Algo list by class:


Testing the Sniffer Algo


Autoscalp price engine ©

Dedicated modules

Dedicated algos

Algo clusters

Geopolitical Sniffer ©

Virtual Grey Pool ©

Example of specific strategies for Oil Sector

Prototype of a Proprietary Oil Price feed

Examples of Event Driven algos

Examples of Geopolitically triggered Algos

Examples of Counter Profiling Algos (samples)

Example of a raw Alpha from idea to algo.







Autoscalp Sniffer ©




Algo Class: price engine core.
Trigger: counter-profiling model.
Covering : eurusd sp500 dax dollar index gold.

Specifically designed for:
Asset Managers, Managed Futures Funds, Alternative Investment Funds, Dark Pools


Uses a proprietary HFT Alpha seeking model covering: eurusd, sp500, dax, dollar index, gold
Results of a proof of concept tests for a Fund done in Dubai :
1433 Safal - Shaban 2012
January to July. Max leveraged exposure: 6.
Instrument used were: euro dollar s&p500 dax oil gold.
Performance resulted steadily positive on a monthly basis.


More data on performances are available only after mutual NDA and NCA have been signed.
See the Partnership Page after reading the disclaimer below.
Further MAQASID SHARI‘A based performance evaluation are also available on partners' request.

Autoscalp model works as a distributed architecture capturing multiple signals
conveyed from different sources
into the Autoscalp Simulated Dark Pool , as follows:


Module Name Icon About the Module...



Artistic interpretation of our Dark Pool Sniffer ©

identifying a potentially profitable trade opportunity into a dark pool.


Schematics of the Signal Sniffer Connector linked to the Virtual Dark Pool Conveyor

Multiple signals Conveyor

(old version)

Autoscalp Signal Matrix

the conveyor manages mutiple incoming raw signal messages.

While the Signal Conveyor creates an internal feed from multiple inbound trading signals coming from different sources, the virtual dark pool creates

four internal concurring prices for each index,

concept which is the base of our proprietary autoscalp pricing model .

Autoscalp Virtual Dark Pool



the virtual dark pool creates 4 concurring internal prices built from:

a dark pool price

an order book price

a block trade price

a sentiment price

Sniffer and counter-profiling module

Counter-profiling Module

Like our grey pool , the autoscalp pricing model sniffers concur to simulate and profie most players.

We are given a definite competitive advantage and a robust situational awareness in return.




In this case we would enter long below prime brokers at 1.3470
and stay long until dark pools high price offer is touched at 1.3556.

Vertical Markets
(oil and gas)

Oil tracker signal feed

Using many proprietary HFT tecniques our Oil Tracker model
covers currency risk while seeking alpha on oil prices either on buyside and sellsde.

Event Driven Algo Manager




See: Event driven algo examples below .

Geopolitical Event Manager


geopolitical event manager


See: Geopolitical algo examples below.

Scenario based trading

First a Think-tank meeting creates a set of dynamic rules for and against a certain event . Then our model tracks if the situation is going toward a certain pole or away from it and
trades accordingly

Price Blocker Module

autoscalp price filter

[Under development]
filters prices before they enter the platform using swarm intelligence tecniques.

It behaves like a firewall identifying and blocking bad price feeds such as: quote baits, gaps, spikes, lags, artificiously injected or manipulated prices.

Asyncronous pre-processing module.



Sipping Market data with many straws.
Rather than allocating a datastream path to a specific machine we left the task to a non-sqlDB running in a shared ram which distributes the data to be processed in non sequential fifo chunks, idea which increased signal generation and analysis process speed by 30%.

Pseudo-events spotter


counter-manipulation module.

Pool counter-profiling
Typical price manipulations
(price gaming)




Pool counter-profiling


Example of artificious price feeds
(spiking and spoofing)

Our false price firewall and its anti-gaming module identify and defend from many dangers among them:
    • Quote Stuffing: an HFT trader sending huge numbers of orders and cancels them rightaway

    • Layering/masking:
      multiple, large orders are placed passively with the goal of “pushing” the book away of the reai price

    • Book Fading:
      fast reactions to news put pressure on order book to make liquidity disappear and force the triggering of sudden price spikes.

