As per last week, LiquidBot has officially started its Alpha test.

This is the first public test after several private pre-Alpha tests. For now, there are no plans to release this robot to the general public in the future, especially if it proves to be lucrative as the EA commercial sales market is not big enough to make it worthwhile to invest in strong enough protect against decompilation. Alternatives are under consideration, but for now our focus is on development instead.

LiquidBot trades on any currency pair (commodities is a work in progress), and employs a self-learning algorithm, server based to avoid stressing a VPS's CPU. At the time of writing, it has 'learned' 9 pairs. It automatically adds all pairs a broker offers and decides optimal entry and exit strategies, risk levels, lot sizes, news avoidance, correlation computations, etc. This bot is probably by far the most advanced system that has been written to date while being publicly 'known' and not sold to or in use by some huge hedge fund. At the current speed of my poor CPU's, it will learn additional pairs at a rate of about 3-4 per week.

I've made all trades and transactions fully available for your scrutiny.

LiquidBot's alpha test comes in two flavors;

Conservative: 1.5% risk per trade, entry tolerance / 10

Aggressive: 1.5% risk per trade, entry tolerance / 10 * 3

Simply put, we expect the aggressive version to trade at least three times as often, but less accurately. This may cause larger drawdowns. For both versions, arbitration, averaging, grid, hedging, cross-currency hedging or martingale is NOT used. We're contemplating a version that will double the lot size for the next trade, as most losing trades are followed by a winning trade but this is on a wishlist with very low priority.
When you fall you learn, then you get up and you try again until you master it.