Leading Indicator for Short Term Price Movement

H

hyperscalper

Here's a typical example of a Leading Indicator of future price movement in USD/JPY. The prediction does not use Price but, instead, analyzes the Dukascopy Depth of Market to determine the near term price move.

HyperScalper

TrendPredictor-USDJPY.PNG
 
That looks interesting, the problem is that the lead time is likely dynamic, do you use other tools to decide when to pull the trigger?
 
jelite said:
That looks interesting, the problem is that the lead time is likely dynamic, do you use other tools to decide when to pull the trigger?
More...

Yes, good point. In "micro scalping" there need to be specialized indicators which help the scalper decide when to "pull the trigger".

First of all, my basic approach is "incremental trading", so we rarely go "all in". So taking a position is not so much of an "event" as a series of events, or a process of Entry.

The screenshot in the post above, shows us the direction of the micro trend. We attempt to take multiple Entries in that trend direction, each on a slight pullback, but which pullbacks?

Forex does not offer Time and Sales, so I synthesize a "virtual" Time and Sales. When large trades in the VT&S, specifically, correlated with large Retail sales, then the Trigger System will take a Long Entry, subject to that event plus a few other criteria. So it is "a micro pullback" or "micro countertrend" entry.

In a nutshell, a derivation of a T&S correlate, the VT&S is done by examining successive Depth of Market (aka Book) snapshots, and taking Delta values. If you imagine a large Retail player who sells against the Book, what happens is that one or more BID Tiers of the Book are "gobbled up" in satisfying the size of that sale.

By examining, equally on both sides of the market, the amounts by which this theoretical trade "consumes" Bid tiers or Ask tiers, then we can identify events which are excellent candidates for Entry actions.

In fact, as I was writing this reply, a Long scalp was entered by exactly this method so I am posting the screenshot of the live trade with an automated entry, and fixed target closure. In this case it's just a single entry order, rather than multiple ones; but the key fact is that the trigger to Enter Long was taken on a "big lot" retail sell trigger, as described.

The Red/Blue is price, and the other lines are a couple of other of these secondary indicators which help to support trend detection. Here are the two LMAX order fills with the LMAX Live Order IDs which are depicted on the chart. (Times are CET, where our server is located.) I am controlling the GUI and enabling the basic trigger parameters from North Carolina, USA via VNC.

You see that London time is GMT and GMT+1 is CET where the dedicated Quad Xeon server lives in Germany.

HA HA. By total coincidence in the snapshot the BID price shown is exactly the price at which the Long Entry order filled... sheer coincidence as the price came back down after the Long position was closed. This illustrates the narrow price ranges which micro scalping attempts to trade successfully.

Code: 
01-20 20:15:39.767 USDJPY BUY B<* (2.0)  wPI: 2.00 qty: 0.015m @ prc: 118.775 ask: 118.777 msecs: 75 LMAXoid: AAGJIwAAAAJtmbRu
01-20 20:18:02.874 USDJPY SEL *S (1.0)  wPI: 1.00 qty: 0.015m @ prc: 118.799 bid: 118.798 msecs: 22 LMAXoid: AAGJIwAAAAJtmbUU

HyperScalper
LMAX-Long-MicroScalp.PNG
 
Last edited:
Just a little more on the prototypical Micro Scalping Incremental Pattern.

Pulling the trigger becomes multiple Entry, so each individual entry is less critical.

Nevertheless we need to know, in this case, that the real trend is UP and, fortunately, we know that, since Price is Lying to us nearly all the time ! :) In this case, despite the fluctuations in Price (...lies, lies...) the Trend was constantly UP.

HyperScalper

Here are the live trades in this sequence shown in the screenshot. Here we are using "resting" limit orders so in most cases we do see price improvement. This is a different style from "triggering" where the trigger strike is normally an immediate fill, similar to a market order, but using a limit order.