    • Momentum ignition:
      (also called liquidity baiting)
      an HFT firm who is using a sniffer to detect a large order on a price range, applies a pre-emptive tecnique to empty the book ahead of the whale order to gather all liquidity of all incoming orders on that level
    • Pseudo-events:
      a pseudo event is something that is apparent, not real and is artificially casted to create a wrong trigger to the counterpart algos. Can be hidden either in a maniuplated price feed or in a very biased news release.

    • Packet Bursts: random packets intruding and obstruding the normal routing path. Our model dynamically reroutes packets elsewhere before undergoing any lags as keeps sniffing packet traffic and reroutes before getting derailed or clogged.

    • Price Shuttering: Shutter is a known malpractice to fake steady prices while market is moving very fast and that allows swallowing customers stops at once at their worst price possible.


Once a Price Shuttering event is spotted or is likely to happen , a dedicated anti-shuttering strategy is automatically deployed and will contrast such behavior before it does harm.

Controlled Algo Pool


Our controlled algo pool is a dynamic containment environment where even the worst possible algos can become profitable without touching the algo code.

HFT Trading HUD

a roller coaster
first person view

autoscalp HFT HUD


Manual trading is the surest and quickest way to destroy an account

but at times one needs a fast interface to correct positions quickly

Such perspective is very similar to the first person view of a roller coaster...

and our think-tank has come up
with a star wars game looking HUD .

4 buttons: buy,sell,exit,reverse.

Replicant Swarm v 6.4

A replicant strategy is an autonomous strategy cluster, partly behaving like a virus and part like a conglomerate of trading systems.


More about Replicant Swarm v 6.4


Class: algo cluster.


alpha idea:
"You can program a machine how to flex an arm or let it generate the code while it sees the movement".
This paper explains how a model can learn from others, subject which which is called imitation learning.

Conceptually similar to the miners algos used to create bitcoins, where plenty of slave machines are
independently dedicated to produce bitcoins . The base difference is evolution and scouting.

If a replicant algo profits, it clones itself on an available algo slot and starts trading. Becoming a "local queen"
making the swarm generate multiple clones from the most profitable spot. If an algo loses ,
it means all its clones are losing then it triggers a "mutation command" to all the cloned strategies.

That triggers a slight mutation in each machine configuration and in meantime all machines
switch to "demo trading" while the swarm saves energy and restarts evolving , the account saves losses.

Mutation continues until one algo starts profiting. From that point its clones are passed to real market...
and the evolutive loop restarts until either all machines available have been conquered and running the
replicant algo or the profit loss bracket has been reached.

Most comments either from think tank and p2p network could have been summarized as :
"all algos sooner or later start losing" and thus the alpha idea has been modified to focus at the core problem
of finding the best way to stop producing on real while still profiting and switch to demo mode
BEFORE the loss happens. It implicates creating a specific monitoring environment based on effective results,
object which is under heavy development these days.


Ant vs bot
Reason of disclosure of the above logic:

is simply applied swarm dynamics, perfectly working in any ant-hill near you.





Geopolitical Sniffer Prototype:


An interactive algo cluster designed to predate against wrong geopolitical decisions


[based on recurrent information filtering and panel filtering , game theory and a special case of Turing test].


Alpha idea:
"don't trade the news, dismantle its noise".
Algo extract facts from many sources,rebuilds the news and profile the actors relatively to that fact.
Reality check (panel vote on coherence of the news summary created by algo).
Algo finds externalities and which indexes are directly or inversely correlated to the news and trades.


1) Automatic part:

Algo is set to harvest breaking news on a user defined event
and creates a summary from opposite information sources.

1a) Algo adds relates externalities whose influence is spiking aside of the main news.
1b) Algo checks coherence for each actor with the ongoing news.
1c) Algo finds correlated and inverseliy correlated index behaviors to the ongoing news.

2) Interactive part:
2a) A panel votes if algo has summarized the situation properly. Otherwise forces summary rebuild.
2b) A different panel votes the event main actors behavioral history was correctly linked to this news.
2c) A recurent filtering process is required to get the fact without the media spins .

3) Cojoined signal generation:
3a) When news summary is approved by panel algo.
3b) Both algo and panels try to spot an entry and exit signals on the related index (apply game theory).

Recent example:
Is the Ceasefire in Ukraine reliable ?
Russia is more experienced on military matters.
Germany is more experienced in banks matters.
NO (panic, gold up). YES (uncertainty, stay out).

