Skip to main content

FIN QUANT

Artificial Intelligence & Machine Learning
Applied to Financial Markets and Trading

  • Use machine learning to automatically build profitable trading strategies.
  • No programming experience and skills required to build fully functional automated algo strategy.
  • Analyze strategies quality using correlations, Monte Carlo, VAR and more..
  • Utilize machine learning to find the best optimal portfolio from populated algos.
  • Autotrade strategies or full portfolio.

Trading Rules Config

1

Traders can choose trading rules based on a type of strategy they want to build. To build the strategy user can use:

Preconfigured Signals
Indicators
Candlestick Patterns
Chart Patterns
Exit Methods

There is no limit as to number of used inputs. User has all freedom to add his own inputs. This means anything can be programmed and added to the builder. Once trader selects strategy rules, these can be further configured according to user preference and target strategy.

AUTOMATIC STRATEGY BUILDER

2

In algo trading, user has to create and program or build automated strategy. It is almost scientific itself to construct the strategy that would be consistently making profit or that would be able to adapt to continuously changing market conditions. Automatic strategy builder does this instead of users and by applying machine learning methods it searches for the best performing trading rules.

This allows to create algo trading strategies without knowledge of any coding, but instead of choosing specific strategy parameters, user needs to configure builder and it starts creating the algo strategies automatically utilizing machine learning methods. In here to generate algos we can use f.e. genetic programming where strategy parameters are considered as members of population and profit would be fitness function. The strategies with the highest fitness – profit survive, create another generation and further evolve to maximize the fitness.

Strategy Quality Evaluation

Performance Analysis

Each strategy can be analyzed in much details and we can look at the strategy from different perspectives. We can analyze return, risk and drawdowns, fees and volumes, view position and orders statistical information, calculate statistical benchmarks and metrics, period and time statistics and more.

Everything is also presented in visual form with many charts and graphs as addition to numbers and statistics.

Robustness Analysis

Users can elect fitness function to evaluate strategy robustness such as strategy standard deviation, earnings distribution, tracking error, profit/loss ratio and also custom functions.

In-Sample and Out-of-Sample testing - check how the strategy would perform on the data it was not optimized for.

Anchored & Rolling Walk-Forward Analysis and Optimization support.

Optimal Portfolio Finder

Optimal Portfolio

Optimal portfolio is a such portfolio that produces the highest return and the lowest drawdown. Return to drawdown ratio is a key factor here, we want to trade the portfolio that has the highest R/D ratio.

This means trader can generate bigger earnings without loosing too much - less risk and safer investment. This is an objective of the most traders and investors.

We are giving traders tools to create the best portfolio from generated strategies. Once the strategies are added into the pool, system will employ various methods to find the best portfolio composed from different algo strategies.

Machine Learning

As Automatic Strategy Builder can generate thousands of trading strategies and implement various money management methods, it gets quite hard to build the optimal portfolio as there can be so many alternatives.

We are using machine learning methods to make this process more effective and to find the optimal portfolio faster. Similar way as ASB is running and generating new trading strategies, Optimal Portfolio Finder is running and creating different portfolios.

OPF can be configured to apply different filters to get the portfolio that is the best suited for the user. Portfolio of strategies can then be converted into the source codes and further evaluated, optimized and autotraded.

Machine learning shortens time to get successful results, but requires powerful hardware to deliver results. Strategy building is a very delicate matter. We need to use quality data to get accurate results. We can’t create the strategies on 4 hour chart using only close prices for the test. As we don’t know what happens during those 4 hours, the results would be very inaccurate and strategy could perform well if we use such data, but in real trading it would fail. So if we want to use 4H timeframe we need to use at least minute data to be able to reproduce price behavior during 4H interval so the test is more accurate. Once we achieve satisfactory test results using 1 minute data, we can fine-tune it on tick data and see how it would perform there. Testing 1 minute or tick data on f.e. 10 years worth of data takes a lot of time. The more powerful hardware we run the test on, the faster it is.

It may take to generate 1000 of different strategies to find 1 suitable strategy that would pass all filters. With strategy builder it’s absolute necessity to leave PC turned on running 24/7. Strategy building is infinite process and if it’s not stopped, it will run for the indefinite time until infinity.

In case of strategy building number of discovered profitable trading strategies directly depends on the hardware and time it’s been running for. As we are using evolution algorithm, the more it runs the better it is as it is constantly improving and discarding failed strategies that didn’t pass filters. Also time needed to generate strategies shortens with better hardware. Amount of profitable strategies in certain time is directly dependent on how fast processing power we have. The more, the better.

Advantage of Cloud

Scalability

Cloud allows to add more processing power. User can add more CPU’s with more performance and with more cores, increase RAM. If it will take a day to find the strategy that would pass strict filters, on the server we can increase CPU and RAM so it’s done in hour.

It’s very simple to upgrade service anytime you wish. You can add more CPUs and RAM to your server just by couple clicks. System will recognize the changes right away.

Same goes for downgrading. If you feel you are not using configuration at it’s full potential you are free to downgrade anytime.

Pricing

Free tier allows to run the builder only on 1 timeframe and 1 instrument. Period is also limited so you can’t test for more than a year of historical data. Building is also only allowed for 3 parameters that strategy should have.

Strategy building and portfolio finder requires a lot of processing power. If you really need it, you are at the right place – we can provide it and you can test and optimize based on your needs.

Solution is suitable for individual algo traders, but also for professional trading institutions such as funds and asset management companies. As we are able to scale, we can cover technology requirements and give the exact tools and performance to anyone who needs it so they can concentrate on algo strategy design and trading.

LET THE MACHINES LEARN

ON FIN QUANT

Try today! Sign Up Now.

By continuing to use our page you are agreeing to our website Terms & Conditions, Cookies Policy, Disclaimer and Privacy Policy. By accepting these terms you are agreeing to use our site and process your personal data as stated in our policies.