LMAX-Incremental-Scalping.PNG


Code: 
01-20 23:18:04.518 USDJPY BUY B<<<* (4.0)  wPI: 4.00 qty: 0.002m @ prc: 118.716 ask: 118.72 msecs: 69 LMAXoid: AAGJIwAAAAJtm7k+
01-20 23:21:15.469 USDJPY BUY B* (1.0)  wPI: 1.00 qty: 0.002m @ prc: 118.708 ask: 118.709 msecs: 69 LMAXoid: AAGJIwAAAAJtm7rn
01-20 23:24:10.822 USDJPY BUY B<<<* (4.0)  wPI: 4.00 qty: 0.002m @ prc: 118.694 ask: 118.698 msecs: 70 LMAXoid: AAGJIwAAAAJtm8HT
01-20 23:29:33.090 USDJPY BUY B<<* (3.0)  wPI: 3.00 qty: 0.002m @ prc: 118.684 ask: 118.687 msecs: 72 LMAXoid: AAGJIwAAAAJtm8dA
01-20 23:31:20.912 USDJPY BUY B* (0.0)  wPI: 0.00 qty: 0.002m @ prc: 118.622 ask: 118.622 msecs: 0 LMAXoid: AAGJIwAAAAJtm8rK
01-20 23:36:51.418 USDJPY SEL *>S (2.0)  wPI: 2.00 qty: 0.002m @ prc: 118.656 bid: 118.654 msecs: 71 LMAXoid: AAGJIwAAAAJtm8uM
01-20 23:51:37.371 USDJPY SEL *S (1.0)  wPI: 1.00 qty: 0.002m @ prc: 118.729 bid: 118.728 msecs: 70 LMAXoid: AAGJIwAAAAJtm9Hd
01-21 00:04:59.223 USDJPY SEL *>>>>S (5.0)  wPI: 5.00 qty: 0.002m @ prc: 118.738 bid: 118.733 msecs: 70 LMAXoid: AAGJIwAAAAJtm73f
01-21 00:04:59.573 USDJPY SEL *>>S (3.0)  wPI: 3.00 qty: 0.002m @ prc: 118.743 bid: 118.74 msecs: 69 LMAXoid: AAGJIwAAAAJtm70b
01-21 00:05:07.173 USDJPY SEL *>>>>S (5.0)  wPI: 5.00 qty: 0.002m @ prc: 118.748 bid: 118.743 msecs: 73 LMAXoid: AAGJIwAAAAJtm8aU


Again, TREND is absolutely critical, and we had a clear UP TREND during the entire process.
 
If I understand it correctly, you enter a LONG on identifying a large sell order. Please correct me if I am wrong here. That would correspond with a rather general observation that there is 'price to be paid for liquidity'-i.e., a large sell order that didn't consume all the available liquidity will likely be followed by a temporary pullback (i.e. price going up). What I find interesting is the creation of a proxy for volume. Clearly you must adjust for 'fake' depth in your estimates. Do you think your system is applicable to futures? There you have the volume but not so much the knowledge of who the likely participant of the transactions is. In any case, for people still trying to find the right cross of MAs, reading your posts might be a wake up call that things might actually be working slightly differently in reality.
 
jelite said:
If I understand it correctly, you enter a LONG on identifying a large sell order. Please correct me if I am wrong here. That would correspond with a rather general observation that there is 'price to be paid for liquidity'-i.e., a large sell order that didn't consume all the available liquidity will likely be followed by a temporary pullback (i.e. price going up). What I find interesting is the creation of a proxy for volume. Clearly you must adjust for 'fake' depth in your estimates. Do you think your system is applicable to futures? There you have the volume but not so much the knowledge of who the likely participant of the transactions is. In any case, for people still trying to find the right cross of MAs, reading your posts might be a wake up call that things might actually be working slightly differently in reality.
More...

Good observations, good questions. Yes, creating a "correlation" or "proxy" for a real Time and Sales involves some assumptions. In general, however, so long as the same treatment is given to the Bid and Offer sides of the book, then we can assume some fairness even though we know there is "noise" and "spoofing" and other randomness occurring on the book.

Nobody knows how to correctly interpret the Book, or the Time and Sales, not really. But it no longer much matters to me, frankly, so long as our own business is stable.

Can it work for Futures, where there is a "real" Time and Sales? Absolutely Yes, been there, done that; and if you want to discuss, get in touch with me. I left the Futures area because my then clients were "underfunded" and "overexposed" for Futures trading, and I thought moving to the Forex arena would be a good choice. Fact is, it's a crappy market, so it was a mistake to get into it.

But, as most marketers know, you can't easily create a new market. Futures traders will remain there, and most likely will not switch to Forex (considered the dangerous and unregulated "wild west") but remain in the Futures area. But when you control 1 single Euro FX 6E contract, your commitment is $12.50 PER TICK. Forex allows 1/100th that exposure, and that opens up a lot of possibilities, but it won't convince Futures traders to switch because CNBC talks about Futures, not about Forex trading. And the Forex market is saturated by pure SCAMS, so I give up :)

I could probably resurrect my Futures stuff, but that's water under the bridge. Yes, you can certainly apply these same analytics profitably in Futures. But, then, who cares? I've left the "client business" and now have partnered in a real trading business, so "begging" clients for money, especially in the Forex area, will soon be a thing of the past.

I'd be willing to work with a committed Futures trader on this but that's a lot of work, and I've moved on to our own Forex trading. Technology sharing with Forex traders is a possibility but I've specialized myself in such a tight niche that it's also not a viable "client business". However, it is a valuable proprietary trading business, so that's the plan :)

HyperScalper
 
Hi hyper, are you tools for microscalping avaible for LMAX?
I knew you from Dukas forum, i love your stuff.
 
You must log in or register to reply here.
Back
Top