Panel and algo voted: NO and created the panic trigger signal.
Enter long gold before panic, Exit when panic happens.

Reason of algo disclosure:
is just applied game theory ...
a special case of the Turing test where part of the machine is human (panel).





Autoscalp Virtual Grey Pool ©


Example of multiple concurrent prices running inside it.


autoscalp greypool


See also: our Autoscalp Pricing Model - concept.







Example of specific strategies for Oil Sector :


Autoscalp oil Tracker ©


Covering: brent,wti, and partner's owb base currency against USD (eurusd, usdcny...)


Module Name Icon About the Module...

Oil Tracker
for sell side:

  • if oil seller's currency is different from buyer's one
  • if oil seller needs to trim oil price spikes against buyer
Oil Tracker
for buy side:

  • if oil buyer's currency is different from seller's one
  • if oil buyer is exposed to oil price spikes fluctuations
sell side
Oil Tracker

oilyuan for BRICS countries

Objective: sterilize the AED/CNY variations on a weekly basis.

The Model builds an internal syntethic Grey Pool
Oil Price feed
Autoscalp Oil Price Filter An alternative price feed for Oil , a proprietary price feed to run our algo based on our Sniffer filters applied on the WTI Oil price.


sage example : sell side Oil Tracker CNY
Assume an oil producer in UAE has a chinese customer
who agreed to pay a certain oil quantity for a certain price on a certain date.


in this case creating an internal price feed to do counter-profiling of oil/yuan.
Since the longer the delivery period, the higher the currency risk for the oil provider,
Oil tracker model actively trades WTI ,
Brent and their currency components relative to the oil provider (AED,CNY,USD)

For every week yuan moved agains the oil provider currency (AED)
the following week the model actively trades to recover such loss.

Notes on eventual personalization of the Autoscalp Oil Tracker as by Shariah Compliance required steps :


  • The eventual collateral needed for the following week is known before dealing, one week ahead.
  • Collateral is traded at leverage 1. That means no leverage is used, at all.
  • When predefined target brackets are reached the model halts trading.
  • No game of chance as each deal target be it profit or loss is predefined and expected.

    Notably, eventual swap related issues depend from the broker chosen by respective counterparts.




Algo prototype Proprietary Oil Price feed:
A proprietary pre-filtered oil price feed sketch was created during a think-tank while proposing some sketches and ideas.


Research Topic: produce a proprietary oil price feed based on our sniffer filters.
Aan internal oil price feed to be used in our algos rather than relying on the centralized Oil (CL) price.
In this oil specific algo prototype we are applying our proprietary Sniffer filter against:


  • price spoofers
  • HFT gamers
  • paper barrels
  • ghost-collateral
  • bait and switch
  • base currency spikes
  • news gapping
  • ...


Usage Example: often "PAPER BARRELS" are netted (or should we say simply disappear?)
just ahead
of the oil news release thus creating an expected shock on the official oil price.
Objective is producing a far more reliable oil price to be used for our algo computations .
Price won't be necessarily in line with official market price . It will try to filter all possible price manipulations it finds.
Research objective is to increase our proprietary algos effectiveness, not third party market making efficiency.

Here are some discussion highlights emerged during the Think-tank session.
Objective is to create a price feed slightly more efficient than WTI oil price feed.


  • Creating an alternative pricing model for Oil could give us very high visibility in the industry.
  • Consists in applying our Sniffer to filter oil price shocks ( currency shocks, leverage distortions, speculations...)
  • All competitor algos adapt to the centralized price while we would produce an alternative price feed for our algos instead.
  • An internal price feed on oil might give our algos the same advantage market makers get creating prices to their customers.
  • The top speculative moment being when our price feeds dissociates the most from the standard accepted price.
  • Objective of the alternative oil pricing model is increase algo efficiency without sticking to the standard oil (CL) price





Examples of Event Driven algos and Geopolitically triggered Algos:


Autoscalp Geopolitical Event manager: Is a plug-in for autoscalp sniffer composed by various sub modules.
It transforms every geopolitical event in a price feed under any aspect including triggers, envelopes and slopes.

A few event driven algos from Autoscalp are shown below.


