Nova Group · algorithmic research since 2020

Price is a
consequence of
order flow.

We collect the trades and orders in the market's depth at levels 1 to 3, simulate the venue's execution engine precisely, and build and test trading algorithms on top of it.

Our method Projects
Delay
Size
Price

Try it yourself: choose a size and an order type, then send it.

An order-level (MBO) book — a simulation, not market data.
  • Background Since 2020

    Studying financial markets and building trading algorithms, from then until today.

  • Data Prices, trades, book
    PricePrice
    TradesTrades
    BookOrders in depth
  • Markets Three kinds of market
    CMEChicago futures & options
    CryptoCentralised & decentralised
    TSETMCTehran stocks & options
  • Methods Data science & models
    DSData science & statistics
    ML·DLML and deep learning
    RLReinforcement learning

Our method

Three layers, from the bottom up

Every project is built on these three layers. The upper layer does not exist without the one beneath it.

  1. Foundation

    An independent platform, built per project

    Each project gets its own data and execution platform, built from the ground up and independent of any off-the-shelf tool.

  2. Engine

    A deep order-execution simulator

    We rebuild the venue's execution engine: orders queue, wait their turn and fill in part.

  3. Intelligence

    Analytics, charts and predictive models

    From the flow of trades and orders, through statistics and model training, to a prediction algorithm.

The method in full

Two tracks

We make liquidity, and we take it

Both tracks draw on the same understanding of order flow, and each has its own projects.

Maker

The market-maker view

Supplying liquidity and making markets: our orders wait in the queue on both sides of the price, and others trade against them.

Taker

The taker view

Consuming liquidity and entering trades: our order crosses the spread and takes the resting orders from the front of the queue.

Projects

What we are working on

In testing

CME Market Maker

A deep-learning market-making algorithm for the Chicago futures market, on the way to an automated market-making desk

Several years of the Chicago exchange, gathered order by order. From that data we have arrived at a market-making algorithm that is tested inside a close reconstruction of the exchange itself.

Simulator report to download Read the project
  • Multi-yearorder-by-order history
  • Deep learningmethod
  • Python · C# · Rusttechnology
In testing

Fast Scalp

An ML-based short-term trading algorithm for the Chicago futures market, on the way to an automated trading desk

Several years of the Chicago exchange, gathered order by order. From that data we have arrived at a short-term trading algorithm that is tested inside a close reconstruction of the exchange itself.

Simulator report to download Read the project
  • Multi-yearorder-by-order history
  • Machine learningmethod
  • Python · C# · Rusttechnology
Data service running

BlockTrade

Behavioural analysis of block trades in the Chicago futures market, on the way to trading and investment algorithms built on the trail of big money

Our server receives and processes the block trades of every Chicago exchange product without pause and delivers them to the trader's chart. Their history builds up day by day.

No simulator report published Read the project
  • 36 secondsmedian arrival of each block after the exchange's notice
  • 24 shapesblock classification
  • Python · C#technology
In development

TSETMC

Order-flow analysis of the Tehran Stock Exchange, on the way to an algorithmic money-management fund

Eleven years of Iran's capital market, gathered and rebuilt trade by trade and order by order. We build and test our models on that history.

No simulator report published Read the project
  • 11 yearsof market history
  • 1,148symbols
  • Python · C#technology
In testing

Gold ETF Arbitrage

An algorithm that moves capital between the gold funds of the Tehran Stock Exchange, on the way to an automated gold portfolio that aims to earn more than holding gold

The gold funds of the Tehran Stock Exchange, recorded order by order and measured on one yardstick. We test moving between them inside a close reconstruction of the exchange, against simply holding gold.

Simulator report to download Read the project
  • 21 gold fundsmeasured on one yardstick
  • 395 trading daysreplayed in the simulator
  • Pythontechnology
Under study

DeFi LP

A money-management algorithm for the liquidity pools of decentralised exchanges, on the way to an automated fund

The full history of 136 pools on four blockchain networks, gathered from each pool's first day and event by event. We simulate capital sitting in a pool, and moving between pools, inside a close reconstruction of the market itself.

No simulator report published Read the project
  • 136 poolsfull history, event by event
  • 4 networksblockchain
  • Python · C#technology
In development

Crypto Market Maker

Data infrastructure and a market-making algorithm for the crypto market, on the way to an automated market-making desk

The market of 22 heavily traded cryptocurrencies on 16 global exchanges, gathered trade by trade and order by order. Before writing the algorithm, we have recorded this whole market, and keep recording it.

No simulator report published Read the project
  • 22 cryptocurrencieson 16 global exchanges
  • 60 terabytesscope of raw data
  • Rust · C#technology
All projects

From order flow to capital flow.

Get in touch