Spartan is a multi-asset-class, multi-entry-logic automated trading algorithm. While most conventional algos focus on long or short entries with the view to capitalise on mean reversion or convergence, after a significant move in the asset price, the algorithm does not limit itself to just one entry logic. It has various competing entry logics that can be mean-reversion or trend driven price action strategies. As a result, with a focus on major FX pairs, Bitcoin and certain global indices, it's average win rate over thousands of trades over decades, is c. 74% on shorts and c.82% on longs with a profit factor of c. 1.7x.

​

​

Duration

The longer one exposes a strategy to duration, the higher the risk of macro economic events resulting in trade losses. Besides macro, market pricing aberrations such as price gaps and the increasing costs of holding overnight positions when adding leverage (e.g cost of carry) can mean that holding trades open for longer is less than optimal. Spartan is purposely designed to minimise these as it has hard stops in place for trades that are open for more than a few days (eg. trades on average are just four days long) and dynamic trailing stop losses that de-risk trades, in real-time, when trades enter 'in-the-money' (ITM) status. This means, over the last 14 years in backtests and so far, in live results, the drawdown has been minimal (under 10% on an annual basis). 

​

​

​

​

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Correlation

The inability to reduce correlation across assets that are being traded is one of the biggest risks for any automated strategy. The key is to limit cross contamination in drawdowns when one open trade goes awry because of a macro economic move that impacts multiple asset classes. Spartan trades 10 global assets, entering, on average, a few trades per business day. The Pearson Correlation calculated to address cross correlations in these assets is low. This means, the team has optimised the algorithm such that it trades on low-correlated assets.​​​​​​​​​​​​​​​​​​​​​

​

​​​

Frequency and Spreads

The trade frequency measures approximately 1-5 trades any business day (and on rare instances, it will identify weekend opportunities on digital assets such as Bitcoin). Unlike ultra-high-frequency automated strategies that rely on very small price moves, high leverage and very tight spreads in order to be profitable, Spartan wins trades by a healthy double digit pip margin. While durations are limited, they give the price enough room to move.

​

​​​

News Filter

Many automated strategies rely on convergence. They trade in the opposite direction of a prevailing trend, hoping to profit from price exhaustion, or a macro or statistical reversion to a mean. Therefore, major news events tend to disrupt such strategies significantly, often rendering them unprofitable. Spartan uses news events to often trade in the direction of the prevailing trend, when certain conditions have been met. Therefore a news filter in place would be antithetical to some of it's key entry logics. 

​

​​​

Trade Quality

Adding to losing positions with the hope that the market will eventually turn is an outdated and unproven strategy. The market can remain trending longer than any trader or fund can remain solvent, if they keep throwing more money at the problem. This is why Spartan never adds to a losing trade. It absorbs the loss and moves on to another trade at another time.

​

​​​​

Curve Fitting

Besides correlation and improper risk management, over-optimisation of algorithms on historical data is another downfall of traditional algos. Our team has taken extensive steps to reduce over-optimisation and stress test the algorithm over 14 years, across a range of market conditions and importantly, unseen data. The underlying optimisation techniques are based on a genetic algorithm.

​​