Icon Algo Brief
Trigger Type

Test Result

A.I. created

  cryptocurrencies were the last human invented currencies. Next one will be made by A.I. Our Think-tank not fucused on timing and trigger yet.

[to be updated after Davos 2018]
 [waiting trigger]


  massive liquidity drain and market crash, 20 times bigger than 2008 flashcrash Within 2025 ,
as Quantum A.I. will crash blockchain univocal transaction ID
 [waiting trigger]


  if public is granted access to Shor algo computation crack snippets on quantum computers as soon as Quantum computers become affordable all cryptocurrencies might disappear in just one day .  [waiting trigger]

Cryptocrash day zero.

[muppets trap]

  IF Bitcoin ever had some sense was when value was 0.15$ not at 20000$. We expect to abruptly crash.
within Chrismas holidays  2017
as trading volume shutsdown
we'll see

Muppets washing

+33% Profit overshoot.

Algo halted.

Game Over

Eurocrash Specific Algo

Can't create a national superstate imposing a foreign currency.

Never worked in history.

Foreign pressure and PIIGS austerity mishap made
E.U. lack national identity
Brexit ok.

Now waiting next trigger:

or Itacrash
or EU war with Russia.
 [waiting trigger]

EU version

  70 trillions european derivatives are concentrated in  a few german banks but it relies on 3 wrong bets:

1) China and US are to keep buying EU bonds without ever checking what they are buying

2) they give for granted interests will be paid and thus are not considering altenatives

3) such loophole justifies reciprocal central banks fiat money policiy

but what if money flows elsewhere.
Example to places where conditions start improving ?

Trigger is
economic boom
or in GCC
or Africa
or next 11 countries

 [waiting trigger]

Quitter le navire !

 Frexit Specific Algo

Not a new concept, is just an European crisis consequence.

Although firmly denied by official media.
After Italy and UK voted to EXIT EU at their referendum, was up to France to leave ths sinking ship and accelerate EU implosion. Wait for trigger


Frexit posponed as Le Pen lost elections.



Itacrash Specific Algo

Hypothesis is that Greece and Italy bankrupted in 2012 but was denied by official media.

Free elections were forbidden while
austerity and IMF inspections increased their weights in PIIGS economies.

spike spotter triggers are:

ECB and IMF inspections triggering Ita-Banks domino.

as  btp-bund spread touches max : buy.

(buy on panic rather than sell on hope).

+ 8%
still running

A muted

Gold Panic on US Election

While mainstream media chose a pseudo-statistical "denial" of real information We expected a Gold spike on Trump victory...


Algo halted after US election result


Skyfall Brexit Specific Algo

Used HFT to position on the day before brexit with the following basket:
WTI short
EURUSD short XAUUSD long
since feb. 2016 our Think-tank expected Brexit to happen.
"Brexit will be just first of many".

Algo halted
as eurusd touched 1.0666



Backfire Algo

Active defense Algo for Oil tanker management companies .

Counter-profiling oil price shocks and price manipulations.
Does price arbitrage between a delocalized pool and centralized oil market pricing

+20% Profit overshoot. Algo halted.


Silppery Oil As Russian economy can't withstand oil price touching 15$... Sell oil till 20
buy below 20

+20% Profit overshoot. Algo halted.
Algo started in Dec 2015


  Centralized pricing , shale oil and paper barrels keep oil price down but in meantime a deteriorating geopolitical situation says the opposite. For each week oil stays below 42 $ algo will buy .

After a stop at -4% algo was allowed to restard and reached a
+20% Profit overshoot. Algo halted.

Started september 2015.

Shanghai Crash

great volume spike but whose collateral ? Short HK Exchange when media posts concept of Shanghai improving. Shanghai increased trading speed without a coherent collateral increment.

Algo will short until one is to match the other.
Algo started in august 2015 .
Profited 2% month
and stopped out at -4% in first days of 2016


Grexit Countdown

Domino For each "Varoufakis"
ttweetcount heap
short 10M eurusd
Each ECB or IMF 'help' for Greece has had a progressively reduced timespan:

6 months,
4 mohths,
2 months.

+20% Profit overshoot. Algo halted.

Was expected to run until end of october 2016

Reversing the

' no need to know '

EU: puppets don't need to know Sentiment harvester:

the higher number of C.V. we receive from PIIGS countries each week

the more
we buy gold
EU govts keep selling the lie that EU crisis is 'almost' over

but they know is a lie and speculate on it. How about reversing the trick?

