Hyperscalper

I had heard of some original HFT trader bots on FX getting over 90% success rates - even taking over 1000's of trades a day at milliseconds rates. This was soon "caught out" as others saw what was happening and then many of the advantages were lost as more of these companies with super computers and extremely super fast cable connections joined in. I believe this activity is still going on with the regulators accepting that these players are assisting with liquidity etc.

Obviously - thats on another level requiring millions being spent on systems and development to even just take part in way of trading. I am not aware of any automated systems costing under a few hundred thousands being able to adapt and make on going monies continually from the markets. The sophistication etc needed is extremely complex and costly.

Also - I reckon maybe up to 50 lots per pip - yes you might be able to play their game - but you would not be able to undertake your type of trades with large sums of over a few hundred lots a pip - the market would delay them and cause problems

I think its fine for retail traders and if you have say a quarter of a million dollar capital account and could make say 5% per day return - using only 20 -40 lots sizes per pip - then absolutely great.

Would be interested to find out how large a size you or your investors have used so far ?

Regards


F

Click to expand...

You make some good points and have a good knowledge of HFT systems.

Of course, as you say, ordinary retail traders cannot and should not attempt to "play their game" on the sub-millisecond timescales, which we know are being used.

Rather, the game I propose is for ordinary traders with modest accounts, who may commit anywhere from 10k nominal position sizes up to several multiples of 100k stakes but nothing which is out of reach of an ordinary account.

BUT I'd like to say that it's very important to "modulate exposure" through multiple staggered pricing entries in order to achieve the desired position sizes.

This is "position trading" and not anything "exotic" like arbitrage. We are just taking a position (Long or Short) through multiple entries, for a short term trade which is usually closed within a fairly short period of time, I'd say from seconds out to perhaps 10 minutes.

The objective for a day trader using such a platform and methodology is to "pull" a few hundred to maybe a thousand or so consistently on a per-day basis. But as there are so many variables, you'll realize that there's no real generalization that can be made.

But this is an approach for "small to medium retail traders" who wish to "make a living" from day trading. Whether that is a huge daily take, or whether it's a modest one that "keeps the lights on" would be a matter of expertise, account size, risk tolerance, judgment and LUCK !

A reasonably "colocated" server would be desirable because their is a "reactive" element to the semi-automated strike logic on my platform, since we always try to make that "extra pip" just on each Entry strike. In my case, I may control the user interface from the U.S. to a European server, but that server has a fairly "tight" network path to the broker's servers.

This is not HFT really, since it's not realistic to attempt to play that game against the supercomputers and Huge lot sizes. Rather it's just reasonably precise micro scalping which can be used by ordinary traders who remain within their "risk envelope" and do not get overly exposed to the market beyond what their accounts and risk tolerances could support, using an Incremental trading technique.

In the end there are many ways to "attack" the problem of profits in trading, and so long as they yield positive outcomes, I guess they are all "valid".

Good Trading !
Brent

A Short Entry using Incremental Entry

Just an idealized example of an "incremental" short entry guided by a market bias indicator. Nothing special about this except that the trader tries to be guided by the market bias, which should give confidence as She adds to the Short position, even though price is rising against Her during the Entry Phase.

Entry to a market should not be an "event" but should be a multi-step process. Exit from the market can be "all out" or can scale out.

We don't want to get overly exposed relative to our ability to tolerate "price adversity" during Entry.

HyperScalper

P.S. I don't know many female traders.....

hyperscalper said:

Just an idealized example of an "incremental" short entry guided by a market bias indicator. Nothing special about this except that the trader tries to be guided by the market bias, which should give confidence as She adds to the Short position, even though price is rising against Her during the Entry Phase.

Entry to a market should not be an "event" but should be a multi-step process. Exit from the market can be "all out" or can scale out.

We don't want to get overly exposed relative to our ability to tolerate "price adversity" during Entry.

HyperScalper

P.S. I don't know many female traders.....

Click to expand...

If the price kept rising when would she get out for a loss.

Reducing the need to Stop Out

12WBT said:

If the price kept rising when would she get out for a loss.

Click to expand...

I think the answer is when the Bias indicator no longer strongly predicts a lower price, then there is consideration of either reducing commitment or stopping out.

If the indicator is "good" then price will not continue to rise unless the Bias indicator shows it. That would be a Great indicator

What I would advise is that Stopping Out should be, in the mind of the trader, the most remote "safety net" possibility. By using an incremental approach, often Stopping Out can be eliminated, or transformed into a "Soft Landing" with minimal damage to the airframe

This Entry process should be a gradual process and by not getting "overly committed" or taking on too much commitment too soon, the need to stop out is reduced.

The graphic is really a bit too simple. Really, in order to maximis(z)e the Short entry price, rather than to Sell in a "linear" distribution fashion using equal size at equal price increments, it would be better to use a "biased" distribution where the larger sizes were progressively used to "weight" the VWAP for Selling even higher than a linear distribution would yield.

So, in simple terms, your later entries would have larger size to pull the Average Selling price or cost basis even higher.

Anyway, the idea is not to have a catastrophic "stop out", but to have a "soft landing" through controlling the exposure level throughout the trade, instead of having just a fixed plan.

HyperScalper

It would certainly make sense for the key MM to be actively involved in several exchanges and that in itself would allow a degree of price manipulation on a short term basis. If your analysis has shown that there is consistent predictability then it sounds very exciting. I can see why being unable to pull your quote has a major advantage to retail traders but how would market orders affect this ?