+20% Profit overshoot. Algo halted.

Started august 2015. Profited 2% month.

Running till end of may 2016

Paper barrels can just fuel...
paper cars!

Sotland no and oil up

Artificial price filtering:
WTI oil (paper barrel) was at 103 while in real world oil barrels were sold at 67.

Our Sniffer set an expected profit target at 85 ,mid price between them, and thus started selling from 96 .
Paper barrels can't be delivered
paper/real spread widens,
correction is imminent.

Shorted at 96
exited with profit profit at 85.6

+10.4% at leverage 1
with a single deal.

Bear Trap

beartrap panic

spike/killbox trap:
application of the Square Wave Filter Model:
Filter the geo-political situation from currency price

Expecting USD/UAH return to 9.5 as panic eases.

No WW3 for Crimea!

counter sentiment
and breaking news harvester:

if string "Russia,war,Ukraine" found in breaking news

Sell oil until price level returns before the Ukraine crisis
(102 $)
Crimea which is far less strategic today than was in 1853.

Almost all positions were closed at profit at leverage 1 maximum exposure with a +5% final profit in less thean 2 months.

Say Cyprus?
Say Short !

Say Cyprus say short: +18% in one day.

find specific word in the news:

news harvester
trend spotter
Alpha idea: if you hear "Cyprus" time to sell.

Test result: +18%
in one day.


Crash !

international market failsafe:

panic propagation

market failsafe exploit
when market failsafe is "on"it means to stay short

Test result: +2.5%
a day for a week.


the 4 july

multiple news handler:

ECB interest decision,
4 july holiday,
Non farm payrolls
spike hunter
with 3 news and no liquidity... stop hunting rules the show

Test result:+20%
in two days.





Examples of Counter Profiling Algo :


Name: sniffer swarm
Class: counter profiling
Type: undisclosed
trigger: undisclosed

Alpha idea:
the fine art of muppets hunting, aka the six "S" : skewing, slippage, spreading, spiking, spoofing, sniping.


On 11 december 2013 we have set our autoscalp proprietary Sniffer and our proprietary pricing model togo stop hunting against retail positions (counter profiling) during the 1.38 eurusd barrier breakout day. Result astonished us too: +18% at leverage 20 , that is 0.9% at leverage 1.





25 november 2013 autoscalp proprietary Sniffer and our proprietary pricing model operating on spot.
This shows the prototype of a new class of sniffer algos, which are mostly concentrating deals hunting for diluted counterpart stops , hidden in fragmented iceberg positions .



5 november 2013 autoscalp proprietary Sniffer and our proprietary pricing model operating on spot.
It profited merely exploiting the flawed OHLC pricing model used by most counterparts.
HFT Model positioning was broadcasted live on our twitter page resulting in +0.6% at leverage 1.


Testing the Sniffer Algo







Example of a raw Alpha idea:


(Algo creation in think-tank meeting)


Name: anti failsafe (US SHUTDOWN DAYS)
Class: counter profiling
Type: spike hunting
trigger: sentiment over-reaction exploit

Alpha idea:
While shutdown was public domain and its panic was not eased
but just stabilized , a smaller amout of news was enough to trigger
a market over-reaction as volatility spikes
tended to ingnite with much less energy than usual.
Producing very fast high spikes on few instruments.

Liquidity pockets were funnelled much faster than usual into fewer instrument ,
situation which tends to dilute and dissolve fast at first signs of panic easing.


average profit result: +3.83% each day it traded.


The days the algo was allowed to trade during shutdown days were
only the ones that triggered further panic :


  • the day when #shutdown hashtag reached its peak
  • the day when payrolls data were NOT released
  • the day when gold market crashed
  • the day when shutdown was removed




Example of different algos running in sequence:




Switch from trend follower to event driven


october 16 2013


Testing the A.I. Algo while U.S. Shutdown ended


october 11 2013


A.I. model on the Gold Market crash day


october 4 2013


A.I. model in a US shutdown day


october 3 2013


A.I. model in a US shutdown day






Copyright Notice: our proprietary algos have to be considered intellectual property
and are NOT avaliable for public access or grid distribution. See disclaimer on bottom of page.